1. Developmental Biology
Download icon

E proteins sharpen neurogenesis by modulating proneural bHLH transcription factors’ activity in an E-box-dependent manner

Research Article
  • Cited 0
  • Views 548
  • Annotations
Cite this article as: eLife 2018;7:e37267 doi: 10.7554/eLife.37267

Abstract

Class II HLH proteins heterodimerize with class I HLH/E proteins to regulate transcription. Here, we show that E proteins sharpen neurogenesis by adjusting the neurogenic strength of the distinct proneural proteins. We find that inhibiting BMP signaling or its target ID2 in the chick embryo spinal cord, impairs the neuronal production from progenitors expressing ATOH1/ASCL1, but less severely that from progenitors expressing NEUROG1/2/PTF1a. We show this context-dependent response to result from the differential modulation of proneural proteins’ activity by E proteins. E proteins synergize with proneural proteins when acting on CAGSTG motifs, thereby facilitating the activity of ASCL1/ATOH1 which preferentially bind to such motifs. Conversely, E proteins restrict the neurogenic strength of NEUROG1/2 by directly inhibiting their preferential binding to CADATG motifs. Since we find this mechanism to be conserved in corticogenesis, we propose this differential co-operation of E proteins with proneural proteins as a novel though general feature of their mechanism of action.

https://doi.org/10.7554/eLife.37267.001

eLife digest

The brain and spinal cord are made up of a range of cell types that carry out different roles within the central nervous system. Each type of neuron is uniquely specialized to do its job. Neurons are produced early during development, when they differentiate from a group of cells called neural progenitor cells. Within these groups, molecules called proneural proteins control which types of neurons will develop from the neural progenitor cells, and how many of them.

Proneural proteins work by binding to specific patterns in the DNA, called E-boxes, with the help of E proteins. E proteins are typically understood to be passive partners, working with each different proneural protein indiscriminately. However, Le Dréau, Escalona et al. discovered that E proteins in fact have a much more active role to play.

Using chick embryos, it was found that E proteins influence the way different proneural proteins bind to DNA. The E proteins have preferences for certain E-boxes in the DNA, just like proneural proteins do. The E proteins enhanced the activity of the proneural proteins that share their E-box preference, and reined in the activity of proneural proteins that prefer other E-boxes. As a result, the E proteins controlled the ability of these proteins to instruct neural progenitor cells to produce specific, specialized neurons, and thus ensured that the distinct types of neurons were produced in appropriate amounts.

These findings will help shed light on the roles E proteins play in the development of the central nervous system, and the processes that control growth and lead to cell diversity. The results may also have applications in the field of regenerative medicine, as proneural proteins play an important role in cell reprogramming.

https://doi.org/10.7554/eLife.37267.002

Introduction

The correct functioning of the vertebrate central nervous system (CNS) relies on the activity of a large variety of neurons that can be distinguished by their morphologies, physiological characteristics and anatomical locations (Zeng and Sanes, 2017). Such heterogeneity is generated during the phase of neurogenesis, once neural progenitors have been regionally specified and are instructed to exit the cell cycle and differentiate into discrete neuronal subtypes (Guillemot, 2007).

Neuronal differentiation and subtype specification are brought together by a small group of transcription factors (TFs) encoded by homologues of the Drosophila gene families Atonal, Achaete-Scute, Neurogenins/dTap and p48/Ptf1a/Fer2 (Bertrand et al., 2002; Huang et al., 2014). These TFs represent a subgroup of the class II of helix-loop-helix proteins and all share a typical basic helix-loop-helix (bHLH) structural motif, where the basic domain mediates direct DNA binding to CANNTG sequences (known as E-boxes) and the HLH region is responsible for dimerization and protein-protein interactions (Massari and Murre, 2000; Bertrand et al., 2002). They are generally expressed in mutually exclusive populations of neural progenitors along the rostral-caudal and dorsal-ventral axes (Gowan et al., 2001; Lai et al., 2016). They are typically referred to as proneural proteins, since they are both necessary and sufficient to switch on the genetic programs that drive pan-neuronal differentiation and neuronal subtype specification during development (Guillemot, 2007). This unique characteristic is also illustrated by their ability to reprogram distinct neural and non-neural cell types into functional neurons (Masserdotti et al., 2016).

Regulating the activity of these proneural proteins is crucial to ensure the production of appropriate numbers of neurons without prematurely depleting the pools of neural progenitors. In cycling neural progenitors, the transcriptional repressors HES1 and HES5 act in response to Notch signalling to maintain proneural TF transcripts oscillating at low levels (Imayoshi and Kageyama, 2014). The proneural proteins are also regulated at the post-translational level. Ubiquitination and phosphorylation have been reported to control their stability, modify their DNA binding capacity or even terminate their transcriptional activity (Ali et al., 2011; Li et al., 2012; Ali et al., 2014; Quan et al., 2016). Furthermore, the activity of these proneural proteins is highly dependent on proteinprotein interactions, and particularly on their dimerization status. It is generally admitted that these TFs must form heterodimers with the more broadly expressed class I HLH/E proteins to produce their transcriptional activity (Wang and Baker, 2015). In this way, the activity of proneural proteins can be controlled by upstream signals that regulate the relative availability of E proteins. Members of the Inhibitor of DNA binding (ID) family represent such regulators. As they lack the basic domain required for direct DNA-binding, ID proteins sequester E proteins through a physical interaction and thereby produce a dominant-negative effect on proneural proteins (Massari and Murre, 2000; Wang and Baker, 2015). Hence, several sophisticated regulatory mechanisms are available during development to control proneural protein activity and fine-tune neurogenesis.

Bone morphogenetic proteins (BMPs) contribute to multiple processes during the formation of the vertebrate CNS (Liu and Niswander, 2005; Le Dréau and Martí, 2013). Yet it is only in the past few years that their specific role in controlling vertebrate neurogenesis has begun to be defined (Le Dréau et al., 2012; Segklia et al., 2012; Choe et al., 2013; Le Dréau et al., 2014). During spinal cord development, SMAD1 and SMAD5, two canonical TFs of the BMP pathway (Massagué et al., 2005), dictate the mode of division that spinal progenitors adopt during primary neurogenesis. Accordingly, strong SMAD1/5 activity promotes progenitor maintenance while weaker activity enables neurogenic divisions to occur (Le Dréau et al., 2014). This model explains how inhibition of BMP7 or SMAD1/5 activity provokes premature neuronal differentiation and the concomitant depletion of progenitors. However, it does not explain why the generation of distinct subtypes of dorsal interneurons are affected differently (Le Dréau et al., 2012), nor how BMP signaling affects the activity of the proneural proteins expressed in the corresponding progenitor domains.

Here, we have investigated these questions, extending our analysis to primary spinal neurogenesis along the whole dorsal-ventral axis. As such, we identified a striking correlation between the requirement of canonical BMP activity for the generation of a particular neuronal subtype and the proneural protein expressed in the corresponding progenitor domain. Inhibiting the activity of BMP7, SMAD1/5 or their downstream effector ID2 strongly impaired the production of neurons by spinal progenitors expressing either ATOH1 or ASCL1 alone, while it had a much weaker effect on the generation of the neuronal subtypes derived from progenitors expressing NEUROG1, NEUROG2 or PTF1a. We found that this differential responsiveness originates from an E-box dependent mode of co-operation of the class I HLH/E proteins with the proneural proteins. E proteins interact with proneural proteins to aid their interaction with CAGSTG E-boxes, facilitating the ability of ASCL1 and ATOH1 to promote neurogenic divisions and hence, neuronal differentiation. Conversely, E proteins inhibit proneural protein binding to CADATG motifs, consequently restraining the ability of NEUROG1/2 that preferentially bind to these motifs to trigger neurogenic division and promote neuronal differentiation. Similar results were obtained in the context of corticogenesis, suggesting that this differential co-operation of E proteins with the distinct proneural proteins is a general feature of their mode of action.

Results

The canonical BMP pathway differentially regulates the generation of spinal neurons derived from progenitors expressing ASCL1/ATOH1 or NEUROG1/NEUROG2/PTF1a

We previously reported that BMP7 signalling through its canonical effectors SMAD1 and SMAD5, is differentially required for the generation of the distinct subtypes of dorsal spinal interneurons (Figure 1A and Le Dréau et al., 2012). Here, we extend this analysis to the generation of neuronal subtypes produced in the ventral part of the developing chick spinal cord. Inhibiting BMP7 or SMAD1/5 expression by in ovo electroporation of specific sh-RNA-encoding plasmids at stage HH14-15 produced a significant reduction in the generation of p2-derived Chx10+ (V2a) and Gata3+(V2b) subtypes 48 hr post-electroporation (hpe), whereas Evx1+ (V0v), En1+ (V1) interneurons and Isl1+ motor neurons were not significantly affected (Figure 1—figure supplement 1). These results revealed a correlation whereby the requirement of the canonical BMP pathway for the generation of discrete spinal neuron subtypes is linked to the proneural protein expressed in the corresponding progenitor domain (Figure 1B,C). The neuronal subtypes strongly affected by BMP7/SMAD1/5 inhibition (dI1, dI3, dI5: Figure 1B,C) were generated from spinal progenitors expressing ATOH1 (dP1) or ASCL1 alone (dP3, dP5). By contrast, all the neuronal subtypes deriving from spinal progenitors expressing either NEUROG1 alone (dP2, dP6-p1) or NEUROG2 (pMN) were much less severely affected (Figure 1B,C). Intriguingly, the V2a/b interneurons that display intermediate sensitivity to BMP7/SMAD1/5 inhibition are derived from p2 progenitors that express both ASCL1 and NEUROG1 (Misra et al., 2014), while the relatively insensitive dI4 interneurons are derived from dP4 progenitors that express PTF1a together with low levels of ASCL1 (Figure 1B,C, Glasgow et al. 2005).

Figure 1 with 1 supplement see all
The canonical BMP pathway differentially regulates the generation of spinal neurons derived from progenitors that express ASCL1/ATOH1 or NEUROG1/NEUROG2/PTF1a.

(A) Actors of the canonical BMP pathway (BMP7, SMAD1 and SMAD5) known to regulate spinal neurogenesis. (B) Dot-plot representing the spinal neuronal subtypes generated 48 hpe with plasmids producing sh-RNA targeting cBmp7 (sh-B7), cSmad1 (sh-S1) or cSmad5 (sh-S5), comparing the electroporated side to the contra-lateral side. The colour code corresponds to the proneural proteins expressed in the corresponding progenitor domains, as shown in C. (C) Drawing of a transverse section of the developing spinal cord at mid-neurogenesis, highlighting: (left) the neuronal subtypes strongly (white) or moderately (grey) affected by inhibiting canonical BMP activity, and (right) a colour-coded representation of the proneural proteins expressed in the corresponding progenitor domains. (D) Working hypothesis whereby we propose to test if (i) the canonical BMP activity is mediated by ID proteins; (ii) ID proteins act by sequestering E proteins (E, orange), thereby inhibiting the activity of class II HLH/proneural proteins (II, grey); and (iii) E proteins co-operate equally or differentially with the distinct proneural proteins as a function of their preferential binding to specific E-box sequences. .

https://doi.org/10.7554/eLife.37267.003

These correlations were particularly interesting in view of recent genome-wide ChIPseq studies that identified the optimal E-box (CANNTG) motifs bound by these TFs: ATOH1 and ASCL1 both preferentially bind to CAGCTG E-boxes (Castro et al., 2011; Klisch et al., 2011; Lai et al., 2011; Borromeo et al., 2014), whereas the optimal motif for NEUROGs is CADATG (where D stands for A, G or T: see Seo et al., 2007; Madelaine and Blader, 2011). Interestingly, most of the E-boxes bound by PTF1a in the developing spinal cord correspond to the CAGCTG motif favored by ASCL1 and ATOH1, yet PTF1a can bind to the NEUROG-like CAGATG motifs in a significant proportion of its targets genes (Borromeo et al., 2014). These observations suggested that the sensitivity of a given progenitor domain to canonical BMP activity originates from the intrinsic DNA-binding preferences of the different proneural bHLH TFs (Figure 1D). In many cell contexts, BMP signaling is mediated by ID proteins (Hollnagel et al., 1999; Moya et al., 2012; Genander et al., 2014), which physically sequester class I HLH/E proteins to produce a dominant-negative effect on proneural proteins (Figure 1D). While this hypothetical signaling cascade could explain the response of spinal progenitors expressing ASCL1 or ATOH1 to altered canonical BMP activity, it would not explain the relative insensitivity of the progenitors expressing NEUROG1, NEUROG2 or PTF1a. Therefore, we tested the veracity of these functional relationships to identify the basis of this differential response (Figure 1D).

ID2 acts downstream of the canonical BMP pathway to differentially regulate the generation of spinal neurons derived from progenitors expressing ASCL1/ATOH1 or NEUROG1/NEUROG2/PTF1a

To test whether ID proteins act downstream of the canonical BMP pathway during spinal neurogenesis, we focused on ID2 (Figure 2A), not least because canonical BMP signalling is necessary and sufficient to promote cId2 expression in the developing spinal cord (Figure 2—figure supplement 1 and Le Dréau et al., 2014). Moreover, the pattern of cId2 expression closely overlaps that described for the canonical BMP activity during spinal cord development (Le Dréau et al., 2012; Le Dréau et al., 2014). At early stages during neural patterning, cId2 expression is restricted to the dorsal region of the developing spinal cord (Figure 2B). Later on during neurogenesis this pattern spreads ventrally throughout the D-V axis, showing expression within all ventral progenitor domains except the pMN domain (Figure 2C–D and Figure 2—figure supplement 2). Inhibition of endogenous ID2 activity was triggered by in ovo electroporation of a sh-RNA specifically targeting chick Id2 transcripts (sh-Id2, Figure 2—figure supplement 3A–E). This ID2 inhibition caused premature cell-autonomous differentiation at 48 hpe similar to that provoked by inhibiting SMAD1/5 (Figure 2E–K and Le Dréau et al., 2014). Conversely, overexpression of a murine ID2 construct reduced the proportion of electroporated cells that differentiated into neurons (EP+;HuC/D+ cells, Figure 2H,K and Figure 2—figure supplement 3F,G). ID2 overexpression could also partially impede the premature differentiation caused by both sh-Id2 and sh-Smad5 (Figure 2I–K). Similar results were obtained when measuring the activity of the pTubb3:luc reporter 24 hpe (Figure 2L).

Figure 2 with 3 supplements see all
ID2 acts downstream of the canonical BMP pathway and it differentially regulates the generation of spinal neurons derived from progenitors expressing ASCL1/ATOH1 or NEUROG1/NEUROG2/PTF1a.

(A) Hypothesis: ID2 mediates the canonical BMP activity during spinal neurogenesis. (B, C) Detection of cId2 transcripts by in situ hybridization in transverse spinal sections at stages HH14 (B) and HH24 (C). Lhx1/5 immunoreactivity (brown) was detected a posteriori (C). (D–D’) Endogenous cID2 immunoreactivity and DAPI staining at stage HH24. (E–J) Transverse spinal cord sections of electroporated cells (H2B-GFP+) that differentiated into neurons (HuC/D+) 48 hpe with: a control plasmid (E), plasmids producing sh-RNAs against cId2 (sh-Id2, (F) or cSmad5 (sh-S5, (G), a murine ID2 construct (H), and its combination with sh-Id2 (I) or sh-S5 (J). (K) Box-and-whisker plots obtained from n = 7–16 embryos; one-way ANOVA + Tukey’s test; *p<0.05. (L) Activity of the pTubb3:luc reporter quantified 24 hpe in the conditions cited above, expressed as the mean fold change ± sem relative to the control, obtained from n = 8–9 embryos; one-way ANOVA + Tukey’s test; *p<0.05. (M) Representative images of the spinal neuron subtypes (identified with the combinations of the markers indicated) generated 48 hpe with control or sh-Id2. (N) Mean ratios ± sem or (O) dot-plots comparing the mean number of neurons on the electroporated and contralateral sides, obtained from n = 8–11 embryos; one-way ANOVA + Tukey’s test; *p<0.05. Scale bars, 50 µM.

https://doi.org/10.7554/eLife.37267.005

We next analysed the consequences of ID2 inhibition on the generation of the different subtypes of spinal neurons and detected a significant dose-dependent reduction in the generation of many neuronal subtypes (Figure 2M,N). The overall phenotype caused by ID2 inhibition was comparable to that triggered by inhibiting BMP7, SMAD1 or SMAD5: the neuronal subtypes deriving from spinal progenitors expressing either ATOH1 or ASCL1 alone were globally more sensitive to ID2 inhibition than those derived from progenitors expressing NEUROG1, NEUROG2 or PTF1a (Figure 2O). Together, these results suggest that ID2 acts downstream of the canonical BMP pathway in spinal neurogenesis and that it regulates distinctly the generation of spinal neurons derived from progenitors expressing ASCL1/ATOH1 and NEUROG1/NEUROG2.

