Her6 regulates the neurogenetic gradient and neuronal identity in the thalamus
- Steffen Scholppa,b,1,
- Alessio Delogua,
- Jonathan Gilthorpea,2,
- Daniela Peukerta,b,
- Simone Schindlerb and
- Andrew Lumsdena
- aMRC Centre for Developmental Neurobiology, New Hunt's House, Guy's Campus, King's College London, London SE1 1UL, United Kingdom; and
- bInstitute of Toxicology and Genetics, Institute of Technology Karlsruhe, Postfach 3640, 76021 Karlsruhe, Germany
-
Communicated by Thomas M. Jessell, Columbia University College of Physicians and Surgeons, New York, NY, September 30, 2009 (received for review June 10, 2009)
Abstract
During vertebrate brain development, the onset of neuronal differentiation is under strict temporal control. In the mammalian thalamus and other brain regions, neurogenesis is regulated also in a spatially progressive manner referred to as a neurogenetic gradient, the underlying mechanism of which is unknown. Here we describe the existence of a neurogenetic gradient in the zebrafish thalamus and show that the progression of neurogenesis is controlled by dynamic expression of the bHLH repressor her6. Members of the Hes/Her family are known to regulate proneural genes, such as Neurogenin and Ascl. Here we find that Her6 determines not only the onset of neurogenesis but also the identity of thalamic neurons, marked by proneural and neurotransmitter gene expression: loss of Her6 leads to premature Neurogenin1-mediated genesis of glutamatergic (excitatory) neurons, whereas maintenance of Her6 leads to Ascl1-mediated production of GABAergic (inhibitory) neurons. Thus, the presence or absence of a single upstream regulator of proneural gene expression, Her6, leads to the establishment of discrete neuronal domains in the thalamus.
Neurogenesis in the developing vertebrate CNS is regulated with a high degree of temporal and spatial precision, with stereotypic patterns of neuronal differentiation and extensive neuronal migration (1, 2). Dynamic patterns of mitotically active neuronal precursors, known as ‘neurogenetic gradients’ (3) have been described in several brain regions, including the neocortex (4, 5), the dorsal midbrain colliculi (6), and the dopaminergic region of the ventral midbrain (7). In the mammalian diencephalon, in particular the thalamus (formerly known as dorsal thalamus), two main neurogenetic gradients have been described: from posterior to anterior and from lateral to medial (8, 9). In rodents, all thalamic neurons are generated in about 6 days, and the orthogonal gradients of glutamatergic neurogenesis sweep across the boundaries of future nuclei. The underlying molecular mechanisms responsible for generating the neurogenetic gradients of the thalamus are unknown.
The major constituent of the thalamus is a population of excitatory neurons generated in the caudal thalamus (cTh), whereas a minor population of inhibitory neurons is generated in the rostral thalamus (rTh). The latter is thought to give rise to the reticular nucleus and the ventral lateral geniculate nucleus, including the intergeniculate leaflet (10). During development, this rostro-caudal partitioning is seen in the expression domains of proneural basic helix-loop-helix (bHLH) genes: the achaete-scute-like complex genes (Ascl formerly known as Mash in mouse and Zash in zebrafish) mark the GABAergic rTh and the prethalamus (PTh) and the Neurogenin genes (Neurog, formerly known as Ngn) mark the glutamatergic cTh (2, 11, 12). Several lines of evidence show that these genes function as determinants of transmitter phenotype: in the mouse telencephalon, Neurog1/2 are required to specify the glutamatergic character of cortical neurons, while simultaneously repressing an alternative subcortical GABAergic fate (13, 14), whereas forced expression of Ascl1 in the dorsal forebrain induces ectopic differentiation of GABAergic neurons (15).
Another subfamily of bHLH proteins, the hairy-related Hes/Her proteins, generally function as DNA-binding transcriptional repressors and antagonize proneural gene function (16, 17). Hairy-related proteins form homodimers through the bHLH region and have a conserved WRPW domain at the carboxyl (C) terminus, which functions as a repression domain by recruiting co-repressors of the Groucho family (18). Some of these hairy-related proteins keep cells in a progenitor state, preventing initiation of the neurogenic program and thereby maintaining local organizer populations at signaling boundaries. Such a role for Hes/Her proteins has been described in relation to the mid-diencephalic organizer (MDO) at the intrathalamic boundary (zona limitans intrathalamica; ZLI) and isthmic organizer (ISO) at the midbrain-hindbrain boundary (19–21).