ID2 and E proteins counterbalance each other’s activity during spinal neurogenesis

We wondered whether ID2 contributes to spinal neurogenesis by sequestering E proteins (Figure 3A). Thus, we analysed the expression of these class I HLH genes during spinal neurogenesis. Transcripts from the cTcf3/E2A gene, which encodes the E12 or E47 alternative splice isoforms (Murre et al., 1989), were readily detected in the ventricular zone throughout the dorsal-ventral axis of the developing spinal cord, with apparently no domain-specific pattern (Figure 3B and Holmberg et al., 2008). Transcripts from the chicken HEB orthologue cTcf12 were detected in the transition zone, following a dorsal-to-ventral gradient (Figure 3C). Previous studies reported that E2-2 transcripts were barely detected in the developing murine spinal cord (Sobrado et al., 2009).

Figure 3 with 1 supplement see all
ID2 and E proteins counteract each other’s activity during spinal neurogenesis.

(A) Hypothesis: ID2 sequesters E proteins during spinal neurogenesis. (B, C) Detection of cTcf3/cE2a (B) and cTfc12 (C) transcripts by in situ hybridization in transverse spinal cord sections at stage HH24. (D–O) Transverse spinal cord sections of electroporated cells (GFP+ or H2B-GFP+) that differentiated into neurons (HuC/D+) 48 hpe with: a control (D), E47 (E), TCF12 (F), ID2 (G) or combinations of these (H, I); a control (K), E47bm (L), sh-Id2 (M) or their combination (N). (J, O) Box-and-whisker plots obtained from n = 7–16 (J) and n = 9–16 (O) embryos; one-way ANOVA + Tukey’s test; *p<0.05. Scale bars, 50 µM. .

https://doi.org/10.7554/eLife.37267.009

The overexpression of E47 or TCF12 both produced a significant increase in the proportion of EP+;HuC/D+ cells (Figure 3D–F,J), a phenotype that was reverted by the concomitant electroporation of ID2 (Figure 3D–J). To inhibit the endogenous activity of E proteins, we took advantage of an E47 construct carrying mutations in its basic domain (E47bm) and that acts in a dominant-negative manner over E proteins in vivo (Zhuang et al., 1998). Electroporation of E47bm inhibited neuronal differentiation in a cell autonomous manner, and it fully compensated for the premature differentiation caused by both E47 and TCF12 (Figure 3—figure supplement 1). This E47bm construct also rescued to a large extent the premature differentiation triggered by sh-Id2 (Figure 3K–O). Together, these results appear to confirm that the role played by ID2 during spinal neurogenesis depends on its ability to sequester E proteins.

E47 modulates in opposite ways the neurogenic abilities of ASCL1/ATOH1 and NEUROG1/NEUROG2 during spinal neurogenesis

The results we obtained so far suggested that E proteins themselves might co-operate differently with the distinct proneural proteins during spinal neurogenesis (Figure 4A). To test this hypothesis, we first analyzed the consequences of expressing the E47bm mutant on the generation of spinal neuron subtypes. There was a marked reduction (≥50%) in the generation of Lhx2/9+ (dI1) and Tlx3+ (dI3/dI5) interneurons, which derive respectively from progenitors expressing ATOH1 and ASCL1 alone (Figure 4B,C,F). By contrast, electroporation of E47bm affected to a lesser extent (<25%) the generation of Lhx1/5+ interneurons (dI2/dI4/dI6-V1) or Isl1+ motor neurons deriving from progenitors expressing NEUROG1 alone (dP2, dP6-V1), PTF1a (dP4) or NEUROG2 (pMN, Figure 4D–F). Alternatively, we used another dominant-negative construct of E47: E47Δnls-RFP, inserted in a plasmid with low electroporation efficiency (see the Materials and methods section). This version of E47 fused to RFP is deleted from its nuclear localization signals and thereby impairs the nuclear import of E47, hence its transcriptional activity (Mehmood et al., 2009). As previously reported in vitro (Mehmood et al., 2009), the subcellular localization of this E47Δnls-RFP mutant after in ovo electroporation was mostly cytoplasmic (Figure 4G–H’). As seen with E47bm, this E47Δnls-RFP mutant also impaired neuronal differentiation cell-autonomously (Figure 4G–I). We analyzed the consequences of E47Δnls-RFP electroporation on the generation of spinal neuron subtypes by quantifying the proportions of differentiated electroporated cells (by focusing on the RFP+ cells that were HuC/D+ or Sox2-) that express distinct neuronal subtype markers (Figure 4J–R). Compared to a control plasmid, electroporation of the E47Δnls-RFP mutant reduced about half the proportions of differentiated electroporated cells that express Lhx2/9 or Tlx3 (Figure 4J–M and R), indicating that inhibiting E47 activity hindered the differentiation of spinal progenitors expressing ATOH1 or ASCL1 alone. By contrast, progenitors electroporated with the E47Δnls-RFP mutant efficiently differentiated into Lhx1/5+ interneurons or Isl1+ motor neurons (Figure 4N–R). Of note, we could even observe Lhx1/5+ or Isl1+ electroporated cells within the ventricular zone (stars in Figure 4O–O’, Q–Q’), indicative of a premature differentiation of NEUROG1/2/PTF1a-expressing progenitors when E47 activity is impaired. Hence, ATOH1 and ASCL1 appear to be much more dependent on the activity of E proteins to promote appropriate neuronal differentiation than are NEUROG1, NEUROG2 and PTF1a.

E47 activity is differentially required for the generation of spinal neurons deriving from progenitors expressing ASCL1/ATOH1 and NEUROG1/NEUROG2/PTF1a.

(A) Hypothesis: E proteins co-operate differently with the distinct proneural proteins during spinal neurogenesis. (B–E) Representative images of spinal neurons expressing Lhx2/9 (dI1, (B–B’), Tlx3 (dI3/dI5, (C–C’), Lhx1/5 (dI2/dI4/dI6-V1, (D–D’) or Isl1 (MN, (E–E’), 48 hpe with a control (B–E) or E47bm (B’–E’). (F) Mean ratios ± sem of neuron numbers on the electroporated side relative to the contralateral side, obtained from n = 8–13 embryos; two-sided unpaired t-test; *p<0.05. (G–H’) Transverse spinal cord sections of electroporated cells (RFP+) that differentiated into neurons (HuC/D+) 48 hpe with a control plasmid (G) or a plasmid expressing an E47Dnls-RFP fusion construct (H). (I) Box-and-whisker plots obtained from n = 9–12 embryos; Mann-Whitney’s test; *p<0.05. (J–Q’) Transverse spinal cord sections showing the electroporated cells (RFP+) that differentiated into Lhx2/9+ (J–K’), Tlx3+ (L–M’) or Lhx1/5+ (N–O’) interneurons or Isl1+ motoneurons (P–Q’) 48 hpe with a control plasmid (J,L,N,P) or a plasmid expressing an E47Dnls-RFP fusion construct (K,M,O,Q). Examples of differentiated RFP+;marker+ cells are highlighted by arrowheads, or by stars for the electroporated cells that were found prematurely differentiated in the ventricular zone. (R) Proportions of differentiated electroporated cells (RFP+;HuC/D+ or RFP+;Sox2-) that express the distinct neuronal subtype markers mentioned above, obtained from n = 10–12 embryos; MannWhitney’s non-parametric test for Lhx2/9 or two-sided unpaired t-test for the three other markers; *p<0.05. Scale bars, 50 µM.

https://doi.org/10.7554/eLife.37267.011

Next, we evaluated how E47 gain-of-function modulates the neuronal differentiation induced when ASCL1, ATOH1, NEUROG1 or NEUROG2 are overexpressed (Figure 5A–H). From 24 hpe onwards, all four proneural bHLH proteins caused premature differentiation in a cell-autonomous and concentration-dependent manner (Figure 5—figure supplement 1A–C). Based on these data, we decided to test how E47 addition would alter the phenotypes caused by sub-optimal concentrations of these four proneural proteins. The addition of E47 accentuated the mild increase in neuronal differentiation provoked by ASCL1 at 24 hpe, and more significantly at 48 hpe (Figure 4A–B’). Accordingly, E47 provoked a significant reduction in the average number of electroporated cells generated 48 hpe of ASCL1 (Figure 4I). A similar tendency, albeit less pronounced, was observed when E47 was combined with ATOH1, especially in terms of the reduced average number of EP+ cells generated 48 hpe (Figure 4C–D’,I). Addition of E47 had the opposite effect when combined with NEUROG1 or NEUROG2: it significantly reduced the proportion of EP+;HuC/D+ cells obtained at 24 hpe and consequently increased the final numbers of EP+ cells observed at 48 hpe (Figure 5E–I). These results suggested that E47 differentially regulates the ability of ASCL1/ATOH1 and NEUROG1/NEUROG2 to promote cell cycle exit.

Figure 5 with 1 supplement see all
E47 modulates in opposite ways the neurogenic abilities of ASCL1/ATOH1 and NEUROG1/NEUROG2 during spinal neurogenesis.

(A–H’) Transverse spinal cord sections of electroporated cells (GFP +or H2B-GFP+) that differentiated into neurons (HuC/D+) 24 and 48 hpe with ASCL1 (A–B), ATOH1 (C–D), NEUROG1 (E–F) or NEUROG2 (G–H) alone (white) or together with E47 (black, A‘-H’). Box-and-whisker plots obtained from n = 6–9 (A), 6–8 (C), 6–14 (E) and 7–12 (G) embryos; two-way ANOVA + Sidak’s test; *p<0.05. (I) Mean number ±sem of electroporated cells quantified 48 hpe with the proneural proteins on their own (white) or together with E47 (black), obtained from 6 to 14 embryos; two-sided unpaired t-test; *p<0.05. (J) Cell cycle exit assay. (K) Mean Violet fluorescence intensity measured 48 hpe with a control, ASCL1 and NEUROG1 on their own (white) or together with E47 (black). The individual values (dots, n = 11–23 embryos) and the mean (bars) are shown; one-way ANOVA + Tukey’s test and two-way ANOVA + Sidak’s test; *p<0.05. (L) Assessment of the modes of division of spinal progenitors. (M–O) Transverse spinal cord sections showing the activity of the pSox2:GFP and pTis21:RFP reporters at 16 hpe, when electroporated in combination with control, ASCL1 or NEUROG1 on their own (M–O) or together with E47 (M’-O‘). (P) Mean proportion ±sem of cells identified as pSox2+/pTis21- (PP), pSox2 +/pTis21+ (PN) or pSox2-/pTis21+ (NN) when quantified by FACS, obtained from n = 6–10 pools of embryos; two-way ANOVA + Tukey’s test; *p<0.05. Scale bars, 50 µM. .

https://doi.org/10.7554/eLife.37267.012

To assess cell cycle exit, a fluorescent cytoplasmic-retention dye that is only diluted on cell division was added at the time of electroporation and its mean fluorescence intensity was measured in FACS-sorted electroporated (GFP+) cells 48 hr later (Figure 5J). This assay demonstrated that E47 itself increased the mean Violet intensity, and further enhanced the mild increase caused by ASCL1 (Figure 5K), indicating that E47 facilitates ASCL1’s ability to promote cell cycle exit. E47 had an opposite effect when combined with NEUROG1, significantly reducing the strong increase in Violet intensity caused by NEUROG1 (Figure 5K), thereby confirming that E47 restricts NEUROG1’s ability to promote cell cycle exit.

We next studied how E47 influences the respective abilities of ASCL1 and NEUROG1 to regulate the balance between the three different modes of division that spinal progenitors can undergo during neurogenesis: symmetric proliferative divisions (PP), asymmetric divisions (PN), and symmetric neurogenic divisions (NN) (Saade et al., 2013; Le Dréau et al., 2014). To this end, we took advantage of the pSox2:eGFP and pTis21:RFP reporters that are specifically active during progenitor-generating (PP +PN) and neuron-generating (PN +NN) divisions, respectively (Saade et al., 2013). The effects of E47, ASCL1 and NEUROG1 on the reporters’ activities were assayed 16 hpe by immunohistochemistry or quantified by FACS (Figure 5L). E47 caused a significant decrease in the proportion of pSox2:eGFP+;pTis21:RFP- (PP) cells and a reciprocal increase in the proportion of pTis21:RFP+ (PN +NN) neurogenic divisions relative to the controls (Figure 5M,M’,P). While we did not detect any significant change in the proportions of PP, PN and NN cells in response to ASCL1 alone at this concentration, we did observe an increase in neurogenic divisions at the expense of proliferative divisions when ASCL1 was combined with E47 (Figure 5N,N’,P). Conversely, E47 significantly restrained NEUROG1’s ability to trigger neurogenic divisions at the expense of PP divisions (Figure 5O–P). Assessing the activity of the pSox2:luc reporter confirmed these results, further showing that E47 facilitates the ability of both ASCL1 and ATOH1 to repress pSox2 activity, whereas it restricts the repressive effects of both NEUROG1 and NEUROG2 (Figure 5—figure supplement 1D). Together, these results revealed that E47 co-operates distinctly with ASCL1/ATOH1 and NEUROG1/NEUROG2 to fine-tune neurogenic divisions during spinal neurogenesis.

E47 modulates the transcriptional activities of ASCL1 and NEUROG1 in an E-box dependent manner and through physical interactions

To identify the molecular mechanisms underlying the different outcomes caused by E proteins’ co-operation with the distinct proneural proteins, we focused on the interaction of E47 with ASCL1 and NEUROG1. A DNA-binding deficient version of NEUROG1 (NEUROG1-AQ, Sun et al., 2001), was unable to transactivate the NEUROG-responsive pNeuroD:luc reporter or to promote neuronal differentiation (Figure 6—figure supplement 1). Hence, the ability of NEUROG1 to trigger neuronal differentiation during spinal neurogenesis depends on its transcriptional activity, as previously reported for ASCL1 and ATOH1 (Nakada et al., 2004).

Genome-wide ChIP-seq studies have established that the preferential E-box motifs bound by ASCL1, E47 and NEUROG1 correspond respectively to CAGCTG (Castro et al., 2011; Borromeo et al., 2014), CAGSTG (where S stands for C or G: Lin et al., 2010; Pfurr et al., 2017) and CADATG (where D stands for A, G or T: Seo et al., 2007; Madelaine and Blader, 2011). In light of these intrinsic preferences, we tested how E47 modulates the abilities of ASCL1 and NEUROG1 to bind to DNA and activate transcription in different E-box contexts (Figure 6A and Figure 6—figure supplement 2A). E47 acted in synergy with both ASCL1 and NEUROG1 to drive transcription of the pkE7:luc reporter under the control of 7 CAGGTG repeats (Figure 6B,C). By contrast, E47 and ASCL1 only weakly transactivated the pNeuroD:luc reporter, the promoter of which contains 9 CADATG E-boxes and 1 CAGGTG box (Figure 6D). A similar result was obtained with a version of the pNeuroD:luc reporter in which the single CAGGTG motif was destroyed by mutagenesis (Figure 6—figure supplement 2B), reinforcing the idea that both E47 and ASCL1 preferentially bind to CAGSTG sequences. Intriguingly, E47 markedly reduced the ability of NEUROG1 to enhance the activity of both pNeuroD:luc and its mutated version (Figure 6E and Figure 6—figure supplement 2C). Of note, TCF12 was also able to enhance ASCL1-dependent pKE7:luc activity while inhibiting NEUROG1’s ability to induce pNeuroD:luc activity, though with milder capacities than E47 (Figure 6—figure supplement 2D,E). To further define whether the way E47 modulates ASCL1 and NEUROG1’s transcriptional activity directly depends on the E-box content, we used two additional reporters: pDll1-M:luc and pDll1-N:luc. These are based on conserved regulatory elements found in the promoter of the Delta-like1 gene and have been described to respectively respond to ASCL1 and NEUROG2 (Beckers et al., 2000; Castro et al., 2006). When combined with the pDll1-M:luc reporter containing 3 CAGSTG +1 CADATG motifs, addition of E47 to ASCL1 had only an additive effect (Figure 6—figure supplement 2F), compared to the synergistic effect observed on pKE7:luc activity (Figure 6B). When combined with the pDll1-N:luc reporter containing 1 CAGSTG +3 CADATG motifs, addition of E47 still inhibited the activity induced by NEUROG1 (Figure 6—figure supplement 2G), but less potently than when combined with the pNeuroD:luc reporter (Figure 6E). In vitro ChIP assays further demonstrated that E47 can bind to and enhance ASCL1 binding at the 7 CAGGTG-containing promoter region of the pkE7:luc reporter (Figure 6F), consistent with the notion that their heterodimerization is required for optimal binding and subsequent transcriptional activation. By contrast, E47 caused a significant reduction in the amount of NEUROG1 bound to the promoter region of the pNeuroD:luc reporter (Figure 6G). The fact that E47 itself bound to this promoter region suggested that E47 and NEUROG1 compete for binding to CADATG motifs (Figure 6G), although E47 cannot transactivate them as potently as NEUROG1 (Figure 6E). Together, these results revealed that E47 acts in synergy with both ASCL1 and NEUROG1 when binding to its own optimal E-box (CAGSTG), while it somehow impedes NEUROG1 from binding to CADATG motifs.