Here we decipher the molecular mechanism leading to temporally controlled thalamic neurogenesis in zebrafish, and uncover a underlying mechanism leading to the correct acquisition of thalamic neuronal identity. We describe a function for the hairy-related gene her6, which is expressed in the entire presumptive thalamic complex at early stages and is subsequently restricted to ascl1-positive neuronal progenitors within the PTh, the rTh and the MDO. The dynamic regression of her6 expression from the cTh is accompanied by the caudal-to-rostral progression of the neurog1 expression, making her6 a candidate regulator of the neurogenic gradient in the glutamatergic thalamus. We show that Her6 blocks neurog1-mediated neurogenesis cell-autonomously by interaction with the co-factor Groucho1. Furthermore, loss of Her6 leads to ectopic induction of neurog1 in both the rTh and the PTh. Conversely, forced expression of Her6 in the cTh switches cells to a GABAergic fate. In an epistatic analysis, we demonstrate genetic suppression of neurog1 by Her6: double knock-down of both genes rescues the single knock-down phenotype of Her6, such as the maintenance of the MDO and the GABAergic population of the rTh.
In summary, we propose that Her6 determines both the spatial progression of neurogenesis through the thalamic territory and the decision to become a glutamatergic relay neuron or a GABAergic interneuron in this region.
Results
The Neurogenetic Gradient in the Thalamus.
The term ‘neurogenetic gradient’ was coined in the 1970s to describe the process of dynamic neurogenesis in the developing mammalian thalamus. To see whether this is a feature of vertebrates in general, we analyzed the expression dynamics of genes marking different stages of neuronal development in the embryonic zebrafish diencephalon: the proneural gene neurogenin1 (neurog1), the neuronal precursor marker deltaA, the postmitotic marker of glutamatergic neurons, the vesicular glutamate transporter 2.1.
To visualize the dynamics of proneural gene expression in the thalamus in vivo, we used a transgenic strategy by generating embryos that express a nuclear-localized form of the red fluorescent protein (RFP) under the control of the neurog1 promoter (22) and green fluorescent protein (GFP) under the control of the shh promoter (23). The proneural gene neurog1 has been shown to mark glutamatergic precursors (15), and shh determines the anterior and ventral limit of the thalamus (Fig. 1 and Movie S1). In the diencephalon at 24 h post-fertilization (hpf), shh:GFP expression was localized to the MDO and the basal plate (Fig. 1A). Between 24 hpf and 33 hpf, we observed an increase in the number of neurog1:RFP+ cells in a high-posterior to low-anterior gradient (Fig. 1 D and G). Over the same period, we also found that the expression of shh and neurog1 is co-localized in the ventral cells that have been suggested to partially form basal forebrain dopaminergic neurons in fish (24). In addition, the progressive activation of neurog1:RFP and neurog1 mRNA are accompanied by the expression of dla mRNA, which marks cells undergoing active neurogenesis (Fig. S1 A, D, and G). After 33 hpf, the region begins to express the differentiated neuronal marker vglut2.1 (Fig. S1 B, E, and H). Thus, as in the mammalian thalamus, we find a wave of proneural/neurogenic and neuronal gene expression spreading, with time, from posterior to anterior.
The neurogenetic gradient in fish. Glutamatergic neurogenesis spreads in a wave from posterior to anterior in the developing thalamus. Analysis of the dynamic expression of proneural genes during the development of the thalamic complex by in vivo imaging of double transgenic zebrafish and double in situ hybridisations. Upon induction of shh:GFP in the MDO, neurog1:RFP is induced first in the posterior Th (A, arrowheads). Over time, the caudal thalamus is filled with neurog1 positive cells (D and G). At the 20-somite stage, neurog1 mRNA can be detected within the thalamic complex (B). Over time, the expression increases from posterior to anterior and fills the entire cTh at 33 hpf (E and H). At 20 somite stage, her6 expression marks the entire thalamic complex (B and C) and is gradually down-regulated in the caudal part over time. her6 expression is maintained in the PTh, MDO, rTh and absent from the cTh at 33 hpf (H and I). The expression domains of her6 and neurog1 are abutting. In contrast to neurog1, the proneural gene ascl1a is induced from ventral to dorsal within the her6 positive PTh and rTh from the 20-somite stage (C), to 24 hpf (F) and 36 hpf (I). Embryos are shown laterally, white double-headed arrows indicate the increasing width of the cTh. III, third brain ventricle; cTh, caudal Thalamus; MDO, mid-diencephalic organizer; PTh, prethalamus; rTh, rostral Thalamus; ss, somite stage.