Figure 6 with 3 supplements see all
E47 modulates the transcriptional activities of ASCL1 and NEUROG1 in an E-box-dependent manner and through physical interactions.

(A) Hypothesis: E proteins modulate the activity of the proneural proteins differently depending on the E-box context. (B–E) Activity of the pkE7 (B, C) and pNeuroD (D, E) luciferase reporters measured 24 hpe with a control, E47 and ASCL1 (B, D) or NEUROG1 (C, E), expressed as the mean fold change ±sem relative to the control, obtained from n = 8 embryos; one-way ANOVA + Tukey’s test. (F–G) ChIP assays performed on the pkE7 (F) or pNeuroD (G) promoter regions (light grey), or luciferase ORF (Luc, striped grey), in HEK293 cells 24 hr after transfection with HA-ASCL1 (F) or HA-NEUROG1 (G) on their own or together with E47-RFP, expressed as Log2 values of the mean fold change ±sem in DNA binding measured in the presence of E47 relative to absence of E47, obtained from n = 3 (F) or n = 5 (G) experiments; two-sided one sample t-test. The HA-ASCL1, HA-NEUROG1 and E47-RFP proteins probed in western blots, with Tubulin-beta as a transfection control. (H–L) Transverse spinal cord sections of electroporated cells (GFP+) that differentiated into neurons (HuC/D+) 48 hpe with a control (H), ASCL1 or NEUROG1 homodimer (A-A, I; N–N, K), or ASCL1-E47 or NEUROG1-E47 heterodimers (A-E, J; N–E, L). (M) Box-and-whisker plots obtained from n = 12–15 embryos; one-way ANOVA + Tukey’s test. (N) Mean number of electroporated cells (GFP+) generated 48 hpe in the conditions cited above, calculated from n = 11–14 embryos; one-way ANOVA + Tukey’s test. *p<0.05. Scale bars, 50 µM. .

https://doi.org/10.7554/eLife.37267.014

To assess whether these E-box-dependent modulations of proneural proteins’ transcriptional activity by E47 indeed depend on physical interactions, we first prevented E47 from interacting with ASCL1 or NEUROG1 by adding ID2. As expected, addition of ID2 partially rescued both the synergistic effect of E47 on pKE7:luc activity when combined with ASCL1 and the inhibitory effect of E47 on NEUROG1-induced pNeuroD:luc activity (Figure 6—figure supplement 2H,I). Secondly, we compared the activity of tethered constructs that were designed to produce homodimers of ASCL1 (A-A) and NEUROG1 (N-N), or heterodimers with E47 (A-E, N-E: Figure 6—figure supplement 3A–C). Consistent with the results obtained with monomers, A-E heterodimers were significantly more potent than A-A homodimers in driving pkE7:luc activity (Figure 6—figure supplement 3D), while N-N homodimers transactivated pNeuroD:luc much more strongly than N-E heterodimers (Figure 6—figure supplement 3E). A-A and A-E promoted similar neuronal differentiation 48 hpe (Figure 6H–J,M), but the average number of EP+ cells obtained after A-E electroporation was significantly lower than after A-A electroporation (Figure 6N), suggesting that A-E promotes early neurogenic divisions more potently than A-A. This idea was supported by the ability of A-E to repress pSox2:luc activity at 20 hpe, unlike A-A (Figure 6—figure supplement 3F). As for NEUROG1, N-N was significantly more potent than N-E at promoting neuronal differentiation (Figure 6K–M), at reducing the average number of EP+ cells generated 48 hpe (Figure 6N) and at repressing pSox2:luc activity (Figure 6—figure supplement 3F). Thus, the tethered constructs performed like the monomers (Figure 5), supporting the conclusion that E47 facilitates the ability of ASCL1 and restrains that of NEUROG1 to trigger neurogenic divisions during spinal neurogenesis.

E47 modulates in opposite ways the neurogenic abilities of ASCL1 and NEUROG1/NEUROG2 during corticogenesis

We were interested to determine if this differential co-operation of E47 with the distinct proneural proteins could be extended to other regions of the developing CNS. We tested this hypothesis in the developing cerebral cortex, as NEUROG1/2 and ASCL1 all contribute to neurogenesis in this region in mammals (Huang et al., 2014). The development of the cerebral cortex in birds actually shows unexpected similarities to mammalian corticogenesis, including the conservation of its temporal sequence of neurogenesis (Dugas-Ford et al., 2012; Suzuki et al., 2012). As in mammals, corticogenesis in the chick embryo originates from a region of the dorsal pallium expressing PAX6 (Figure 7A,B and Suzuki et al., 2012). From E3 to E5, an early phase of corticogenesis produces the first SOX2-;HuC/D+ cortical neurons, which are generated specifically from PAX6+;TBR2- radial glia-like progenitors that divide at the apical surface, as in mammals (Figure 7C–D). Cortical TBR2+ progenitors that divide basally, similar to mammalian intermediate progenitor cells, appear at around E5 (Figure 7D–D”). The cortical neurons produced during this early phase express TBR1 (Figure 7—figure supplement 1A–A”), as well as other markers typically expressed by mammalian deep-layer neurons (Dugas-Ford et al., 2012; Suzuki et al., 2012). From E3 to E5, cTcf3/cE2A transcripts were detected throughout the whole D-V axis of the developing telencephalon, with apparently no domain-specific pattern (Figure 7—figure supplement 1B–B”). cTcf3/cE2A transcripts were mainly detected in the ventricular zone formed by cortical progenitors (Figure 7E–E’), as previously reported during mouse corticogenesis (Li et al., 2012). At E4, cNeurog1 and cNeurog2 transcripts were detected in a salt-and-pepper fashion in the cortical PAX6+ region (Figure 7F,G), whereas cAscl1 expression was detected strongly in the sub-pallium and more weakly in the developing cerebral cortex (Figure 7H). These expression patterns seen in early chicken embryos are very similar to what is observed in the developing mammalian telencephalon (Huang et al., 2014), suggesting the phylogenetic conservation of the functions played by proneural proteins during corticogenesis.

Figure 7 with 1 supplement see all
E47 modulates in opposite ways the neurogenic abilities of ASCL1 and NEUROG1/NEUROG2 during corticogenesis.

(A) Scheme of the embryonic chick telencephalon at early stages of neurogenesis. (B–D) Coronal telencephalic sections showing PAX6 immunoreactivity and the cell nuclei (DAPI) at low magnification at E4 (B), cortical progenitors and differentiating neurons (SOX2 +and HuC/D+, (C), apical progenitors (PAX6+;TBR2-, (D) and mitotic basal progenitors (TBR2+;pH3+, arrow in D’’) at E3 (C, D), E4 (C’, D’) and E5 (C’’, D’’). (E–H) Detection of cTcf3/cE2a (E), cNeurog1 (F), cNeurog2 (G) and cAscl1 (H) transcripts by in situ hybridization at E4. (I) In ovo electroporation of the chick telencephalon. (J–Q) Coronal telencephalic sections of electroporated cells (GFP+) that differentiated into neurons (HuC/D+) 48 hpe with a control (J), ASCL1 (K), NEUROG1 (L) or NEUROG2 (M) on their own or together with E47 (N–Q). (R) Box-and-whisker plots obtained from n = 5–9 embryos; one-way ANOVA + Tukey’s test; *p<0.05. Scale bars, 50 µM. CCx, cerebral cortex; dT/vT, dorsal and ventral telencephalon; dML, dorso-medial-lateral; VZ, ventricular zone.

https://doi.org/10.7554/eLife.37267.018

To test how E47 modulates the activity of ASCL1 and NEUROG1/2 in the developing chick cerebral cortex, we electroporated the corresponding proneural constructs in the dorsal telencephalic region in ovo at E3 and analysed their effects on neuronal differentiation 2 days later (Figure 7I). Both NEUROG1 and NEUROG2 triggered significant neuronal differentiation in the developing cerebral cortex in a cell autonomous and dose-dependent manner, whereas ASCL1 overexpression had only a minor effect per se (Figure 7J–M,R and Figure 7—figure supplement 1C). E47, which itself had no obvious effect at this concentration (Figure 7N,R), significantly increased neuronal differentiation when combined with ASCL1 (Figure 7O,R). Conversely, E47 markedly reduced the ability of NEUROG1, and to a lesser extent that of NEUROG2, to promote neuronal differentiation (Figure 7P–R). These results suggest that E47 also modulates in opposite ways the neurogenic activities of ASCL1 and NEUROG1/2 in the context of cortical neurogenesis.

Discussion

Class I HLH/E proteins are generally described as obligatory and permissive co-factors for proneural proteins, which must form heterodimers to become active and regulate transcription (Wang and Baker, 2015). The main findings of our study are that the co-operation between E proteins and proneural proteins might be more complex than originally thought. Our results indeed revealed that E proteins can facilitate or restrain the transcriptional activity of the proneural proteins, depending both on their intrinsic DNA-binding preferences and on the E-box content (Figure 8).

Model of the E-box-dependent co-operation of E proteins with proneural proteins.

In neural progenitors, ID proteins (ID) physically sequester E proteins (E), thereby regulating their ability to interact with ASCL1 and ATOH1 (A) or NEUROG1/2 (N). When E protein availability is limited, ASCL1/ATOH1 cannot bind optimally to high affinity CAGSTG E-box motifs, resulting in poor regulation of their target genes and favouring symmetric proliferative (PP) divisions and hence, progenitor maintenance. The release of E proteins from IDs allows heterodimerization with ASCL1/ATOH1, resulting in optimal binding to CAGSTG motifs, correct regulation of the target genes and the appropriate increase in neurogenic asymmetric (PN) and self-consuming (NN) divisions. In the absence of E proteins, NEUROG1/2 bind to high affinity CADATG motifs, possibly as homodimers, and regulate the expression of target genes in an exacerbated manner. This deregulation results in excessive neurogenic divisions that cause premature neuronal differentiation and depletion of the progenitor pool. In the presence of E proteins and when N-E heterodimers are formed, the activity of NEUROG1/2 is moderated and the proportions of the different modes of divisions are balanced appropriately to sustain the progenitor population while promoting correct neuronal differentiation.

https://doi.org/10.7554/eLife.37267.020

On the one hand, our results support a revised model whereby E proteins synergize with proneural proteins specifically at CAGSTG E-boxes, the preferential motifs of E proteins (Lin et al., 2010; Pfurr et al., 2017). Therefore, E proteins facilitate the activity of the proneural proteins that share their preferential binding to CAGSTG motifs, such as ASCL1 and ATOH1 (Castro et al., 2011; Klisch et al., 2011; Lai et al., 2011; Borromeo et al., 2014). Inhibiting the activity of E proteins by overexpressing the E47bm or E47Δnls-RFP mutants strongly impaired the generation of interneurons derived from spinal progenitors that express ATOH1 or ASCL1 alone. Conversely, enhancing the expression of E47 reinforced the ability of ATOH1 and more markedly, that of ASCL1 to promote neuronal differentiation. Our results suggest that this results from the capacity of E47 to increase the ability of these proneural proteins to trigger neurogenic divisions at the expense of proliferative ones (Figure 8). Such co-operation appears to be particularly crucial in the case of ASCL1, whose overexpression could barely increase neurogenic divisions per se, at least at low concentration. These observations support a growing body of evidence that ASCL1 possesses a mild neurogenic potential. For instance, the broad dP3-dP5 domain of spinal progenitors, in which ASCL1 is expressed alone or in combination with PTF1a, expands at the end of primary neurogenesis before producing large numbers of dILA/B neurons during the second neurogenic wave (Wildner et al., 2006; Borromeo et al., 2014). Later on, ASCL1 is also involved in promoting oligodendrogenesis in both the developing brain and spinal cord (Huang et al., 2014). Moreover, recent studies have reported cell cycle promoting-genes among the targets bound by ASCL1 in the ventral telencephalon and that it also sustains the proliferation of adult neural stem cells (Castro et al., 2011; Urbán et al., 2016), suggesting that its mild neurogenic ability might actually be required to sustain long-term production of the neural lineages. Whether the ability of ASCL1 to maintain neural progenitor pools is related to its dependence on the availability of E proteins is an intriguing hypothesis that would be worth testing.

On the other hand, our findings demonstrate that E proteins inhibit proneural protein binding to CADATG motifs. In consequence, E proteins restrict the activity of the proneural proteins that preferentially bind to these motifs, such as NEUROG1/2 (Seo et al., 2007; Madelaine and Blader, 2011). Spinal progenitors electroporated with the E47Δnls-RFP mutant efficiently differentiated into Lhx1/5+ interneurons or Isl1+ motor neurons. Some of these differentiated cells were even observed within the ventricular zone, evidencing that NEUROG1/2/PTF1a-expressing progenitors are prone to differentiate prematurely when E47 activity is impaired. Since E47 restrains the capacity of NEUROG1/2 to promote neuronal differentiation in the context of both spinal neurogenesis and corticogenesis, this would appear to be a general feature. Depletion of the murine E47 isoform was recently shown to increase the production of both TBR1+ and SATB2+ neurons at mid-corticogenesis (Pfurr et al., 2017). In fact, the loss of E47 in early cortical progenitors, for which NEUROG2 constitutes the main proneural protein, causes premature neuronal differentiation. This is consistent with our results and model and it contrasts with the block in neuronal differentiation that would be expected if E47 was essential for NEUROG2 activity.

Our results also suggest that NEUROGs do not necessarily need to form heterodimers with E proteins to trigger neuronal differentiation. Indeed, forced NEUROG1 homodimers drive CADATG-dependent transcription and neuronal differentiation more efficiently than NEUROG1-E47 heterodimers. Similarly, NEUROG2 homodimers better transactivate neuronal differentiation genes than NEUROG2-E47 heterodimers (Li et al., 2012), and EMSA experiments suggested the existence of multiple combinations of proneural homo- and heterodimers (Henke et al., 2009). The physiological relevance of such proneural homodimers is worthy of further study but to date, our attempts to determine whether NEUROG1 homodimers are formed in vivo during spinal neurogenesis remain inconclusive for technical reasons (data not shown). Nevertheless, the strong capacity of NEUROGs to trigger neurogenic divisions independently of E proteins, including self-consuming NN divisions, correlates well with the fact that neural progenitors expressing NEUROG1/NEUROG2 are usually depleted during the neurogenic phase (Simmons et al., 2001; Kim et al., 2011). Together, these results support the notion that E proteins are required to dampen the strong capacity of NEUROGs to trigger neurogenic divisions, thereby avoiding the premature depletion of neural progenitor pools (Figure 8).

The findings that E proteins modulate in opposite ways the activities of ASCL1/ATOH1 and NEUROG1/2 would also explain why modulating canonical BMP activity affects differently the generation of the distinct neuronal subtypes produced during primary spinal neurogenesis. Inhibiting BMP7 or SMAD1/5 would result in the release of E proteins from their complexes with IDs. In turn, this would facilitate ATOH1 and ASCL1 activity, prematurely increasing the proportion of neurogenic divisions undertaken by the corresponding dP1 and dP3/dP5/p2 progenitors, causing their premature differentiation and exhaustion, and ultimately leading to a production of fewer neurons. As NEUROGs are less dependent on E proteins, the inhibition of canonical BMP signalling only mildly impairs the generation of the neuronal subtypes that derive from progenitors expressing NEUROG1/NEUROG2.

In summary, the results presented here led us to propose that E proteins fine-tune neurogenesis by buffering the activity of the distinct proneural proteins. As such, these data add another layer of sophistication to the molecular mechanisms that regulate the activity of proneural bHLH proteins and hence, neurogenesis.

Materials and methods

Key resources table
Reagent type (species)
or resource
DesignationSource or referenceIdentifiersAdditional
information
Strain, strain
background
(Gallus gallus domesticus)
Chicken
fertilized
eggs
Granja Gibert
Cell line
(human)
HEK293ATCCRRID:CVCL_0045
Transfected
construct (empty)
pCAGGS:_ires_
GFP
(Megason and McMahon, 2002)
Transfected
construct (Rattus
norvegicus)
pCAGGS:ASCL1OtherGift from François
Guillemot, Francis Crick
Institute, London, UK
Transfected construct (Rattus
norvegicus)
pCAGGS:ASCL1-t-
E47_ires_GFP
(Geoffroy et al., 2009)
Transfected construct (Rattus norvegicus)pCAGGS:ASCL1-t-
ASCL1_ires_GFP
This paperSee Materials and
methods
Transfected construct (Mus
musculus)
pCAGGS:ATOH1_
ires_GFP
This paperSee Materials and
methods
Transfected construct (Mus
musculus)
pCAGGS:E47(Holmberg et al., 2008)
Transfected
construct (Mus
musculus)
pCAGGS:E47bm_
ires_GFP
This paperSee Materials and
methods
Transfected construct (Rattus
norvegicus)
pCAGGS:HA-
ASCL1
(Alvarez-Rodríguez and Pons, 2009)
Transfected
construct (Rattus
norvegicus)
pCAGGS:HA-
NGN1
This paperSee Materials and
methods
Transfected
construct (Mus
musculus)
pCAGGS:ID2_ires
_GFP
This paperSee Materials and
methods
Transfected
construct (Rattus
norvegicus)
pCAGGS:NGN1OtherGift from François
Guillemot, Francis
Crick Institute,
London, UK
Transfected
construct (Rattus
norvegicus)
pCAGGS:NGN1-t-
E47_ires_GFP
This paperSee Materials and
methods
Transfected
construct (Rattus
norvegicus)
pCAGGS:NGN1-t-
NGN1_ires_GFP
This paperSee Materials and
methods
Transfected
construct (Rattus
norvegicus)
pCAGGS:NGN2OtherGift from François
Guillemot, Francis
Crick Institute,
London, UK
Transfected
construct (Rattus
norvegicus)
pCAGGS:SMAD5-
SD_ires_GFP
(Le Dréau et al., 2012)
Transfected
construct (Mus
musculus)
pCAGGS:TCF12(Holmberg et al., 2008)
Transfected
construct (Mus
musculus)
pCMV2:Flag-
E47-RFP
(Mehmood et al., 2009)
Transfected
construct (Mus
musculus)
pCMV2:Flag-
E47Δnls-RFP
(Mehmood et al., 2009)
Transfected
construct (human)
pCS2:H2B-GFP(Le Dréau et al., 2014)
Transfected
construct (Mus
musculus)
pCS2:Somitabun(Beck et al., 2001)
Transfected
construct (Mus
musculus)
pMiW:myc-
NGN1
(Gowan et al., 2001)
Transfected
construct (Mus
musculus)
pMiW:myc-
NGN1-AQ
(Gowan et al., 2001)
Transfected
construct (empty)
pSuperOligoengine
Transfected
construct (empty)
pShin(Kojima et al., 2004)
Transfected
construct (Gallus
gallus domesticus)
sh-Bmp7(Le Dréau et al., 2012)
Transfected
construct (Gallus
gallus domesticus)
sh-Smad1(Le Dréau et al., 2012)
Transfected
construct (Gallus
gallus domesticus)
sh-Smad5(Le Dréau et al., 2012)
Transfected
construct (Gallus
gallus domesticus)
sh-Id2This paperSee Materials and
methods
Transfected
construct (Mus
musculus)
pTubb3:lucThis paperSee Materials and
methods
Transfected
construct (Mus
musculus)
pId2(backbone):
luc
(Kurooka et al., 2012)
Transfected
construct (Mus
musculus)
pId2(Full):luc(Kurooka et al., 2012)
Transfected
construct (Mus
musculus)
pId2(BREonly):
luc
(Kurooka et al., 2012)
Transfected
construct (Mus
musculus)
pId2(BREmut):
luc
(Kurooka et al., 2012)
Transfected
construct (Gallus
gallus domesticus)
pSox2:luc(Saade et al., 2013)
Transfected
construct (artificial)
pKE7:luc(Akazawa et al., 1995)
Transfected
construct (Mus
musculus)
pNeuroD:lucOtherGift from François
Guillemot, Francis
Crick Institute,
London, UK
Transfected
construct (Mus
musculus)
pNeuroDmut:
luc
This paperSee Materials and
methods
Transfected
construct (Mus
musculus)
pDll1-M:luc(Castro et al., 2006)
Transfected
construct (Mus
musculus)
pDll1-N:luc(Castro et al., 2006)
Transfected
construct (Gallus
gallus domesticus)
pSox2:eGFP(Saade et al., 2013)
Transfected
construct (Mus
musculus)
pTis21:RFP(Saade et al., 2013)
Biological sample
(Gallus gallus
domesticus)
developing
spinal cord
(HH14-HH24)
Biological sample
(Gallus gallus
domesticus)
developing
telencephalon
(E3-E5)
Antibodyanti-Ascl1
(mouse)
BD
Pharmingen
cat #556604;
RRID:AB_396479
AntibodyAnti-Chx10
(guinea pig)
OtherGift from Sam Pfaff,
The Salk Institute
for Biological Studies,
La Jolla, CA, USA;
4% PFA fixation;
dilution 1/10,000
AntibodyAnti-E2A
(rabbit)
Santa Cruzcat #sc-349;
RRID:AB_675504
AntibodyAnti-En-1
(mouse)
OtherGift from Alexandra
Joyner, Sloan Kettering
Institute, New York,
NY, USA; 4% PFA
fixation; dilution
1/100
AntibodyAnti-Evx1
(mouse)
DSHBcat #99.1-3A2;
RRID:AB_528231
AntibodyAnti-Gata3
(mouse)
Santa Cruzcat #sc-268;
RRID:AB_2108591
AntibodyAnti-HA
(rabbit)
Abcamcat #ab20084
RRID:AB_445319
AantibodyAnti-HuC/D
(mouse)
Life
Technologies
cat #A-21271;
RRID:AB_221448
AntibodyAnti-ID2
(mouse)
Abcam/Thermo
Fisher Scientific
cat #ab166708; RRID:
AB_2538566
AntibodyAnti-ID2
(rabbit)
Calbio Reagentscat #M213; RRID:
AB_1151771
AntibodyAnti-Isl1
(mouse)
DSHBcat #402D6; RRID:
AB_528315
AntibodyAnti-Lbx1
(guinea pig)
OtherGift from Thomas
Müller and Carmen
Birchmeier, Max
Delbrück Center
for Molecular
Medicine, Berlin,
Germany; 4% PFA
fixation; dilution
1/10,000
AntibodyAnti-Lhx1/5
(mouse)
DSHBcat #4F2; RRID:
AB_2314743
AntibodyAnti-Lhx2/9
(rabbit)
OtherGift from Thomas
Jessell,
Zuckerman Institute,
Departments of
Neuroscience, and
Biochemistry and
Molecular Biophysics,
Columbia University,
New York, NY, USA;
4% PFA fixation;
dilution 1/5,000
AntibodyAnti-Lmx1b
(mouse)
DSHBcat #50.5A5;
RRID:AB_2136850
AntibodyAnti-Neurogenin-1
(rabbit)
Milliporecat #AB15616;
RRID:AB_838216
AntibodyAnti-Nkx6.1
(mouse)
DSHBcat #F55A10;
RRID:AB_532378
AntibodyAnti-Pax2
(rabbit)
Zymedcat #71–6000;
RRID:AB_2533990
AntibodyAnti-Pax6
(mouse)
DSHBCat#pax6;
RRID:AB_528427
AntibodyAnti-Pax6
(rabbit)
Covance/Biolegendcat #901301;
RRID:AB_2565003
AntibodyAnti-pH3
(rat)
Sigma Aldrichcat #H9908;
RRID:AB_260096
AntibodyAnti-RFP
(rabbit)
(Herrera et al., 2014)
AntibodyAnti-Sox2
(rabbit)
Abcamcat #ab97959;
RRID:AB_2341193
AntibodyAnti-Trb1
(rabbit)
Abcamcat #ab31940;
RRID:AB_2200219
AntibodyAnti-Tbr2
(rabbit)
Abcamcat #ab23345;
RRID:AB_778267
AntibodyAnti-Tlx3
(guinea pig)
OtherGift from Thomas
Müller and Carmen
Birchmeier, Max
Delbrück Center
for Molecular
Medicine, Berlin,
Germany; 4% PFA
fixation; dilution
1/10,000
AntibodyAnti-Tubulin,
beta (mouse)
Milliporecat #MAB3408;
RRID:AB_94650
Sequence-
based reagent
Primers for
ChIP of the pKE7
promoter region:
This study#FW:CAggTgg
CAggCAgg; #REV:
AAgCAgAACg
Tgggg
Sequence-
based reagent
Primers for ChIP
of the pNeuroD
promoter region:
This study#FW:CTggCTTCTAA
TCTCCATCAC;
#REV:gTTTCCAg
ATgAggTCgT
Sequence-
based reagent
Primers for ChIP
of the Luciferase
ORF:
This study#FW:gAAgACgC
CAAAAACATA;
#REV:CgTATCTCT
TCATAgCCTTA
Commercial
assay or kit
CellTrace Violet
Cell Proliferation
Kit
Thermo Fisher
Scientific
Cat #C34557
Software,
algorithm
Adobe Photoshop
CS5
AdobeRRID:SCR_014199
Software,
algorithm
FlowJoTreeStar IncRRID:SCR_000410
Software,
algorithm
ImageJ(Schneider et al., 2012)RRID: SCR_003070http://imagej.net/Welcome
Software,
algorithm
Leica LAS AF
Image Acquisition
Leica
Microsystems
RRID:SCR_013673

In ovo electroporation

Fertilized white Leghorn chicken eggs were provided by Granja Gibert, rambla Regueral, S/N, 43850 Cambrils, Spain. Eggs were incubated in a humidified atmosphere at 38·C in a Javier Masalles 240N incubator for the appropriate duration and staged according to the method of Hamburger and Hamilton (HH, (Hamburger and Hamilton, 1951). According to EU animal care guidelines, no IACUC approval was necessary to perform the experiments described herein, considering that the embryos used in this study were always harvested at early stages of embryonic development (at E5 at the latest). Sex was not identified at these stages.

Unilateral in ovo electroporations in the developing chick spinal cord and dorsal telencephalon were performed respectively at stages HH14-15 and HH18 (54 and 69 hr of incubation). In the telencephalon, corticogenesis was studied specifically in the dorsal-medial-lateral (dML) subregion to minimize any possible variability along the medial-lateral axis. Plasmids were diluted in RNAse-free water at the required concentration [0 to 4 µg/µl] and injected into the lumen of the caudal neural tube or the right cerebral ventricle using a fine glass needle. Electroporation was triggered by applying 5 pulses of 50 ms at 22.5 V with 50 ms intervals using an Intracel Dual Pulse (TSS10) electroporator. Electroporated chicken embryos were incubated back at 37C and recovered at the times indicated (16–48 hr post-electroporation).

Cell lines

The human embryonic kidney-derived HEK293T cell line was obtained from ATCC (#CRL-3216; STR profile = CSF1PO: 11,12; D13S317: 12,14; D16S539: 9,13; D5S818: 8,9; D7S820: 11; TH01: 7, 9.3; TPOX: 11; vWA: 16,19; Amelogenin: X). This cell line is not listed in the commonly misidentified cell lines list from the International Cell Line Authentication Committee. Cells were Mycoplasma-free, as routinely assessed using the LookOut Mycoplasma PCR Detection Kit (Sigma #MP0035-1KT). HEK293T cells were grown at sub-confluent density in DMEM supplemented with 10% fetal bovine serum and penicillin/streptomycin at 37°C, 5% CO2.

Plasmids

To facilitate comparisons in gain-of-function experiments, all the constructs used in this study were inserted under the control of a pCAGGS promoter that harbors high activity in chick (pCAGGS or pCAGGS_ires_GFP, kindly provided by Andy McMahon, Megason and McMahon, 2002), and were electroporated at similar concentrations (0, 0.1, 0.5 or 1 µg/µl as specified in the respective figure legends). Non-fluorescent pCAGGS plasmids were combined with 0.25 µg/µl of pCS2_H2B-GFP for visualization. The pCAGGS:ASCL1, pCAGGS:NEUROG1 and pCAGGS:NEUROG2 plasmids were kindly provided by François Guillemot. The pCAGGS:ATOH1_ires_GFP plasmid was obtained by subcloning from a pCMV:ATOH1 kindly provided by Nissim Ben-Arie (Krizhanovsky et al., 2006). The pCAGGS:E47 and pCAGGS:TCF12 were kindly provided by Jonas Muhr (Holmberg et al., 2008). The pCAGGS:E47bm_ires_GFP plasmid was derived from a pGK:E47_CFP plasmid kindly provided by Yuan Zhang (Zhuang et al., 1998). The pCAGGS:ID2_ires_GFP was derived from a pCMV:ID2 plasmid, and the pCAGGS_SMAD5-SD_ires_GFP was described previously (Le Dréau et al., 2012). Only E47Δnls-RFP and E47-RFP (in a pFlag-CMV2 vector, kindly provided by Yoshihiro Yoneda, Mehmood et al., 2009), Somitabun (pCS2:Somitabun, kindly provided by Jonathan Slack, Beck et al., 2001) and NEUROG1-AQ and its wild-type NEUROG1 version (pMiW:myc-NGN1 and pMiW:myc-NGN1-AQ, kindly gifted by Jane Johnson, Gowan et al., 2001) were used in a different backbone. HA-tagged versions of ASCL1 (pCAGGS:HA-ASCL1, Alvarez-Rodríguez and Pons, 2009) and NEUROG1 (pCAGGS:HA-NEUROG1) and a Flag-E47-RFP construct were used for chromatin immuneprecipitation assays. Inhibition of cBmp7, cSmad1, cSmad5 or cId2 expression was triggered by electroporation of short-hairpin constructs inserted into the pSuper (Oligoengine) or pSHIN (Kojima et al., 2004) vectors. Electroporation of 2–4 µg/µl of these constructs caused a specific and reproducible 50% inhibition of the target expression (see Le Dréau et al., 2012). The pSox2:GFP and pTis21:RFP reporters used to assess the modes of divisions undergone by spinal progenitors were previously described in details (Saade et al., 2013). The pSox2:luc derived from the pSox2:GFP (Saade et al., 2013). The different versions of the pId2:luc reporter were kindly provided by Yoshifumi Yokota (Kurooka et al., 2012) and the pkE7:luc by Masashi Kawaichi (Akazawa et al., 1995). The pNeuroD:luc, pDll1-M:luc and pDll1-N:luc reporters were kindly provided by François Guillemot (Castro et al., 2006). The pNeuroDmut:luc reporter was obtained by site-directed mutagenesis of the single CAGGTG E-box contained in the NeuroD promoter region. The pTubb3:luc reporter was obtained by subcloning the Tubb3 enhancer region present in the pTubb3enh:GFP plasmid (kindly provided by Jonas Muhr, Bergsland et al., 2011) into the pGL3:luc vector (Promega). Please head for the Key Resources Table for additional information.

Generation of tethered constructs

The tethered bHLH dimers were derived from the pCAGGS:ASCL1-t-E47_ires_GFP kindly provided by François Guillemot (Geoffroy et al., 2009). This plasmid and pCAGGS:NEUROG1 were used as templates to generate the ASCL1-t-ASCL1, NEUROG1-t-E47 and NEUROG1-t-NEUROG1 constructs inserted into pCAGGS_ires_GFP, using a tether peptide AAAGTSAGGAAAGTSASAATGA flanked by SpeI and ClaI restriction sites as described previously (Henke et al., 2009). Expression of the tethered bHLH dimers was assessed by western blot after transfection into HEK293 cells. Transient cell transfections were obtained by electroporation applying 2 pulses of 120V, 30 ms (Microporator MP‐100, Digital Bio). Cells were grown for 24 hr onto poly-L-Lysine-coated 6-well dishes in DMEM/F12 supplemented with 10% fetal bovine serum and 50 mg/L of Gentamicin until reaching 70‐80% confluence. The typical transfection efficiency of this procedure was 40–60%. Cells were lysed in 1X SDS loading buffer (10% glycerol, 2% SDS, 100 mM dithiothreitol, and 62.5 mM Tris-HCl, pH 6.8) and DNA was disrupted by sonication. Protein extracts were separated by SDS-PAGE electrophoresis, transferred to Immobilon-FL PVDF membranes (IPFL00010, Millipore), blocked with the Odyssey Blocking Buffer (927–40000, LI-COR), and incubated with antibodies against ASCL1 (BD Pharmingen, cat#556604, 1:1000), NEUROG1 (Millipore, cat#AB15616, 1:3000) or E2A (Santa Cruz, cat#sc-763, 1:1000). Detection was performed using fluorescence-conjugated secondary antibodies and an Odyssey Imaging System (LI-COR).

Immunohistochemistry

For immunohistochemistry experiments, chicken embryos were carefully dissected, fixed for 2 hr at 4°C in 4% paraformaldehyde and rinsed in PBS. Immunostaining was performed on either vibratome (40 µm) or cryostat (16 µm) sections following standard procedures. After washing in PBS-0.1% Triton, the sections were incubated overnight at 4C with the appropriate primary antibodies (see Key Resources Table) diluted in a solution of PBS-0.1% Triton supplemented with 10% bovine serum albumin or sheep serum. After washing in PBS-0.1% Triton, sections were incubated for 2 hr at room temperature with the appropriate secondary antibodies diluted in a solution of PBS-0.1% Triton supplemented with 10% bovine serum albumin or sheep serum. Alexa488-, Alexa555- and Cy5-conjugated secondary antibodies were obtained from Invitrogen and Jackson Laboratories. Sections were finally stained with 1 μg/ml DAPI and mounted in Mowiol (Sigma-Aldrich).