Our observations suggest the presence of a mechanism that regulates both temporal and spatial aspects of thalamic neurogenesis in vertebrates. Since members of the Hairy/Enhancer of Split (HES) family have been shown to be important regulators of neurogenesis in a number of contexts (17), we focused our attention on this family of bHLH transcription factors. We found that the hairy-related factor her6, a close relative of mammalian HES1, was also dynamically expressed in the developing diencephalon (25). Initially, her6 is broadly expressed in the presumptive telencephalon and mid-diencephalon at the open neural plate stage (Fig. S2 A–C). By the early somitogenesis stage, her6 becomes refined toward the mid-diencephalon and marks the entire thalamic complex including the anteriorly located PTh, the MDO, and the posteriorly located thalamus (Fig. 1B). Interestingly, her6 expression was found to be reciprocal to that of neurog1. At 33 hpf, her6 expression abuts the expanding expression domain of neurog1 precisely in the cTh. The second major neuronal population in the diencephalon consists of the GABAergic interneurons in the PTh and the rTh and their precursors, which are marked by the expression of achaete-scute complex genes. Therefore, we studied the expression of ascl1a relative to her6 in the developing Th. At the 20-somite stage, expression of ascl1a is induced within the her6+ PTh and at 24 hpf ascl1a is further found in the her6+ rTh (Fig. 1 C and F). At the 33 hpf, overlapping expression domains of her6 and ascl1a are maintained in the PTh and the rTh (Fig. 1I), whereas neurog1 marks the her6− cTh (Fig. 1H).
Knock-Down of Her6 Leads To an Increase of neurog1 Expression.
The expression of her6 is consistent with its regulating progression of the glutamatergic neurogenetic wave by upstream repression of neurog1. To test whether this is the case we used Morpholinos to create an antisense knock-down in vivo (26). In her6 morphant embryos, we found induction of neurog1 in the entire thalamic complex including the PTh, MDO and thalamus (Fig. 2 A and B). Notably, the timing of neurog1 induction at 24 hpf seems unchanged, suggesting that the inductive signal for neurog1 is unaffected. At 33 hpf in her6 morphant embryos, the PTh, the MDO and the entire thalamus becomes neurog1 positive and shh expression in the MDO is down-regulated (Fig. 2 C and D). These results indicate that Her6 regulates the progression of the neurogenetic gradient by acting as a brake to suppress premature or ectopic glutamatergic neurogenesis in the thalamus. In a second set of experiments, we analyzed the phenotype of the GABAergic cells in the thalamus of her6 morphants. We analyzed the expression of markers of the PTh such as the Ascl1 downstream factor dlx4 (27) by using the dlx4/6:GFP transgenic line (28). For the rTh marker we used the tal1:GFP line (29). tal1 (formerly known as scl) is essential for the development of the interneuron population originating from the rTh such as reticular nucleus, and the ventral lateral geniculate nucleus, including the intergeniculate leaflet (30). We find that both expression domains – dlx:GFP and tal1:GFP - are strongly reduced in the diencephalon (Fig. 2 E–H), whereas vasculature development such as of the tal1+ anterior cerebral vein (31) is unaltered (Fig. 2 G and H). As both domains are normally ascl1a+ (Fig. S1I), we asked whether this determinant of GABAergic fate is also affected. Indeed, we find that ascl1a is strongly down-regulated in the morphant embryos (Fig. 4 A and B), suggesting that the loss of dlx4/6:GFP and tal1:GFP expression is a consequence of the down-regulation of ascl1a. In support of this conclusion, we find that a marker for terminal GABAergic neurons, glutamic acid decarboxylase 65/67 (GAD 65/67), is strongly down-regulated in her6 morphant embryos (Fig. 4 E and F).