Image acquisition, treatment and quantification

Optical sections of fixed samples (transverse views of the spinal cord, coronal views for the telencephalon) were acquired at room temperature with the Leica LAS software, in a Leica SP5 confocal microscope using 10x (dry HC PL APO, NA 0.40), 20x (dry HC PL APO, NA 0.70), 40x (oil HCX PL APO, NA 1.25–0.75) or 63x (oil HCX PL APO, NA 1.40–0.60) objective lenses. Maximal projections obtained from 2 µm Z-stack images were processed in Photoshop CS5 (Adobe) or ImageJ for image merging, resizing and cell counting.

Quantification of endogenous ID2 intensity was assessed using the ImageJ software. Cell nuclei of H2B-GFP + electroporated and neighboring non-electroporated cells were delimitated by polygonal selection, and the mean intensity of ID2 immunoreactivity quantified as mean gray values. Quantifications were performed on at least six electroporated and six non-electroporated cells per image, in at least three different images per embryo.

In situ hybridization

Chicken embryos were recovered at the indicated stage, fixed overnight at 4°C in 4% PFA, rinsed in PBS and processed for whole mount RNA in situ hybridization following standard procedures. Probes against chick cId2 (#chEST852M19) and cNeurog2 (#chEST387d10) were purchased from the chicken EST project (UK-HGMP RC). Probes against cTcf3/E2a, cAscl1 and cNeurog1 were kindly provided by Drs Jonas Muhr, José-Maria Frade and Cristina Pujades. The probe against cTcf12/cHeb was obtained by PCR from genomic DNA of E4 chicken embryonic tissue and the purified 623 nucleotides insert was sub-cloned into the pGEM-T vector (Promega). Hybridized embryos were post-fixed in 4% PFA and washed in PBT. 45µM-thick sections were cut with a vibratome (VT1000S, Leica), mounted and photographed using a microscope (DC300, Leica). The data show representative images obtained from three embryos for each probe.

Luciferase assay

Transcriptional activity was assessed following electroporation of a luciferase reporter together with a renilla luciferase reporter used for normalization, in combination with the indicated plasmids required for experimental manipulation. Embryos were harvested 24 hr later and GFP-positive neural tubes were dissected and homogenized in a Passive Lysis Buffer on ice. Firefly- and renilla-luciferase activities were measured by the Dual Luciferase Reporter Assay System (Promega).

Cell cycle exit assay

The average number of divisions undergone by electroporated spinal progenitors was assessed in vivo using the CellTrace Violet Cell Proliferation Kit (Invitrogen). The Violet cell tracer (1 mM), a cytoplasmic retention dye that becomes diluted as cells divide, was injected into the lumen of the neural tube at the time of electroporation. Embryos were recovered 48 hr later, the neural tubes were carefully dissected and recovered and the cells dissociated following a 10–15 min digestion in Trypsin-EDTA (Sigma). The fluorescence intensity of the Violet tracer was measured in viable dissociated electroporated GFP+ cells in the 405/450 nm excitation/emission range on a Gallios flow cytometer (Beckman Coulter, Inc).

Assessment of the modes of divisions

Chicken embryos were recovered 16 hr after co-electroporation of the pSox2:eGFP and pTis21:RFP reporters together with the indicated bHLH TF-encoding plasmids. Cell suspensions were obtained from pools of 6–8 dissected neural tubes after digestion with Trypsin-EDTA (Sigma) for 10–15 min, and further processed on a FACS Aria III cell sorter (BD Biosciences) for measurement of eGFP and RFP fluorescences. At least 1,000 cells for each progenitor population (PP, PN and NN) were analyzed per sample.

Chromatin immunoprecipitation assay

HEK293T cells were transfected by a standard calcium phosphate co-precipitation protocol with combinations of pCAGGS_ires_GFP, pCAGGS:HA-ASCL1, pCAGGS:HA-NEUROG1, pCMV2_Flag-E47-RFP together with the pkE7:luc or pNeuroD:luc reporters, with a total of 10 µg of DNA per 100 mm dish. 24 hr later, cells were collected and 10% of the material was reserved to check transfection by Western blot. For chromatin immunoprecipitation assays, approximately 1 million transfected HEK293T cells were fixed with 1% formaldehyde for 10 min at room temperature. Fixation was quenched by adding 0.125M glycine for 5 min. After two washes with PBS, cells were lysed on ice for 20 min in a lysis buffer containing protease inhibitors (1% SDS; 10 mM EDTA pH8.0; 50 mM Tris-HCl pH8.1). Sonication was performed with a Bioruptor sonicator to obtain 200–500 bp shredded chromatin fragments. Chromatin purification was carried out by spinning samples down at maximum speed at 4C during 30 min. Purified chromatin was pre-cleared with protein A agarose (Millipore #16–125) for 30 min. 25 µg of chromatin were immunoprecipitated with 5 µL of anti-RFP serum (Herrera et al., 2014), 2 µg of anti-HA (Abcam, cat#20084), anti-NEUROG1 (Millipore, cat#15616) or unspecific rabbit IgG (Diagenode, cat#C15410206) antibodies. Antibody-chromatin complexes were recovered using magnetic beads (Magna ChIP, Millipore, cat#16–661) and immuno-complexes were washed once with TSE I (0.1% SDS; 1% Triton-X100; 2 mM EDTA pH8.0; 20 mM Tris-HCl pH8.1; 150 mM NaCl), TSE II (0.1% SDS; 1% Triton-X100; 2 mM EDTA pH8.0; 20 mM Tris-HCl pH8.1; 500 mM NaCl), TSE III (0.25M LiCl; 1% NP-40; 1% Sodium Deoxicholate; 1 mM EDTA pH8.0; 10 mM Tris-HCl pH8.1) and twice with TE (Tris-HCl 10 mM, EDTA 1 mM). Reversal of crosslinking was done by incubating samples in elution buffer (1% SDS, 0.1M NaHCO3) overnight at 65C. DNA was purified by phenol-chloroform extraction followed by ethanol precipitation. Quantification of the DNA target regions and negative control (luciferase ORF) was assessed by qPCR in a Lightcycler 480 (Roche) using specific primers (see Key Resources Table).

Proteins extracts were obtained by incubation in a RIPA buffer (150 mM NaCl, 1.0% NP-40,0.5% sodium deoxycholate, 0.1% SDS and 50 mM Tris, pH 8.0) supplemented with protease and phosphatase inhibitors for 20 min on ice and centrifugation (20 min at maximum speed). 30 μg of protein samples were mixed with the Laemmli buffer (375 mM Tris pH = 6.8, 12%SDS, 60% glycerol, 600 mM DTT, 0.06% bromphenol blue), heated to 95°C and then separated on a SDS-PAGE gel in running buffer (25 mM Tris base, 190 mM glycine, 0.1% SDS, pH = 8,3). Proteins were transferred to a nitrocellulose membrane using transfer buffer (190 mM glycine, 25 mM Tris, 20% Methanol, 0.1% SDS) for 90 min at 80V. Membranes were blocked for 1 hr with a solution of PBS-5% milk, 1% Tween (PBST) and further incubated overnight at 4C with appropriate primary antibodies diluted in PBST: rabbit anti-HA (Abcam, cat #ab20084), rabbit anti-RFP serum (Herrera et al, 2014) and mouse anti-Tubulin beta (Millipore, cat #MAB3408). After three washes in PBST, membranes were incubated with Horseradish peroxidase-conjugated anti-rabbit IgG or anti-mouse IgG secondary antibodies (Sigma-Aldrich, cat#GENA934-1ML and cat#GENA931) for 1 hr at room temperature and the signals detected by chemiluminescence using the Immobilon western chemiluminiscent HRP substrate (Sigma-Aldrich, cat# WBKLS0100).

Statistical analyses

No statistical method was used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments or outcome assessment. Statistical analyses were performed using the GraphPad Prism six software (GraphPad Software, Inc.). For in vivo experiments, cell counts were typically performed on 2–5 images per embryo and n values correspond to different embryos, except for the assessment of the modes of divisions where n values correspond to pools of embryos. For in vitro chromatin immunoprecipitation assays, n values represent the numbers of independent experiments performed. The n values are indicated in the corresponding figure legends. The normal distribution of the values was assessed by the Shapiro-Wilk normality test. Significance was then assessed with a two-sided unpaired t-test, one-way ANOVA + Tukey’s test or two-way ANOVA + Sidak’s test for data presenting a normal distribution, or alternatively with non-parametric Mann–Whitney or Kruskal-Wallis +Dunn’s multiple comparisons’ tests for non-normally distributed data. n.s: non-significant; *: p<0.05 or less, as indicated in individual figures.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
    A mitogen gradient of dorsal midline wnts organizes growth in the CNS
    1. SG Megason
    2. AP McMahon
    (2002)
    Development 129:2087–2098.
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63

Decision letter

  1. Jeremy Nathans
    Reviewing Editor; Johns Hopkins University School of Medicine, United States
  2. Marianne E Bronner
    Senior Editor; California Institute of Technology, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: this article was originally rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]

Thank you for submitting your manuscript "E proteins differentially co-operate with proneural bHLH transcription factors to sharpen neurogenesis" to eLife. Your manuscript has been reviewed by a Senior Editor, a Reviewing Editor, and three reviewers.

While the reviewers found the work interesting, they also raised a large number of substantive questions that left us no choice but to reject the paper. This decision reflects the eLife view that a submitted manuscript should be reasonably close to a final version to be accepted. We hope that the reviewers' comments will be useful to you. We apologize for not being able to deliver better news, and we hope that you will continue to consider eLife for future submissions.

Reviewer #1:

In the manuscript by Le Dreau, Escalona and colleagues, the authors have assessed the function and mechanisms of action of E proteins and bHLH transcription factors during neurogenesis and neural fate specification. Using predominantly chick neural tube electroporation assays, as well as molecular approaches, the authors present a model in which E proteins have differential effects on neurogenesis, depending whether they interact with ASCL1/ATOH1 versus NEUROG1/2 bHLH transcription factors. In addition, the authors argue that Bmp signaling contributes to this specificity through induction of ID2 expression, which inhibits the activity of ASCL1/ATOH1 bHLHs.

The authors have performed an impressive number of experiments to generate a model of how various HLH proteins interact to generate the appropriate number of neurons of a specific fate. However, I found the amount of data somewhat overwhelming, and in places disjointed with the overall point the authors are trying to make. In many places, I do not think the authors went far enough to support their model, or provide enough data to convince me their results have biological meaning.

In Figure 1 the authors show that inhibiting Bmp5 or Smad signaling leads to a selective loss of p2-domain interneurons, as assessed by Chx10 and Gata3 expression. These neurons are defined by Ascl1 expression in progenitors. First, it is unclear to me why these cells would be sensitive to Bmp loss, as most Bmps are normally expressed in the dorsal region of the spinal cord. Second, what is the evidence that p2-domain cells are normally responding to Bmp? Is there detectable expression of phospo-Smad1/5 in this domain? Finally, what happens to expression of Ascl1 under these conditions? Most of the markers analyzed are expressed in postmitotic cells, and it is unclear what is happening to the identity of spinal progenitors in these experiments.

Figure 2 addresses the role of Id2 proteins, where the authors first state the Bmp is necessary and sufficient to promote Id2 expression in the spinal cord. This is not actually shown for the endogenous gene but instead using an Id2 promoter-luciferase reporter that was electroporated within the spinal cord. So, I think this is somewhat overstated in the text. Based on the data in Figure 1, one also wonders what happens to endogenous Id2 expression after Bmp7 and Smad1/5 knockdown.

The main part of this Figure analyzes the effects of Id2 misexpression and inhibition, where the authors find that Id2 misexpression inhibits the generation of most ventral cell types, while knockdown of Id2 is argued to cause premature differentiation. The biological relevance of these experiments is unclear to me, largely because there is clearly a domain in the ventral spinal cord that is lacking in Id2 expression (Figure 2B-D). If this domain includes the p2/Ascl1 domain then it hard to understand how these experiments provide insights into what is happening naturally.

In the next experiments the authors asked whether Id2 contribute to neurogenesis through sequestering E proteins. The authors show that overexpression of E proteins increases the production of differentiated cells, an activity that can be inhibited by co-expression of Id2. They also expressed a dominant negative E protein that inhibited neural differentiation, and this effect was lost when Id2 was depleted.

In Figure 4, the authors argue that E47 has selective function in the generation of neurons derived from ASCL1+ domains. This is partly based on experiments expressing dnE47 that the author's state has a more selective effect on ASCL1 derivatives. The quantification of these experiments is based on ratios of total neurons from electroporated and non-electroporated sides of the spinal cord. These results could conceivably reflect the dorsal-ventral gradients of electroporation efficiencies that one normal sees. Ideally one would want to know how these data look in terms of electroporated (GFP+) neurons that expressed a specific marker. If for example, if motor neurons are unaffected, one would expect to see a larger percentage of Isl1+/GFP+ than Chx10+/V2+ neurons in these experiments.

In Figure 5 the authors provide evidence that the specificity of ASCL1/E47 action are due to its ability to selectively activate sequences containing CAGSTG sequences. E47 also inhibited the ability of NEUROG1 activate its CADATG targets, suggested that E47 can functionally inhibit NEUROG1 at certain sequences.

Finally, the authors tested whether differentiation interactions of ASCL1 and NEUROG1 with E47 might operate in other regions of the CNS. The authors show that E47 enhances the ability of ASCL1 promote neuronal differentiation in the cortex. In contrast they found that E47 blocks the ability of NEUROG1 to promote cortical neuronal differentiation. As with many other experiments in this study, the biological relevance of these interactions are unclear, as we are given no detailed information on the temporal or spatial expression profile of E47 during cortical development.

The authors have provided detailed information on the statistical methods used in the methods section, and n values for samples in the Figure legends.

Reviewer #2:

This is a interesting study that puts forward a new mechanistic model to understand how different/limiting levels of E-proteins could have differential effects on neurogenesis through modulating activity of proneural bHLH proteins. The model is supported by manipulating levels of BMP signaling, ID levels, and E-protein/proneural bHLH ratios in the chick neural tube and assessing consequences to division patterns and neuronal subtype specification during embryogenesis. Additional experiments assess the different activities on the proneural bHLH factors with and without the E-protein E47 with respect to transcriptional activity of reporter genes with different motifs. The study is written clearly, and data are shown carefully and thoroughly for the most part. The ChIP data are not particularly strong, and the activities rely heavily on over expression assays, nevertheless, multiple types of experiments are used to develop the model and support the conclusions presented.

1) The authors come up with the phrase 'dual co-operation of E proteins with proneural proteins' to describe the differential consequences to neurogenesis that result when E-proteins are limiting. It is a confusing description. Based on the model put forward, it is a differential activity of homodimers of ASCL1/ATOH1 versus NEUROG1/NEUROG2 that form when E-proteins are limiting reflecting the intrinsic binding preferences of the different proneural bHLH TFs. To call this differential co-operation of E-proteins does not really help understand what is going on. I recommend changing how this is stated throughout the manuscript.

2) Some of the differences in activities could be affected by different levels of the proneural proteins made from the expression constructs. Assessment of the protein levels will be difficult to compare because no tag is included, and the antibodies can work at different levels. Minimally it should be considered. Ideally it should be evaluated. This only becomes important because arguments are made about the different effects of E-proteins on the factors. It may be possible that the protein levels of the proneurals are substantially different and are showing differential effects in these assays. This could be true even though pCIG is used for them all.

Also related to this issue, in the Discussion section it is stated that: "Such co-operation appears to be particularly crucial in the case of ASCL1, whose overexpression could barely increase neurogenic divisions per se." This contrasts with other studies ie Nakada 2014 that show Ascl1 is very efficient at increasing differentiation even at 24 hrs. Is it possible that expression levels are relatively low from the construct used here relative to Nakada, or relative to the other bHLH in the current study as mentioned above? This may need some consideration or ruled out by determining the comparative levels of the overexpressed bHLH proteins.

3) All the results in the study are from E47 but the conclusions are generalized to E-proteins. Was HEB tested in these assays? This should be shown if the generalization is being used.

Reviewer #3:

In this study, LeDréau et al., explore how differences in the formation of proneural bHLH-E protein transcription factor complexes can influence neuronal differentiation in the developing spinal cord and cortex. A link to BMP influences on this process is further explored though an examination of the function of Id2, a BMP-regulated inhibitor of proneural bHLH-E protein complexes. Some of the initial observations include findings that shRNA knockdown of Bmp7, Smad1, or Smad5 potently suppresses the formation of certain classes of spinal neurons (dI1, dI3, and V2a/b) while having less of an effect on other classes (dI6, V0, V1, and MNs). These trends are argued to align with the association of these two groups with the activity of the proneural factors Ascl1 and Atoh1 vs. Neurog1 and Neurog2. The connection of BMP signaling to proneural genes is supposed by evidence that BMP7/Smad1/5 can regulate Id2. shRNA knockdown of Id2 appears to broadly enhance neurogenic differentiation.

The focus of the study then shifts to an exploration of the requirement and sufficiency of E proteins, primarily E47, for differentiation of different classes of neurons. The key finding is an unexpected discrepancy in how E47 modulates the neurogenic activity of Ascl1/Atoh1 vs. Neurog1/Neurog2. E47 potentiates the activities of Ascl1/Atoh1 while showing an inhibitory effect on Neurog1/Neurog2 in some assays. It is further argued that Neurog1/Neurog2 but not Ascl1/Atoh1 may primarily function as homodimeric complexes and that E47 addition may shift the capacity of the bHLH proteins to bind to certain E box motifs, favoring interactions with CAGSTG sites vs. CADATG sites. Lastly, the authors show that the differential cooperativity of Ascl1 and E47 vs. Neurog1/2 and E47 also holds true in the embryonic cortex, suggesting that this may be a widespread phenomenon.

Overall, the study makes some interesting and important observations. First, E proteins do not universally increase the activity of proneural bHLH proteins, a presumption that comes from classic studies of the cooperativity of myogenic bHLH factor and E proteins. Second, the study argues that certain proneural bHLH proteins such as Neurog1 and Neurog2 may function as homodimeric complexes while others are heterodimeric, which has implications for whether the complex could be regulated by Id proteins. However, the study also has significant weaknesses, most notably inconsistencies with some results and the conclusions/models that are drawn. In addition, while it is suggested that the DNA binding specificity of Neurog1/2 may be altered by the addition of E47, conclusive proof is not provided. Lastly, the studies showing the importance of E47 for neurogenesis are based on misexpression studies without complementary loss of function data. It thus remains unclear whether Neurog1/Neurog2 functions are indeed independent of E protein activities as proposed.

1) A central premise of the study is that BMP signaling differentially regulates neurogenesis by controlling Id2 expression. However, the Id2 shRNA experiments in Figure 2 seems to show broader effects of Id2 loss encompassing both Ascl1/Atoh1 and Neurog1/Neurog2-associated groups. How does one reconcile this discrepancy with the overall model?

2) The inhibitory effects of E47 expression appear to be only transient- in Figures4K and 4M, suppression is seen at the 24 h time point, but not at 48 h. How does this difference play into the central model?

3) Evidence that E47 inhibits Neurog1/2 function solely comes from gain of function experiments, which can be artifactual. The authors should examine the consequences of E47 loss using shRNA or CRISPR mutation approaches. If it is true that Neurog1/2 act as homodimers, then there should be little change in the differentiation of the Neurog1/2-associated neuron groups, while Ascl1/Atoh1-associated groups should be severely impacted.

4) The stoichiometry of various proneural proteins and E protein needs to be carefully considered- too much of either the proneural factor or E protein could throw off activities. Ideally, the amount used should be titrated to identify the optimal amounts to be used. It might be further useful to judge whether equivalent amounts of each protein are present in the experiments using epitope tagged versions of the proteins and immunoblotting (or semi-quantitative fluorescence microscopy).

5) The authors show that another E protein, Tcf12 is also expressed in the spinal cord, yet few experiments are performed to explore its relevance. Does Tcf12 similarly provide enhancement of Ascl1/Atoh1 function while inhibiting Neurog1/2?

6) The data collectively suggest that the presence of E47 alters the preferences of proneural bHLH proteins for certain E box motifs, yet direct proof remains lacking. These conclusions could be strengthened by the inclusion of DNA binding experiments (EMSA) showing that the mixing of proneural proteins with E47 changes their affinity for certain E boxes as is implicated in the Figure 7 model. Id proteins could also be added in to test whether certain complexes (such as Neurog1/2 homodimers) are indeed resistant to Ids. Another way to achieve these results might be to build out the use of reporter constructs adding modified versions of the pkE7 construct containing various E box motifs and ask whether there are switches in the activity levels of the different reporters.

7) The ChIP studies shown in Figure 5 are performed in 293 cells, while all of the luciferase assays and cell differentiation experiments are conducted in the chick spinal cord or brain. It would be best to perform ChIP in a system where some signs that the proneural bHLH factors with or within E47 are transcriptionally active.

8) The studies here do not address the secondary wave of proneural proteins, i.e. Neurod1 and Neurod4 which come on as cells begin to differentiate. Is the expression and/or function of these factors also impacted by E47 presence/absence and Id2 inhibition?

https://doi.org/10.7554/eLife.37267.023

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

In the manuscript by Le Dreau, Escalona and colleagues, the authors have assessed the function and mechanisms of action of E proteins and bHLH transcription factors during neurogenesis and neural fate specification. Using predominantly chick neural tube electroporation assays, as well as molecular approaches, the authors present a model in which E proteins have differential effects on neurogenesis, depending whether they interact with ASCL1/ATOH1 versus NEUROG1/2 bHLH transcription factors. In addition, the authors argue that Bmp signaling contributes to this specificity through induction of ID2 expression, which inhibits the activity of ASCL1/ATOH1 bHLHs.

The authors have performed an impressive number of experiments to generate a model of how various HLH proteins interact to generate the appropriate number of neurons of a specific fate. However, I found the amount of data somewhat overwhelming, and in places disjointed with the overall point the authors are trying to make. In many places, I do not think the authors went far enough to support their model or provide enough data to convince me their results have biological meaning.

R1C1) In Figure 1 the authors show that inhibiting Bmp5 or Smad signaling leads to a selective loss of p2-domain interneurons, as assessed by Chx10 and Gata3 expression. These neurons are defined by Ascl1 expression in progenitors.

We wish to precise that inhibiting BMP7 or SMAD1/5 activity does not only cause a reduction of p2-derived V2a/V2b interneurons in the ventral part of the developing spinal cord, but also markedly reduces the generation of dorsal dI1, dI3 and dI5 interneurons (previously published in Le Dreau et al., 2012 and presented again in Figure-1B).

R1C2) First, it is unclear to me why these cells would be sensitive to Bmp loss, as most Bmps are normally expressed in the dorsal region of the spinal cord.

While it is true that the expression of BMP ligands and their canonical signaling (dependent on SMAD1/5/8 activity) are confined to the most dorsal part of the developing spinal cord at early stages during neural patterning, the expression of several BMP ligands, such as BMP7, spreads ventrally in the ventricular zone at the onset of interneuron neurogenesis around HH18 in chick embryos (see Annex 1C taken from Le Dreau et al., 2012). Concomitantly, the canonical BMP activity, as assessed by in ovo electroporation of a SMAD1/5/8-responsive GFP reporter, can be detected throughout the Dorsal-Ventral (D-V) axis of the developing spinal cord (see Annex 1D and Annex 2G taken from Le Dreau et al., 2014). We believe that these previously published results are strong evidences that the canonical BMP pathway is active throughout the D-V axis of the developing spinal cord from the onset of interneuron generation.

R1C3) Second, what is the evidence that p2-domain cells are normally responding to Bmp? Is there detectable expression of phospo-Smad1/5 in this domain?

In addition to the arguments mentioned above, we had moreover reported that phospho-SMAD1/5/8 staining (which specifically detects the active form of SMAD1/5/8) can be observed in mitotic spinal progenitors along the entire D-V axis from the onset of interneuron generation (see Annex 3A-G taken from Le Dreau et al., 2014). In addition, work from Misra et al., reported that the generation of V2a and V2b interneurons from p2 progenitors is dependent on BMP signaling (Misra et al., 2014, reference included in our manuscript, subsection “The canonical BMP pathway differentially regulates the generation of spinal neurons derived from progenitors expressing ASCL1/ATOH1 or NEUROG1/NEUROG2/PTF1a”). They showed for instance that in ovo electroporation of the extracellular BMP antagonist Noggin markedly reduces the generation of both Chx10+ and Gata3+ neurons. Altogether, we believe that these previously published results demonstrate that p2 are submitted to BMP signaling during neurogenesis and depend on it for proper V2a/b generation.

R1C4) Finally, what happens to expression of Ascl1 under these conditions? Most of the markers analyzed are expressed in postmitotic cells, and it is unclear what is happening to the identity of spinal progenitors in these experiments.

In a previous work, we demonstrated that the activity of the canonical BMP pathway spreads ventrally at the onset of interneuron generation (Le Dreau et al., 2012). We further brought evidences that this canonical BMP activity is not required for maintenance of neural patterning as first anticipated (see Annex 4). When we analyzed the expression pattern of the proneural bHLH genes in response to modulations of canonical BMP signaling (using sh-Smad1/5 and constitutively active SMAD1/5 mutants), we did not observe any significant alteration of their D-V pattern (see Annex 5). Later on, we demonstrated that the primary function of the canonical BMP pathway during spinal neurogenesis is to dictate the mode of divisions of spinal progenitors: high levels of SMAD1/5 activity promote symmetric proliferative PP divisions, whereas lower levels enable spinal progenitors to undergo neurogenic divisions (Le Dreau et al., 2014).

Altogether, we believe that the issues raised by the reviewer #1 in his/her first comments were largely addressed and answered in previous studies.

R1C5) Figure 2 addresses the role of Id2 proteins, where the authors first state the Bmp is necessary and sufficient to promote Id2 expression in the spinal cord. This is not actually shown for the endogenous gene but instead using a Id2 promoter-luciferase reporter that was electroporated within the spinal cord. So, I think this is somewhat overstated in the text. Based on the data in Figure 1, one also wonders what happens to endogenous Id2 expression after Bmp7 and Smad1/5 knockdown.

In our previous work, we reported the effects of both SMAD1/5 gain-of-function and inhibition of the canonical BMP activity (mediated by in ovo electroporation of Noggin) on the endogenous expression of the four ID members, which showed that expression of cId1, cId2 and cId3 is dependent on and positively modulated by the canonical BMP pathway (see Annex 6 taken from Le Dreau et al., 2014). We moreover showed that cId2 transcript levels in PP, PN and NN progenitors follow a gradient that correlates well with the gradient of endogenous SMAD1/5 activity (see panels H-J from Annex 3). These results already argued for cId2 being directly regulated by the canonical BMP pathway during spinal neurogenesis. While we agree with the reviewer #1 that the additional experiments performed in our present study with the pId2:luc reporter do not demonstrate per se that cId2 is a direct target of the canonical BMP pathway, they however reinforce this idea. These results taken altogether support the notion that ID2 might participate in mediating the function of the canonical BMP pathway during spinal neurogenesis, which we believe was the essential demonstration required to test the involvement of ID2 in this process.

R1C6) The main part of this Figure analyzes the effects of Id2 misexpression and inhibition, where the authors find that Id2 misexpression inhibits the generation of most ventral cell types, while knockdown of Id2 is argued to cause premature differentiation. The biological relevance of these experiments is unclear to me, largely because there is clearly a domain in the ventral spinal cord that is lacking in Id2 expression (Figure 2B-D). If this domain includes the p2/Ascl1 domain then it hard to understand how these experiments provide insights into what is happening naturally.

As stated by the reviewer #1, knockdown of cId2 causes an overall premature neuronal differentiation. Conversely, ID2 gain-of-function causes an overall delay/reduction in neuronal differentiation (rather than specifically on ventral cell types). We believe that these results are relevant and fit well the classical role assigned to ID factors in inhibiting differentiation.

Concerning the expression pattern of cId2, we agree with the reviewer #1 that cId2 transcripts were not detected in the ventral part of the developing spinal cord at HH14 (Figure 2A), a stage at which neuronal production has not started yet. We added this panel to illustrate that cId2 expression is restricted to dorsal domains at early developmental stages (when neural patterning is being established) as is the canonical BMP activity (Le Dreau et al., 2012; Tozer et al., 2013). Around the onset of neurogenesis, cId2 transcripts appear expressed also in ventral areas, in correlation with the ventral spreading of cBmp7 expression and the canonical BMP activity (as explained above in the answer to the points R1C2-4). This ventrally-spread expression pattern was illustrated with the images corresponding to Figure 2C-D.

To answer more accurately the reviewer #1’s concern, we performed additional experiments and now bring new data that show unequivocally that cId2 transcripts and protein are expressed by all ventral progenitor domains during spinal neurogenesis, except the pMN domain (new Figure—figure supplement 2). The manuscript has been modified accordingly (subsection “ID2 acts downstream of the canonical BMP pathway to differentially regulate the generation of spinal neurons derived from progenitors expressing ASCL1/ATOH1 or NEUROG1/NEUROG2/PTF1a”). We hope these precisions will satisfy the reviewer #1’s concern.

R1C7) In the next experiments the authors asked whether Id2 contribute to neurogenesis through sequestering E proteins. The authors show that overexpression of E proteins increases the production of differentiated cells, an activity that can be inhibited by co-expression of Id2. They also expressed a dominant negative E protein that inhibited neural differentiation, and this effect was lost when Id2 was depleted.

We understand from this comment that the reviewer #1 was convinced by this part of the study.

R1C8) In Figure 4, the authors argue that E47 has selective function in the generation of neurons derived from ASCL1+ domains. This is partly based on experiments expressing dnE47 that the author's state has a more selective effect on ASCL1 derivatives. The quantification of these experiments is based on ratios of total neurons from electroporated and non-electroporated sides of the spinal cord. These results could conceivably reflect the dorsal-ventral gradients of electroporation efficiencies that one normal sees. Ideally one would want to know how these data look in terms of electroporated (GFP+) neurons that expressed a specific marker. If for example, if motor neurons are unaffected, one would expect to see a larger percentage of Isl1+/GFP+ than Chx10+/V2+ neurons in these experiments.

Indeed, to inhibit the endogenous activity of E47 and E proteins at large, we opted for a dominant-negative mutant of E47 (E47bm), given that this mutant construct has been reported to act as a dominant-negative over not only E47 but also HEB/TCF12 (as demonstrated by Zhuang et al., 1998, and explained in our manuscript, Subsection “ID2 and E proteins counterbalance each other’s activity during spinal neurogenesis”). We took care of sub-cloning this mutant into a pCIG backbone (which harbors high activity when electroporated in ovo) and first confirmed that this E47bm mutant was able to counteract the premature differentiation caused by both E47 and TCF12 (as shown in Figure 3—supplement 1). These latter results led us to consider that E47bm could be used to inhibit the endogenous activity of the different E proteins expressed in the developing spinal cord (and not only E47).

As presented in Figure 4B-F, electroporation of E47bm strongly impaired the generation of neurons derived from progenitors expressing ATOH1 or ASCL1 and to a lesser extent those derived from progenitors expressing NEUROG1, PTF1a or NEUROG2. As explained by the reviewer #1, the results are presented as the ratio of numbers of cells expressing a specific neuronal marker in the electroporated side vs the number of cells expressing this marker in the contral-lateral (control) side. In our experience, this type of quantification is classical in studies of spinal neuronal specification (see for instance the work from Jane E. Johnson’s lab, an expert in the field) and presents the advantage of showing results in a very simple and visual manner, when the missexpression generates a strong phenotype as the one caused by E47bm electroporation.

In the revised version of our study, we decided to provide an additional set of data that, we believe, answers the reviewer #1’s concern as well as similar concerns raised later on by the reviewers #2 and #3 (points R2C4, R3C2 and R3C5). In addition to E47bm, we assessed the phenotype caused by in ovo electroporation of another dominant-negative construct of E47: E47Δnls-RFP (see the panels G-R from the new Figure 4). This construct of E47 fused to RFP, inserted in a plasmid with low electroporation efficiency, consists of a version of E47 deleted from its nuclear localization signals which thereby impairs the nuclear import of E47, hence its transcriptional activity (Mehmood et al.et al., 2009). As previously reported in vitro, the subcellular localization of this E47Δnls-RFP mutant after in ovo electroporation was mostly cytoplasmic (Figure 4G-H’). As seen with E47bm, this E47Δnls-RFP mutant also impaired neuronal differentiation in a cell-autonomous manner (Figure 4G-I). As suggested by the reviewer #1, we analyzed the consequences of E47Δnls-RFP electroporation on the generation of spinal neuron subtypes by quantifying the proportions of differentiated electroporated cells (by focusing on the RFP+ cells that were HuC/D+ or Sox2-) that express one of the neuronal subtype markers (Figure 4J-R). Compared to a control plasmid, electroporation of this E47Δnls-RFP mutant reduced about half the proportions of differentiated electroporated cells that express Lhx2/9 or Tlx3 (Figure 4J-M and R), indicating that inhibiting E47 activity hindered the differentiation of spinal progenitors expressing ATOH1 or ASCL1 alone. By contrast, spinal progenitors electroporated with the E47Δnls-RFP mutant efficiently differentiated into Lhx1/5+ interneurons or Isl1+ motor neurons (Figure 4N-R). Of note, we could even observe Lhx1/5+ or Isl1+ electroporated cells within the ventricular zone (stars in Figure 4O-O’, Q-Q’), indicative of a premature differentiation of NEUROG1/2/PTF1a-expressing progenitors when E47 activity is impaired. Therefore, the results obtained by inhibiting E47 activity using this new mutant are in agreement with the phenotypic consequences expected by the reviewer #1. They are also in agreement with the results previously obtained with the E47bm mutant, further suggesting that E47 activity is required to avoid NEUROG1/2-expressing progenitors to differentiate prematurely.