Her6 is required to repress the proneural gene neurogenin1. In vivo analysis of the loss of Her6 function in double transgenic zebrafish embryos by confocal microscopy. Knock-down of her6 (her6MO) leads to an increase of neurog1 expression in the thalamic complex at 27 hpf (62/86; A and B). At 33 hpf, lack of Her6 function leads to a massive up-regulation of neurog1 in the entire thalamic complex including PTh and MDO (72/91; C and D). The Shh positive MDO is fragmented. Up-regulation of neurog1 leads to a down-regulation of markers, such as dlx4/6:GFP in the PTh at 36 hpf (53/78; E and F). Similarly, the expression of the rTh marker tal1:GFP is completely lacking in her6 morphant embryos 42 hpf (23/28; G and H). By delivering the her6 MO antisense oligomers into a few cells of the rTh by electroporation, neurog1 expression was induced cell-autonomously, as shown by the expression of RFP tagged with a nuclear localisation sequence (5/11; J, higher magnification of the electroporated cell clone marked by arrowheads in J′). Electroporation of a control MO showed no effect (I and I′). III, third brain ventricle; ACeV, anterior cerebral vein; cTh, caudal thalamus; PTh, prethalamus; PTec, pretectum; rTh, rostral thalamus.
It is possible that a global knock-down of Her6 could have an early non-specific effect on thalamic patterning. To assess this we performed a temporally and spatially regulated knock-down experiment using an in vivo electroporation assay (32). Fluorescently tagged her6 Morpholinos were delivered into a small group of cells in the rTh of neurog1:RFP embryos at 24 hpf (Fig. 2 I–J′). Expression of neurog1:RFP was analyzed 12 h later. We found that neurog1:RFP was induced in a cell-autonomous manner in rostral thalamic cells that had taken up the antisense her6 MO (Fig. 2 I and J′).
Maintenance of Her6 Leads To the Acquisition of rTh Cell Fate.
In a further set of experiments, we investigated the Her6 maintenance phenotype in the cTh. Therefore, we used an inducible vector consisting of heat-shock responsive elements that drive bidirectional expression of her6 and gfp (33) (Fig. 3A–B′). Injection of the plasmid DNA into the neurog1:RFP line at the one-cell stage leads to mosaic distribution of the plasmid. Embryos were heat-shocked at the 10-somite stage, before the endogenous down-regulation of her6 expression in the thalamus. This experimental setting allowed us to analyze the phenotype resulting from maintained Her6 expression with high temporal precision. Although located in the normally neurog1+ cTh, cells that ectopically expressed her6 were unable to activate neurog1:RFP expression (Fig. 3 B and B′). In a control setting, we found that cells expressing a her6 construct that lacks the interaction domain for the Groucho co-repressor co-expressed neurog1:RFP (26) (Fig. 3 A and A′). To assess the requirement for Groucho in more detail, we performed a MO-mediated knock-down experiment for Groucho1. This led to the ectopic induction of neurog1:RFP as well as to down-regulation of shh:GFP at the MDO, a similar phenotype to that of her6 morphant embryos (Fig. 3 D and E). We then injected embryos with 50 pg her6 mRNA. This led to an down-regulation of neurog1 at 24 hpf followed by an up-regulation of ascl1a expression at 30 hpf (Fig. S3 C–F). We then analyzed the fate of small groups of clonally related cells maintaining Her6 in the cTh by injecting her6 mRNA in a single blastomere at the 32-cell stage. These cells express the rTh marker tal1:GFP ectopically in the cTh (Fig. 3 H and I). These experiments suggest that Her6 is able to repress neurog1 expression via a Groucho-dependent mechanism and that the down-regulation of Her6 is required for the formation of the cTh. If Her6 is not down-regulated, cells acquire the identity of the rTh.
Maintenance of Her6 is required of fate of the rTh. Mis-expression of her6 represses neurog1 expression and activates rostral thalamic fate. Activation of a heat-shock inducible construct driving her6 and the lineage tracer GFP leads to the cell-autonomous down-regulation of neurog1:RFP at 33 hpf (9/15; B and B′; arrowheads). If Her6 lacks the Groucho interaction motif WRPW, the construct does not alter neurog1:RFP expression in a control experiment and shows a co-localization of GFP and nuclear-localized RFP (8/12; A and A′; arrowheads). Knock-down of grg1 resembles the her6 morphant phenotype (23/43; D and E) and leads to an increase in the caudal expression of neurog1:RFP (white arrowhead) as well as a decrease of shh:GFP in the MDO (yellow arrowhead) at 30 hpf. Clonal overexpression of her6 and the membrane-bound red lineage marker mCherry in the cTh leads to the activation of the GABAergic rTh marker tal1:GFP (14/21; I; arrowhead). Notably, dividing cells at the ventricular surface are only positive for the mCherry lineage marker (asterisk) but become tal1:GFP positive after radial migration (arrowhead). Cells containing the lineage tracer only do not express Tal1 (H). III, third brain ventricle; cTh, caudal thalamus; PTec, pretectum; PTh, prethalamus; MDO, mid-diencephalic organizer; rTh, rostral thalamus.
Genetic Suppression of neurog1 by her6.
We then performed a knock-down experiment for her6 and neurog1. If Her6 acts solely via the repression of neurog1, it should be possible to rescue the her6 morphant phenotype by the double knock-down of both her6 and neurog1.
her6 morphant embryos display ectopic expression of neurog1 throughout the thalamus, down-regulation of ascl1 in the rTh and PTh, leading to an decrease in GABAergic neurons, and fragmented shh expression in the MDO (Fig. 4B, F, and J). In contrast, neurog1 morphants display posterior expansion of the ascl1a expression domain resulting in an increase in the domain of GABAergic neurons of the rTh (Fig. 4 C and G), while shh expression at the MDO appears normal (Fig. 4K). In her6/neurog1 double morphant embryos, both the ascl1a expression domains in the PTh and in the rTh and the corresponding domains of GABAergic neurons, as well as the shh expression in the MDO were rescued (Fig. 4 D, H, and L). Furthermore, the rTh showed an expansion of ascl1a expression comparable to the neurog1 morphant embryos, suggesting genetic suppression of neurog1 by her6/neurog1. To further test our hypothesis for this epistatic relationship, we performed an over-expression analysis of neurog1. Forced expression of neurog1 mRNA led to a down-regulation of ascl1a and gad1 expression in the fore- and midbrain area (Fig. S4 A–D), consistent with the observation in mouse (13, 14) and a cell autonomous down-regulation of shh in the MDO (Fig. S4 E and F) and of tal1 in the rTh (Fig. S4 G and H). These outcomes mimic the her6 morphant phenotype. Thus, the main function of Her6 is to repress neurog1 and Her6 is necessary but not sufficient to induce a subsequent GABAergic fate.
Her6 acts to genetically suppress neurog1. Analysis of ascl1a and irx1b, a pan-thalamic marker, at 30 hpf (A–D) and vglut2.1 and GAD65/67 at 48 hpf (E–H) in her6 and neurog1 morphant embryos. Knock-down of her6 leads to the down-regulation of ascl1a (45/61; A and B) as well as GAD65/67 (22/35; E and F) in the rTh (arrowheads) and in the PTh. Knock-down of neurog1 leads to an unaltered PTh but an increase in width in ascl1a expression (22/34; C) as well as GAD65/67 expression (12/21; G) in the rTh (yellow arrows). Compared to the her6 single knock-down, double knock-down of her6 and neurog1 rescues the prethalamic as well as the rostral thalamic expression domain (ascl1a: 17/22; GAD: 8/17; D and H). Analysis of the MDO in shh:GFP transgenic animals co-expressing ubiquitously the membrane-bound fluorophore mCherry (I–L): Down-regulation of her6 leads to disintegration of the Shh positive MDO (20/26; I and J; arrowheads). In embryos knocked-down for neurog1, the MDO seems unaltered in extension and width compared to the control (8/10; K, arrowhead). Double knock-down of her6 and neurog1 rescues the extend of the shh:GFP expression within the MDO (6/8; L, arrowhead). III, third ventricle; cTh, caudal thalamus; MDO, mid-diencephalic organizer; PTh, prethalamus; rTh, rostral thalamus.