R1C9) In Figure 5 the authors provide evidence that the specificity of ASCL1/E47 action are due to its ability to selectively activate sequences containing CAGSTG sequences. E47 also inhibited the ability of NEUROG1 activate its CADATG targets, suggested that E47 can functionally inhibit NEUROG1 at certain sequences.

We understand from this comment that the reviewer #1 was convinced by this part of the study.

R1C10) Finally, the authors tested whether differentiation interactions of ASCL1 and NEUROG1 with E47 might operate in other regions of the CNS. The authors show that E47 enhances the ability of ASCL1 promote neuronal differentiation in the cortex. In contrast they found that E47 blocks the ability of NEUROG1 to promote cortical neuronal differentiation. As with many other experiments in this study, the biological relevance of these interactions are unclear, as we are given no detailed information on the temporal or spatial expression profile of E47 during cortical development.

The idea behind this set of experiments was to try to determine whether the differential cooperation between E47 and ATOH1/ASCL1 and NEUROG1/2 that we had observed in the context of spinal neurogenesis is specific of this region of the developing CNS or if it could be generalized. We believed that the functional assays shown (now in Figure 7I-R and Figure 7—figure supplement 1C) answered this question. It is however true that we did not provide any information about the temporal or spatial expression of E47 in the developing chick cortex. To rectify this shortage of information, we analyzed the expression of cTcf3/cE2A by in situ hybridization. These new results (presented in Figure 7E-E’ and Figure 7—figure supplement 1B-B”) show that cTcf3 transcripts are found located in the ventricular zone throughout the D-V axis of the developing chick telencephalon from the beginning of neurogenesis (E3) until at least mid-cortigenesis (E5). This expression pattern is similar to the one described for its murine orthologue during early corticogenesis (Li et al., 2012).

R1C11) The authors have provided detailed information on the statistical methods used in the methods section, and n values for samples in the Figure legends.

We understand from this comment that the reviewer #1 is satisfied with the statistical information provided.

Reviewer #2:

This is a interesting study that puts forward a new mechanistic model to understand how different/limiting levels of E-proteins could have differential effects on neurogenesis through modulating activity of proneural bHLH proteins. The model is supported by manipulating levels of BMP signaling, ID levels, and E-protein/proneural bHLH ratios in the chick neural tube and assessing consequences to division patterns and neuronal subtype specification during embryogenesis. Additional experiments assess the different activities on the proneural bHLH factors with and without the E-protein E47 with respect to transcriptional activity of reporter genes with different motifs. The study is written clearly and data are shown carefully and thoroughly for the most part. The ChIP data are not particularly strong, and the activities rely heavily on over expression assays, nevertheless, multiple types of experiments are used to develop the model and support the conclusions presented.

R2C1) The authors come up with the phrase 'dual co-operation of E proteins with proneural proteins' to describe the differential consequences to neurogenesis that result when E-proteins are limiting. It is a confusing description. Based on the model put forward, it is a differential activity of homodimers of ASCL1/ATOH1 versus NEUROG1/NEUROG2 that form when E-proteins are limiting reflecting the intrinsic binding preferences of the different proneural bHLH TFs. To call this differential co-operation of E-proteins does not really help understand what is going on. I recommend changing how this is stated throughout the manuscript.

We totally agree with the reviewer #2 in that the different functional outcomes caused by modulating E proteins’ activity in the ASCL1/ATOH1-expressing versus NEUROG1/2-expressing neural progenitors indeed originate from the differences in intrinsic binding preferences displayed by the distinct proneural proteins per se. We came up with the concept of “dual co-operation of E proteins with proneural proteins” to highlight the notion that E proteins modulate in opposite ways the activities of ASCL1/ATOH1 versus NEUROG1/2. However, we understand the concern of the reviewer #2 as “dual co-operation” might be understood as “dual mechanism of action”, which would indeed be mistaken. Taking this comment into consideration, we thus decided to erase the term “dual co-operation” from our manuscript. To avoid causing ambiguity in the molecular mechanisms at play between the proneural and E proteins, we limited the use of the terms “differential co-operation” throughout the manuscript. In particular, this expression was removed from the subheadings of the Results section and from the title of our study, henceforth entitled “E proteins sharpen neurogenesis by modulating proneural bHLH transcription factors’ activity in an E-box-dependent manner”. We hope that these changes make our message clearer and that they satisfy the reviewer #2’ concerns.

R2C2) Some of the differences in activities could be affected by different levels of the proneural proteins made from the expression constructs. Assessment of the protein levels will be difficult to compare because no tag is included and the antibodies can work at different levels. Minimally it should be considered. Ideally it should be evaluated. This only becomes important because arguments are made about the different effects of E-proteins on the factors. It may be possible that the protein levels of the proneurals are substantially different and are showing differential effects in these assays. This could be true even though pCIG is used for them all.

Indeed, we realized along the course of this study that differences in expression levels of the different proneural and E protein constructs might impact the results obtained and their interpretation. That is why we had decided to sub-clone all these constructs in a similar backbone and thereby get comparable levels of expression of the different proteins (as explained in the Materials and methods section). As shown in the Western blots presented in the Figure 6—figure supplement 3A-C, transfection of the different constructs of ASCL1, NEUROG1 and E47 and their respective hetero- and homodimers (all inserted under the control of the same promoter) yielded the expression of protein amounts that, if not similar, appear at least to be in the same expression range.

To further ensure that differences in the concentrations of electroporation might not alter the functional outcome, we performed dose-dependent analyses for each proneural protein (tested at 0.1 and 1 µg/µl in the developing spinal cord and 0.5 or 1.0 µg/µl in the developing cerebral cortex). These control experiments were presented in the Figure 4—figure supplement 1A and Figure 6—figure supplement 1 (henceforth Figure 5—figure supplement 1 A and Figure 7—figure supplement 1) and explained in the manuscript (subsection “E47 modulates in opposite ways the neurogenic abilities of ASCL1/ATOH1 and NEUROG1/NEUROG2 during spinal neurogenesis” and subsection “E47 modulates in opposite ways the neurogenic abilities of ASCL1 and NEUROG1/NEUROG2 during corticogenesis”). As can be observed, while the concentration did impact on the extent of the effects caused by the different proneural proteins, in no case did it change the way these TFs affected differentiation (always increasing the proportion of electroporated cells that differentiated into neurons). Based on these data, we thus decided to test how the combined electroporation of E47 would affect the effects caused by the sub-optimal concentration of each proneural protein, in order to get a wider range of putative modulation. We apologize for not making it clear enough in the manuscript and have added this precision in the revised version (subsection “ID2 and E proteins counterbalance each other’s activity during spinal neurogenesis”).

In the case of luciferase assays (Figure 6B-E and Figure 6—figure supplement 2B-I), the effects of ASCL1 or NEUROG1 alone were assessed at both 0.5 and 1.0 µg/µl to ensure that the effects caused by E47 addition (electroporated at 0.5 µg/µl) were not simply resulting from an increase in the total concentration.

Therefore, while we agree with the reviewer #2 that differences in the expression levels of the distinct bHLH proteins might somewhat affect the results obtained in terms of strength of effect (promoting more or less differentiation), we believe that we did consider this issue and performed and provided control conditions and experiments which are thorough enough to support the validity of our main findings.

R2C3) Also related to this issue, in the Discussion section it is stated that: "Such co-operation appears to be particularly crucial in the case of ASCL1, whose overexpression could barely increase neurogenic divisions per se." This contrasts with other studies ie Nakada 2014 that show Ascl1 is very efficient at increasing differentiation even at 24 hrs. Is it possible that expression levels are relatively low from the construct used here relative to Nakada, or relative to the other bHLH in the current study as mentioned above? This may need some consideration or ruled out by determining the comparative levels of the overexpressed bHLH proteins.

We could not find a reference “Nakada et al., 2014” in the literature that assessed the function of ASCL1. We believe that the reviewer #2 might be referring to the study by “Nakada et al., 2004” from Jane E. Johnson’s lab in which the authors assessed the consequences of in ovo electroporation of ASCL1 (named MASH1 at that time) on neuronal differentiation in the developing chick spinal cord (Nakada et al., 2004, reference already cited in our manuscript). As shown in their Figure 1Q, about 55% of the cells electroporated with ASCL1 were differentiated into neurons after 24 hours (using an ASCL1 construct inserted into a pMiwIII plasmid electroporated at 2 µg/µl as stated in their Materials and methods section). We obtained 45% of EP+;HuC/D+ cells 24 hours after in ovo electroporation of ASCL1 at 0.1µg/µl and 57% after 48 hours (Figure 5—figure supplements 1B and C, respectively). We believe that the results obtained in these two independent studies show a comparable efficiency of ASCL1 at increasing differentiation and suggest that the lower proportion of differentiated cells obtained in response to ASCL1 electroporation in our study is a consequence of a diluted concentration of electroporation, and not a defect of the construct we used.

We also bring to the attention of the reviewers another study in which the authors compared the ability of ASCL1 and NEUROG1/2 to induce neuronal differentiation after adenoviral transfection of mouse cortical progenitor cells in vitro(Ge et al., 2006). They reported that infection with NEUROG1/2 caused nearly 100% of cells to differentiate into neurons after 24-48 hours, whereas only ≈25% of the cortical progenitors had differentiated into neurons after infection with ASCL1 (see their Figure 1C). Interestingly, ASCL1 showed an ability to promote migration comparable to that of NEUROG1/2 (see their Figure 1D), arguing against the possibility that the distinct differentiation-inducing capacities of ASCL1 and NEUROG1/2 observed in their assay were due to differences in adenoviral infection efficiencies. This is thus another independent study performed in another neural context that reported, as the authors themselves stated, that ASCL1 has less-potent neurogenic effect than NEUROG1/2 when misexpressed in neural progenitors.

R2C4) All the results in the study are from E47 but the conclusions are generalized to E-proteins. Was HEB tested in these assays? This should be shown if the generalization is being used.

We presented results showing that HEB/TCF12 overexpression promotes neuronal differentiation comparably to E47, and that the effects of both could be rescued by ID2 co-electroporation (Figure 3D-J). As explained in our answer to the point R1C8, we moreover demonstrated that the effects of TCF12 overexpression could be rescued by co-electroporation with the dominant-negative mutant E47bm, which led us to consider that E47bm could be used to inhibit the endogenous activity of the different E proteins expressed in the developing spinal cord.

In this revised version of the manuscript, we moreover provide new results that show that TCF12 addition is able to enhance ASCL1-dependent pKE7:luc activity while inhibiting NEUROG1’s ability to induce pNeuroD:luc activity, similarly to E47 though with milder capacities (Figure 6—figure supplement 2D,E, cited in subsection “E47 modulates the transcriptional activities of ASCL1 and NEUROG1 in an E-box dependent manner and through physical interactions”).

We believe that these results altogether support the notion that E47 and TCF12 modulate the activity of the proneural proteins in a comparable manner. That is why we took the liberty of using the general term “E proteins” in our final model and conclusions. However, if the editor and reviewers consider it more appropriate, we would accept to tone down this generalization in our manuscript and focus our findings on E47.

Reviewer #3:

In this study, LeDréau et al., explore how differences in the formation of proneural bHLH-E protein transcription factor complexes can influence neuronal differentiation in the developing spinal cord and cortex. A link to BMP influences on this process is further explored though an examination of the function of Id2, a BMP-regulated inhibitor of proneural bHLH-E protein complexes. Some of the initial observations include findings that shRNA knockdown of Bmp7, Smad1, or Smad5 potently suppresses the formation of certain classes of spinal neurons (dI1, dI3, and V2a/b) while having less of an effect on other classes (dI6, V0, V1, and MNs). These trends are argued to align with the association of these two groups with the activity of the proneural factors Ascl1 and Atoh1 vs. Neurog1 and Neurog2. The connection of BMP signaling to proneural genes is supposed by evidence that BMP7/Smad1/5 can regulate Id2. shRNA knockdown of Id2 appears to broadly enhance neurogenic differentiation.

The focus of the study then shifts to an exploration of the requirement and sufficiency of E proteins, primarily E47, for differentiation of different classes of neurons. The key finding is an unexpected discrepancy in how E47 modulates the neurogenic activity of Ascl1/Atoh1 vs. Neurog1/Neurog2. E47 potentiates the activities of Ascl1/Atoh1 while showing an inhibitory effect on Neurog1/Neurog2 in some assays. It is further argued that Neurog1/Neurog2 but not Ascl1/Atoh1 may primarily function as homodimeric complexes and that E47 addition may shift the capacity of the bHLH proteins to bind to certain E box motifs, favoring interactions with CAGSTG sites vs. CADATG sites. Lastly, the authors show that the differential cooperativity of Ascl1 and E47 vs. Neurog1/2 and E47 also holds true in the embryonic cortex, suggesting that this may be a widespread phenomenon.

Overall, the study makes some interesting and important observations. First, E proteins do not universally increase the activity of proneural bHLH proteins, a presumption that comes from classic studies of the cooperativity of myogenic bHLH factor and E proteins. Second, the study argues that certain proneural bHLH proteins such as Neurog1 and Neurog2 may function as homodimeric complexes while others are heterodimeric, which has implications for whether the complex could be regulated by Id proteins.

R3C1) However, the study also has significant weaknesses, most notably inconsistencies with some results and the conclusions/models that are drawn. In addition, while it is suggested that the DNA binding specificity of Neurog1/2 may be altered by the addition of E47, conclusive proof is not provided.

To demonstrate that E47 restrains the DNA binding and transcriptional activity of NEUROG1 on CADATG motifs, we presented results obtained from luciferase (Figure 6E, Figure 6—figure supplement 2C) and ChIP assays (Figure 6G). In the revised version presented herein, we bring novel results obtained using TCF12 (Figure 6—figure supplement 2E), an additional luciferase reporter (Figure 6—figure supplement 2G) and new experimental conditions presenting the effects of ID2 addition on top of NEUROG1 and E47 (Figure 6—figure supplement 2I). To further demonstrate that these results were due to a direct/physical interaction between E47 and NEUROG1, we provided luciferase assays in which we compared the activity of tethered constructs of NEUROG1 homodimers and NEUROG1-E47 heterodimers (Figure—figure supplement 3E). We moreover provided results showing the phenotypical consequences of this modulation of NEUROG1/2 transcriptional activity by E47 on the mode of division (using the pSox2:luc, pSox2GFP and pTis21:RFP reporters as shown in the Figures henceforth renamed 5L-P and Figure 5—supplement 1D), the final EP+ cell number (Figure 5I) and the proportion of neuronal differentiation (Figure 5E,F and Figure 5—supplement 1B, C) as well as results obtained with the tethered NEUROG1-NEUROG1 and E47-NEUROG1 dimers (Figure 6K-N). We sincerely believe that these pieces of evidence put altogether are consistent and strongly support our conclusions.

R3C2) Lastly, the studies showing the importance of E47 for neurogenesis are based on misexpression studies without complementary loss of function data. It thus remains unclear whether Neurog1/Neurog2 functions are indeed independent of E protein activities as proposed.

As explained in details in the answer to the point R1C8, we had opted for a dominant-negative strategy to inhibit the endogenous activity of E47 and E proteins at large, based on the literature suggesting that the consequences of E47 loss-of-function might have been masked and compensated for by the activity of other E proteins expressed in our system (Zhuang et al., 1998). In our opinion, the results obtained with the E47bm mutant (shown in Figure 4B-F) are in agreement with the phenotype expected from our hypothesis.

In addition, we now provide a new set of experiments performed with another dominant-negative E47 mutant (E47Δnls-RFP), as explained in details in our answer to the point R1C8 (please head for this section to read the description of these new results). Although these results represent another misexpression strategy instead of a loss of function strategy as asked by the reviewer #3, we believe that the results obtained and presented in the Figure 4G-R reinforce our conclusions, and we hope that they will satisfy the concerns of the reviewers #1 and #3.

R3C3) A central premise of the study is that BMP signaling differentially regulates neurogenesis by controlling Id2 expression. However, the Id2 shRNA experiments in Figure 2 seems to show broader effects of Id2 loss encompassing both Ascl1/Atoh1 and Neurog1/Neurog2-associated groups. How does one reconcile this discrepancy with the overall model?