Discussion
We have explored the function of the bHLH transcription factor Her6 during development of the zebrafish thalamic complex. We find that her6, the expression of which is initially widespread, represses the expression of neurog1 cell-autonomously throughout the thalamus. The subsequent posterior-to-anterior regression of her6 expression is accompanied by up-regulation of neurog1 in cells that formerly expressed her6. The progression of this gene expression transition presents as a neurogenetic gradient, a phenomenon that was observed, but not explained, in the mammalian thalamus. In addition, we find that the ascl1+ domains in the thalamic complex, the PTh and the rTh, require the maintenance of her6 expression (and her6-mediated repression of neurog1) for proper GABAergic specification. In zebrafish, two orthologues of the mammalian ascl1 gene exist, ascl1a and ascl1b, which show a similar expression pattern in the thalamic complex, suggesting a similar function (Fig. S2I). In parallel, her6 is required to maintain the MDO in a non-neurogenic state to fulfil its function as a signaling boundary. Lack of Her6 leads to ectopic de-repression of neurog1 in the PTh and the rTh, the repression of ascl1a, and the ectopic acquisition of glutamatergic fate. Additionally, the MDO starts to ectopically express neurog1, leading to the down-regulation of shh expression at the organizer. In contrast, cells forced to express her6 ectopically in the cTh adopt a GABAergic fate. Thus, we propose a triple role for Her6 during thalamic development: (i) Her6 determines the spatial activation pattern of neurog1, (ii) it is required for the development of GABAergic cell fate and (iii) it maintains the mid-diencephalic organizer as a non-neurogenic signaling boundary (Fig. S5).
Hes/her Genes and the Regulation of Neurogenesis.
In the developing CNS, Hes or Hes-related (Her) proteins have been shown to be involved in the regulation of neuronal differentiation. In Hes1 and Hes5 knockout mice, neural stem cells cannot be maintained and neurons differentiate prematurely (34, 35), whereas over-expression of Hes1 prevents neuronal differentiation in the brain (36). Furthermore, in the absence of Hes1 and its related genes, Hes3 and Hes5, proneural bHLH genes are ectopically expressed in boundaries, resulting in ectopic neurogenesis and disruption of their local organizing properties; for example, the MDO is disrupted in Hes1 mutant mice (21). These data suggest that Hes genes are critical for the proper spatial control of neuronal differentiation and for the maintenance of local organizing boundaries in the CNS. Following analysis of the Hes1/3/5 triple knockout mouse, a model was suggested in which the Hes genes repress all proneural bHLH factors, including Neurog, Ascl1 or Atonal (21). Similarly in the zebrafish thalamus, we show that neurog1 and coe2 (a member of the Collier/Olf1/EBF family of transcription factors and the ortholog of mouse Ebf2) are repressed by Her6, the orthologue of mouse Hes1, consistent with the presence of a bHLH binding motif (E-box) in the neurog1 promoter (20, 37). In contrast to mouse, however, we observe no repression of ascl1a by Her6. Rather, we find that the mRNAs of ascl1a and her6 are co-expressed, a phenomenon also seen in the figures of (21) but not discussed therein.
As well as disruption of the thalamic complex, the midbrain-hindbrain boundary (MHB) area including the ISO is missing in Hes1 mutant mouse embryos. From this perspective, Hes1 seems to be functionally equivalent to another bHLH factor, her5, in zebrafish (19, 20). her5 morphant embryos show premature up-regulation of the proneural genes neurog1 and coe2 at the ISO and, as with her6, her5 does not repress either ascl1a nor ascl1b (20). No orthologue of her5 has yet been identified in mammals, suggesting that Hes1 has adopted the functions of both her6 at the MDO/Th and her5 at the ISO in zebrafish.
The direct upstream signal regulating the activity of Her6 is unknown. Transcription of her6 seems to be dependent on Hdac1 (38) but independent of Notch signaling (39). However, recent data from the mouse retina suggest that Shh is able to stabilize Hes1 protein (40), suggesting one possible way in which the progressive retreat of Hes/her genes from the cTh could be controlled.
Proneural Genes in the Thalamic Complex.
Following the ectopic activation of her6 we saw first the down-regulation of neurog1, and subsequently the up-regulation of ascl1a. Thus, we suggest that Her6 is required to repress neurog1 expression and does not regulate ascl1a in the thalamus. This is consistent with the finding that Neurog1 is able to repress the transcription of ascl1, whereas Ascl1 is able to direct neuronal precursors toward a GABAergic fate (15). Interestingly, both genes are induced by the same signal, Shh (41–43) (Fig. S6). In the thalamus, concentration dependent Shh activity has been suggested to be involved in regionalization of thalamic nuclei: high Shh concentration defines rTh fate proximal to the source and lower concentration determines cTh fate distal to the source (44, 45). However, one single gradient cannot define a sharp border between abutting expression domains (46) such as the rTh and the cTh. Therefore, we propose that a different mechanism to assign correct neuronal fate - before the induction of proneural genes by Shh and the mutual repressive interaction of neurog1 and ascl1: Her6 perform this function by acting as a neurog1 repressor. When her6 function is abrogated, neurog1 expression is induced prematurely in normally glutamatergic domains of the thalamus, and is induced ectopically in normally GABAergic domains, bringing about a switch of fate. Spatial knock-down of Her6 by electroporation in an area exposed to high Shh concentration leads to the induction of Neurog1, supporting the hypothesis that Shh acts as an concentration-independent but general upstream inducer of proneural genes expression. Furthermore, rescue of the her6 knock-down phenotype by simultaneous knock-down of neurog1, suggesting that the down-regulation of ascl1 expression is a secondary effect of the her6 knock-down and represents the genetic suppression of ascl1 by neurog1. In this light, it would be interesting to examine the thalamic phenotype in Hes1−/− Neurog−/− mice. In support of our findings, it has been shown in a different context that other bHLH factors are required for GABAergic cell fate determination: Helt1 seems to have a function only in the mouse midbrain and acts in a different way as it lacks the Groucho binding domain (47), and Ptf1a is expressed only in the spinal cord (48).
The MDO is fragmented in the her6 morphant embryos, with glutamatergic proneural genes being expressed in place of shh. We suggest that the abrogation of Her6 function converts the organizer cells into neurons and, therefore, the characteristics of an organizing boundary (i.e., expression of the principal signaling molecule Shh) are lost, similar to the loss of Shh expression in Hes1 knock-out mouse (21). Thus, Her6 can be seen as essential for the maintenance of this local organizer.
The activity of Her6 appears to be Groucho-dependent. Indeed, it has been reported that Hairy related factors contain a Groucho interaction domain (49). It is unlikely that the down-regulation of shh expression at the MDO in the gro1 morphant embryos is due to Gro function directly, as it has been shown that Shh expression is unaltered when Groucho function is abrogated in the mouse spinal cord (50).
Temporal Control of neurog1 Induction.
Although we find ectopic neurog1 expression in her6 morphant embryos, the onset of neurog1 expression, at 24 hpf in the thalamus, is unaltered. The timing of induction of neurog1 expression as well as ascl1a strongly correlates with the induction of Shh in the MDO and, indeed, Shh has been demonstrated as a crucial factor for the regulation of proneural gene expression in a variety of neural tissues (41–43) and to be essential for diencephalic maturation (51–54). We show that blockade of Shh before its expression in the MDO is sufficient to block expression of the proneural genes neurog1 and ascl1a in the thalamus (Fig. S6). In support of these data we find that in embryos carrying a mutation in the nodal co-receptor one-eyed pinhead, in which shh expression is expressed in the MDO but is absent from the basal plate, neurog1 expression is unaltered within the thalamus (54), suggesting that Shh expression from the MDO alone is sufficient to induce neurog1 in the cTh. Compared to the phenotype in wild-type embryos, expression of her6 is independent of Shh signaling, as in smu mutant embryos her6 expression is still detectable. Reduction of the expression domain in smu mutant embryos may be explained by the overall reduced size of the diencephalon in the mutant (Fig. S6 F and I).
We therefore propose a model in which Shh from the MDO defines the timing of induction of proneural genes in the thalamus. Subsequently, Her6 determines the spatial dynamics of neurog1 induction and neuronal identity. Thus, Hairy genes are not only required for the spatial control of neurogenesis but are also important for the determination of a fundamental neuronal cell fate decision between excitatory projection neuron and inhibitory interneuron.
Methods
Maintenance of Fish.
Breeding zebrafish (Danio rerio) were maintained at 28 °C on a 14-h light/10-h dark cycle (55, 56). To prevent pigment formation, some embryos were raised in 0.2 mM 1-phenyl-2-thiourea (PTU, Sigma). The data we present in this study were obtained from analysis of King's College wild-type (KWT) fish, the neurog1:RFP transgenic line (22), and the shh:GFP transgenic line (23).
Injections.