We agree with the reviewer #3 on the fact that sh-Id2 in ovo electroporation affected one particular neuronal subtype deriving from NEUROG1-expressing progenitors more than sh-Bmp7 or sh-Smad1/5 (V0v neurons specifically, Figure 2O) and less markedly several neuronal subtypes deriving from ASCL1/ATOH1-expressing progenitors (Figure 2O). Nevertheless, the overall tendencies of these phenotypes are similar enough to be noticed (Figure 2O). It actually surprised us that the phenotype caused by sh-Id2 would reproduce the overall phenotype caused by BMP7 or SMAD1/5 inhibition. We were expecting that the reduced activity of ID2 might have been compensated for by the activity of other IDs, which are expressed in overlapping D-V progenitor domains (see Annex 6).

As explained in our answer to the point R1C5, results from our previous study indeed demonstrated that cId1, cId2 and cId3 are regulated by the canonical BMP pathway during spinal neurogenesis (see Annex 6) and brought evidence that at least cId2 and cId3 (it was less evident for cId1) might represent direct targets of SMAD1/5 mediating their function on the control of the mode of division of spinal progenitors (see panels H-I from Annex 3). In order to get a proof of concept that the ID factors mediate the function of the canonical BMP pathway during spinal neurogenesis, we had chosen to focus on ID2 for the reasons mentioned above and the fact that cId2 presents an expression pattern broader than cId3 for instance. Our intention was to say that ID2 represents one of the factors regulated by the canonical BMP activity and involved in mediated BMP function during spinal neurogenesis, not that ID2 is the sole factor involved. We apologize if we did not make this point clear enough. So, differences in the phenotypes obtained with sh-Id2 as compared to the ones obtained with sh-Bmp7/Smad1/5 are likely to be due to the fact that the canonical BMP activity is mediated not only by ID2 but by additional targets of the pathway (possibly other IDs).

On the other hand, it is also very likely that the expression of cId2 is not regulated solely by the canonical BMP pathway and is probably modulated by additional signaling pathways during spinal neurogenesis. For all these reasons, we did not expect that the phenotype caused by ID2 loss of function would match 100% the one caused by BMP7 or SMAD1/5. However, we believe that they are similar enough to support the idea that ID2 acts downstream of the canonical BMP pathway.

R3C4) The inhibitory effects of E47 expression appear to be only transient- in Figures 4K and 4M, suppression is seen at the 24 h time point, but not at 48 h. How does this difference play into the central model?

Rather than the effect of E47 being only transient, the fact that the inhibitory effects of E47 on NEUROG1-induced differentiation are observed after 24 hours and not anymore after 48 hours results from the way NEUROG1, alone or in combination with E47, affects the mode of division of spinal progenitors (Figure 5L-P). To make our explanation more explicit, we propose you a simulation of the effects of NEUROG1 overexpression alone or combination with E47 on the progressive differentiation of the pool of electroporated cells (Annex 8).

Let assume that 10 spinal progenitors (white circles) have integrated the plasmid at the time of in ovo electroporation. Under the effect of NEUROG1 overexpression, these 10 progenitors will undergo symmetric proliferative (PP, green circle), asymmetric (PN, yellow circle) or self-consuming neurogenic divisions (NN, red circles) according to the proportions assessed with the pSox2:GFP and pTis21:RFP reporters (results presented in Figure 5P). The outcome of this first round of division results in the generation of 7 new progenitors (resulting from the PP and PN divisions) and 13 daughter cells committed to neuronal differentiation (resulting from the PN and NN divisions, black circles). By definition, these differentiating neurons won’t divide anymore. As specified in the bottom of the panel A, this first round of division generated 20 cells in total, of which 13 are already differentiating. This represents a proportion of EP+ cells that differentiated into neurons of 65%, a proportion that is pretty close to the one quantified 24 hours after NEUROG1 overexpression (Figure 5E). We are thus left with 7 electroporated progenitors for the second round of division, which should all undergo PN or NN divisions based on the proportions of PP/PN/NN applied previously. By following on with this simulation, after a fourth round of divisions we are left with only 1 progenitor cell for a total of 33 electroporated cells present in the system. This represents a proportion of EP+ cells that differentiated into neurons of 97%, a proportion that matches also pretty well the one quantified 48 hours after NEUROG1 overexpression (Figure 5E).

If we apply the same modeling for the co-electroporation of NEUROG1 and E47, the proportions of PP/PN/NN divisions are slightly shifted towards proliferative divisions at the beginning. After the first round of division, we would obtain 9 progenitors and 11 differentiating neurons, representing a proportion of EP+ cells that differentiated into neurons of 45%. Again, this proportion obtained by simulation is pretty close to the one quantified (Figure 5E). At the end of the fourth round, the proportion of differentiated cells among the electroporated cells reaches 93%, which again is pretty similar to the one quantified (Figure 5E). It is also interesting to note that although we started with the same number of spinal progenitors electroporated at the beginning and applied the same experimental duration, the final numbers of EP cells are markedly different, in agreement with the experimental results described (Figure 5I). When E47 is combined with NEUROG1, more progenitor daughter cells have been generated and divided throughout the rounds of division, which fits also very well with the results obtained with the Violet assay (Figure 5K). Altogether, these results illustrate how a slight shift in the proportions of the 3 different modes of divisions can impact the final numbers of neurons that are generated and the pace of differentiation.

We hope that these simulations will help convincing the reviewers of the logic and consistency of the data we presented in Figure 4 (now Figure 5).

R3C5) Evidence that E47 inhibits Neurog1/2 function solely comes from gain of function experiments, which can be artifactual. The authors should examine the consequences of E47 loss using shRNA or CRISPR mutation approaches. If it is true that Neurog1/2 act as homodimers, then there should be little change in the differentiation of the Neurog1/2-associated neuron groups, while Ascl1/Atoh1-associated groups should be severely impacted.

We previously detailed in our answers to the points R1C8 and R3C2 the reasons that led us to opt for a dominant-negative strategy over the use of a sh-RNA targeting cE47. We wish to highlight the fact that the consequences that the reviewer #3 would expect from the use of a sh-RNA are fitting exactly with the phenotype obtained with the E47bm mutant (shown in Figure 4B-F). The new results obtained with the E47Δnls-RFP mutant fit also well with these predictions, with the additional demonstration that a fraction of the electroporated spinal progenitors from NEUROG1/2/PTF1a-expressing domains differentiated prematurely (Figure 4O, Q). These latter results support the notion that within these domains E47 acts to restrain NEUROG1/2 activity in order to prevent a premature differentiation.

As discussed in our manuscript (Discussion section), a recent study reported the consequences of the genetic depletion of E47 on mouse cortical development (Pfurr et al., 2017). This study revealed that E47 depletion resulted in an increased production of cortical excitatory neurons at mid-corticogenesis. Considering that it is assumed that NEUROG2 constitutes the main proneural protein at play during early corticogenesis, these results are consistent with our results and model. They moreover suggest that inhibiting E47 function, whether this is achieved by a loss of function strategy or missexpresion of a dominant-negative mutant, can produce comparable phenotypes.

R3C6) The stoichiometry of various proneural proteins and E protein needs to be carefully considered- too much of either the proneural factor or E protein could throw off activities. Ideally, the amount used should be titrated to identify the optimal amounts to be used. It might be further useful to judge whether equivalent amounts of each protein are present in the experiments using epitope tagged versions of the proteins and immunoblotting (or semi-quantitative fluorescence microscopy).

The reviewer #2 emitted a similar concern (points R2C2, R2C3). Please head for these paragraphs to read our answer.

R3C7) The authors show that another E protein, Tcf12 is also expressed in the spinal cord, yet few experiments are performed to explore its relevance. Does Tcf12 similarly provide enhancement of Ascl1/Atoh1 function while inhibiting Neurog1/2?

In this revised version of the study, we provide additional data showing that TCF12 addition is able to enhance ASCL1-dependent pKE7:luc activity while inhibiting NEUROG1’s ability to induce pNeuroD:luc activity, similarly to E47 though with milder capacities (Figure 6—figure supplement 2D,E, cited in subsection “E47 modulates the transcriptional activities of ASCL1 and NEUROG1 in an E-box dependent manner and through physical interactions”). We believe that these results, together with the results previously presented, support the notion that E47 and TCF12 modulate the activity of the proneural proteins in a comparable manner. However as previously said, if the editor and reviewers consider it more appropriate, we would accept to tone down this generalization in our manuscript and focus our findings on E47.

R3C8) The data collectively suggest that the presence of E47 alters the preferences of proneural bHLH proteins for certain E box motifs, yet direct proof remains lacking. These conclusions could be strengthened by the inclusion of DNA binding experiments (EMSA) showing that the mixing of proneural proteins with E47 changes their affinity for certain E boxes as is implicated in the Figure 7 model. Id proteins could also be added in to test whether certain complexes (such as Neurog1/2 homodimers) are indeed resistant to Ids. Another way to achieve these results might be to build out the use of reporter constructs adding modified versions of the pkE7 construct containing various E box motifs and ask whether there are switches in the activity levels of the different reporters.

As already answered in the point R3C2, we sincerely believe that we brought an amount of data convincing enough to support the notion that E47 modulates the transcriptional activity of ATOH1/ASCL1 and NEUROG1/2 in an E-box-dependent manner. However, we agree with the reviewer #3 that his/her suggestions of additional experiments would reinforce the conclusions of our work and the model presented. For this reason, in the revised version of our study we provide additional data in line with the reviewer #3’s suggestions, and which are included in the revised version of the Figure 6—figure supplement 2.

To strengthen the notion that the E-box context determines the way E proteins modulate the activity of the distinct proneural proteins, luciferase assays were performed using two additional reporters: pDll1-M and pDll1-N, which have been previously described to respectively respond to ASCL1 and NEUROG2 (Castro et al.et al., 2006) and are based on conserved regulatory elements found in the promoter of the gene Δ-like1 (Beckers et al.et al., 2000). The promoters of these pDll1-M:luc and pDll1-N:luc reporters contain 3 CAGSTG + 1 CADATG and 1 CAGSTG + 3 CADATG motifs, respectively (Figure 6—figure supplement 2A). In the context of pDll1-M:luc activity, the addition of E47 had only an additive effect to ASCL1 activity (Figure 6—figure supplement 2F), compared to the synergistic effect observed on pKE7:luc activity (Figure 6B). In the context of pDll1-N:luc activity, addition of E47 also inhibited the induction caused by NEUROG1 (Figure 6—figure supplement 2G), though with milder effects compared to the strong repression observed on pNeuroD:luc activity (Figure 6E). These results reinforce our model whereby the outcome of the co-operation of E47 with the distinct proneural TFs depends on the balance between the CAGSTG and CADATG motifs present in each DNA target region and on the intrinsic binding preferences of the proneural proteins per se.

In addition, we present luciferase assays that include experimental conditions in which ID2 was combined with E47 and ASCL1 or NEUROG1. Addition of ID2 partially rescued both the synergistic effect of E47 on pKE7:luc activity when combined with ASCL1 and the inhibitory effect of E47 on NEUROG1-induced pNeuroD:luc activity (Figure 6—figure supplement 2H,I). These results are in agreement with the notion that IDs act by sequestering E proteins from their interaction with proneural proteins, and reinforced the notion that the context-dependent modulatory effect of E47 on ASCL1 or NEUROG1 activities relies on physical interactions.

We hope that these additional results will satisfy the reviewer #3’s curiosity.

R3C9) The ChIP studies shown in Figure 5 are performed in 293 cells, while all of the luciferase assays and cell differentiation experiments are conducted in the chick spinal cord or brain. It would be best to perform ChIP in a system where some signs that the proneural bHLH factors with or within E47 are transcriptionally active.

We wish to emphasize, as the reviewer #3 recognized himself/herself, that all the results presented in our study, except for the ChIP data, were obtained from in vivo experiments.

As an in vitro system, in contrast to the opinion of the reviewer #3, we actually preferred to use a cell line in which no endogenous proneural protein would be expressed to avoid the possibility that these endogenous proneural proteins might interfere with the designed experiments and bias the results’ interpretation. We believe that the results obtained in these ChIP assays are supporting the results obtained in vivo.

R3C10) The studies here do not address the secondary wave of proneural proteins, i.e. Neurod1 and Neurod4 which come on as cells begin to differentiate. Is the expression and/or function of these factors also impacted by E47 presence/absence and Id2 inhibition?

We agree with the reviewer #3 that, in view of our results, studying how the E proteins and ID factors modulate the activity of NEUROD1/4 would be very interesting. As explained by the reviewer #3, NEUROD1/4 represent another group of bHLH proteins that act downstream of the proneural proteins to promote neuronal differentiation (Guillemot, 2007). But for this very same reason, we truly believe that this question is beyond the scope of our study and should rather be tackled in future studies.

Additional References:

Ge, W., He, F., Kim, K.J., Blanchi, B., Coskun, V., Nguyen, L., Wu, X., Zhao, J., Heng, J.I., Martinowich, K. et al. 2006. Coupling of cell migration with neurogenesis by proneural bHLH factors. Proc Natl Acad Sci U S A 103(5): 1319-1324.

Tozer, S., Le Dreau, G., Marti, E., and Briscoe, J. 2013. Temporal control of BMP signalling determines neuronal subtype identity in the dorsal neural tube. Development 140(7): 1467-1474.

https://doi.org/10.7554/eLife.37267.024

Article and author information

Author details

  1. Gwenvael Le Dréau

    Department of Developmental Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Conceptualization, Formal analysis, Supervision, Investigation, Visualization, Methodology, Writing—original draft
    Contributed equally with
    René Escalona
    For correspondence
    gldbmc@ibmb.csic.es
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6877-3670
  2. René Escalona

    Department of Developmental Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Present address
    Departamento de Embriología, Facultad de Medicina, Universidad Nacional Autónoma de México, México City, Mexico
    Contribution
    Formal analysis, Investigation
    Contributed equally with
    Gwenvael Le Dréau
    Competing interests
    No competing interests declared
  3. Raquel Fueyo

    Department of Molecular Genomics, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7106-7163
  4. Antonio Herrera

    Department of Developmental Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
  5. Juan D Martínez

    Department of Developmental Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  6. Susana Usieto

    Department of Developmental Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  7. Anghara Menendez

    Department of Cell Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  8. Sebastian Pons

    Department of Cell Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Resources, Supervision
    Competing interests
    No competing interests declared
  9. Marian A Martinez-Balbas

    Department of Molecular Genomics, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Resources, Supervision
    Competing interests
    No competing interests declared
  10. Elisa Marti

    Department of Developmental Biology, Instituto de Biología Molecular de Barcelona, Barcelona, Spain
    Contribution
    Resources, Supervision, Funding acquisition, Project administration, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5839-7133

Funding

Asociación Española Contra el Cáncer (AIO2014)

  • Gwenvael Le Dréau

Consejo Nacional de Ciencia y Tecnología

  • René Escalona

Ministerio de Educación, Cultura y Deporte (#FPU13/01384)

  • Raquel Fueyo

Ministerio de Economía y Competitividad (#FJCI-2015-26175)

  • Antonio Herrera

Ministerio de Economía y Competitividad (BFU2014-53633-P)

  • Sebastian Pons

Ministerio de Economía y Competitividad (BFU2015-69248-P)

  • Marian A Martinez-Balbas

Fondation Jérôme Lejeune (Fondation Jérôme Lejeune.2016)

  • Marian A Martinez-Balbas

Ministerio de Economía y Competitividad (BFU2016-81887-REDT)

  • Elisa Marti

Ministerio de Economía y Competitividad (BFU2016-77498-P)

  • Elisa Marti

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank the members of the laboratory for helpful comments on the study. We are grateful to Drs N Ben-Arie, C Birchmeier, J-M Frade, F Giraldez, F Guillemot, TM Jessell, JE Johnson, A Joyner, M Kawaichi, A McMahon, J Muhr, T Müller, S Pfaff, C Pujades, J Slack, Y Yokota, Y Yoneda and Y Zhuang for kindly providing reagents. We also thank the Developmental Studies Hybridoma Bank, developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA, USA. We acknowledge E Rebollo and the IBMB Molecular Imaging platform and J Comas and the PCB Flow Cytometry facility for excellent assistance. This work was supported by the grants to EM from BFU2016-81887-REDT and BFU2016-77498-P.

Senior Editor

  1. Marianne E Bronner, California Institute of Technology, United States

Reviewing Editor

  1. Jeremy Nathans, Johns Hopkins University School of Medicine, United States

Publication history

  1. Received: April 5, 2018
  2. Accepted: August 9, 2018
  3. Accepted Manuscript published: August 10, 2018 (version 1)
  4. Version of Record published: September 6, 2018 (version 2)

Copyright

© 2018, Le Dréau et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 548
    Page views
  • 113
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Chromosomes and Gene Expression
    2. Developmental Biology
    Alexander Munden et al.
    Research Article Updated
    1. Developmental Biology
    2. Stem Cells and Regenerative Medicine
    Anna Yoney et al.
    Research Article