Transient knock-down of gene expression was performed as described in (54), The following Morpholino-antisense oligomers (MO, Genetools LLC) were used at a concentration of 0.5 mM. her6MO (26), neurog1MO (57), gro1MO (5′- CGGCCCTGCGGATACATCTTGAATG-3′), and 5-bp mismatch control gro1MO (5′-CGcCCCTaCGGATAgATgTTcAATG-3′).
For mis-expression experiments mRNA was synthesized in vitro (Message Machine Kit, Amersham) from full-length pCS2+-her6 (26) and pCS2+-neurog1 (24). Together with rhodamine dextran as lineage tracer. 120pg mRNA was injected into one out of 64 cells (MiniRuby, Invitrogen).
For temporally controlled over-expression studies, 60 ng of the heat-shock (hs)-inducible pSGH2-her6 and her6-trunc DNA constructs were used per blastomere. At 10-somite stage, embryos were heat-shocked for 30 min at 42 °C.
Single-Cell Electroporation.
The single-cell electroporation technique (32) was adapted to efficiently transfer MOs to a small group of cells in the rostral thalamus of the neurog1:RFP line. Electroporation was performed on the right side of the thalamus at 22 hpf. Control embryos were electroporated with 0.5 μg/μL fluorescein-labeled control MO and experimental embryos with 0.5 μg/μL fluorescein-tagged her6 MO. To ensure stable concentration levels in the capillary, continuous but low outflow was used, causing green fluorescent staining of brain ventricular fluid. The following stimulation parameters were used as described for electroporation of Morpholinos in Xenopus embryos (58): 1-s long trains of 7-ms square pulses at 200 Hz and a voltage of 45 V. Trains were delivered three times with a 1-s interval between trains. Pulses were generated with a Grass SD9 stimulator (Grass-Telefactor).
Confocal Microscopy and Live Imaging.
For live imaging, larvae were embedded in 1.5% low-melting-point agarose dissolved in Ringer medium containing 0.016% tricaine at 22 hpf. Confocal image stacks were acquired using a Nikon C1 confocal laser-scanning microscope. To reconstruct the imaged area, we collected a series of optical planes (z-stacks) spanning one body-half of the larvae and projected them into a single image (maximum intensity projection) or rendered the volume in three dimensions to provide views of the image stack at different angles. The step size for each z-stack was chosen upon calculation of the theoretical z-resolution of the objective used (typically 2-μm) or for multiday imaging, typically 15-μm. The data sets were deconvolved by AutoDeblur X Gold-Edition (AutoQuant) and further processed using Imaris 4.1.3 (Bitplane AG).
Whole-Mount In Situ Hybridization.
Whole-mount mRNA in situ hybridizations (ISH) were performed as in ref. 59. Expression patterns have been described for ascl1a (originally described as zash1a) (60), coe2 (61), irx1b (originally described as Ziro1) (62), neurog1 (also known as ngn1) (41), shha (originally described as shh) (63), and wnt8b (64). For GAD staining post ISH, the polyclonal GAD65/67 antibody (Abcam) was used.
Acknowledgments
We thank Corinne Houart, Jon Clarke (MRC CDN) for experimental suggestions and comments, and Pia Aanstad (KIT) for help with the purmorphamine treatment. Patrick Blader (Centre de Biologie du Développement, Toulouse) supplied the neurog1:RFP fish line; Jon Clarke advised on the electroporation technique; Thomas Czerny (Institute of Animal Breeding and Genetics, Vienna) supplied the heat-shock inducible pSGH2 vector; Andrea Pasini (University of Padua) and David Wilkinson (NIMR, London) supplied the her6 probe; Steve Wilson (UCL) supplied the ascl1 probe. This work was funded primarily by a grant from the Medical Research Council (G0601064) and also by a Deutsche Forschungsgemeinschaft Emmy-Noether Fellowship (SCHO 847/2-1).
Footnotes
- 1To whom correspondence should be addressed. E-mail: steffen.scholpp{at}kit.edu
-
Author contributions: S. Scholpp, A.D., J.G., and A.L. designed research; S. Scholpp, D.P., and S. Schindler performed research; S. Scholpp contributed new reagents/analytic tools; S. Scholpp, and A.L. analyzed data; and S. Scholpp and A.L. wrote the paper.
-
The authors declare no conflict of interest.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0910894106/DCSupplemental.
-
Freely available online through the PNAS open access option.
