Foxd4 is essential for establishing neural cell fate and for neuronal differentiation

Abstract

Many molecular factors required for later stages of neuronal differentiation have been identified; however, much less is known about the early events that regulate the initial establishment of the neuroectoderm. We have used an in vitro embryonic stem cell differentiation model to investigate early events of neuronal differentiation, and to define the role of mouse Foxd4, an orthologue of a forkhead-family transcription factor central to Xenopus neural plate/neuroectodermal precursor development. We found that Foxd4 is a necessary regulator of the transition from pluripotent ES cell to neuroectodermal stem cell, and its expression is necessary for neuronal differentiation. Mouse Foxd4 expression is not only limited to the neural plate, but it is also expressed and apparently functions to regulate neurogenesis in the olfactory placode. These in vitro results suggest that mouse Foxd4 has a similar function to its Xenopus orthologue; this was confirmed by successfully substituting murine Foxd4 for its amphibian counterpart in overexpression experiments. Thus, Foxd4 appears to regulate the initial steps in establishing neuroectodermal precursors during initial development of the nervous system.

1 Introduction

Elucidating the precise steps that direct the transition of pluripotent stem cells to fully differentiated neurons is key for understanding normal and pathological neural development. Although there is an extensive literature that describes the molecules required for the transition of neural stem cells to neural progenitor cells and their subsequent differentiation into neurons (Gaspard and Vanderhaeghen, 2010; Perrier et al., 2004; Tropepe et al., 2001; Yan et al., 2013), much less is known about those that regulate the earlier transitions from a pluripotent stem cell to neural ectodermal cell to neural plate stem cell. This early phase of neural development in the embryo is critical for establishing a pool of proliferative neural stem cells that can produce a nervous system that is correctly scaled to the rest of the body. Defining the molecules that regulate the early steps of this program also is likely to provide important insights for manipulating pluripotent stem cells in vitro to produce specific neuronal phenotypes, and for determining how errors in the program result in neural malignancies.

A number of transcription factor genes that together regulate the size of the neural plate precursor population have been described in Xenopus (Sullivan et al., 2001; Yan et al., 2009). A key factor in the Xenopus network is Foxd4-like1 (foxd4l1), a member of the forkhead family of transcription factors that is expressed in the early neural ectoderm and then is down-regulated as the neural plate matures. In Xenopus, foxd4l1 is required for the expression of a large number of neural plate genes (Sullivan et al., 2001; Yan et al., 2009). It has the dual function of activating neural plate stem cell genes and delaying the onset of neural differentiation gene expression (Klein et al., 2013; Neilson et al., 2012), the net result of which is maintenance of the nascent neural ectoderm in a proliferative, immature state. Homologues of Xenopus foxd4l1 are highly conserved across vertebrates, and the zebrafish, mouse and human homologues of Xenopus foxd4l1, are all similarly expressed in the neural ectoderm of embryos (Kaestner et al., 1995; Katoh and Katoh, 2004; Odenthal and Nusslein-Volhard, 1998). Nonetheless, little is known about the function of the foxd4l1 homologues in other vertebrates, and particularly in mammals. Because of the important role that foxd4l1 plays in Xenopus neural development, and its conserved expression pattern in other vertebrates, we initiated studies to elucidate the role of its mouse orthologue, Foxd4, in regulating neural cell fate.

To accomplish this we utilized a well-described protocol to differentiate mouse embryonic stem cells (ESCs) into neurally committed stem cells and subsequently neurons (Chatzi et al., 2009). This paradigm allowed us to initiate neural fate progression in a precise manner, and thereby monitor gene expression levels as ESCs transitioned from pluripotency to neural ectodermal cells (NEs), to neural stem cells (NSCs), then to neural progenitor cells (NPCs) and finally to differentiating neurons. Each of these stages can be characterized by the expression of specific marker genes. For example, Nanog, Foxd3 and Oct4 are markers of ESCs (Loh et al., 2006; Silva et al., 2006), Nestin marks NSCs (Aubert et al., 2003; Elkabetz et al., 2008), and N-Cad and Zic1 mark committed NPCs (Aiba et al., 2006). Neurofilament (NF) and βIII-Tubulin are cytoskeletal proteins specifically expressed in post-mitotic neurons (Lee et al., 1990; Menard et al., 2002; Moody et al., 1989; Schulz et al., 2004). Using this paradigm we assayed when during the mouse neural developmental program Foxd4 is expressed, and then tested whether its protein is required for the acquisition of neural cell fate and neuronal differentiation.

We identified a direct relationship between the onset of Foxd4 expression and the transition between pluripotent ESCs and NSCs. Knock-down of Foxd4 in ESCs resulted in a maintenance of their pluripotent state without further differentiation to NPCs or neurons. Conversely, increasing the expression of Foxd4 in ESCs repressed their pluripotency, and promoted neuronal differentiation. We demonstrate that Foxd4 expression in the mouse embryo is required for neurogenesis in the olfactory epithelium. We further show that the mouse protein shares the same functional domains as the Xenopus protein, can ectopically induce neural plate stem cell genes in Xenopus embryos, and functions as both a transcriptional activator and repressor. This study is the first to identify a role for mammalian Foxd4 in neural development and demonstrates its key position in the transition from pluripotency to neural stem cells. It is necessary for both down-regulating ESC pluripotency genes and for up-regulating NSC and NPC genes that are required for acquiring a neural cell fate and differentiating into neurons.

2 Results

2.1 Foxd4 is expressed during neuronal differentiation of murine ES cells

In Xenopus, foxd4l1 is one of the earliest expressed neural ectodermal genes; it acts upstream of several neural plate stem cell genes, and it delays the expression of several genes required for neural differentiation (Sullivan et al., 2001; Yan et al., 2009). Mouse Foxd4 also is expressed in the early neural ectoderm (Kaestner et al., 1995), but the relative timing of its expression with respect to neural plate stem cell genes has not been described. To determine whether Foxd4 expression coincides with the acquisition of neural cell fate, we adapted an ESC differentiation protocol into embryoid bodies (Chatzi et al., 2009) that allows cells to be sampled during a defined neural differentiation process. In this procedure, pluripotent ESCs are first allowed to begin to differentiate by withdrawing support from LIF and feeder cell layers, and instead culturing on a non-adhesive substratum on a rotating platform, where they form multipotent “embryoid bodies” (EBs). After 2 days of culture, EBs are given a 2-day pulse of all-trans retinoic acid (RA) to promote neural differentiation, and then allowed to continue differentiation in the absence of RA for an additional 3 days, for a total of 7 days of in vitro culture. During this differentiation protocol, we assayed the expression of Foxd4, as well as for markers characteristic of specific phases of neural differentiation (Figure 1a–c). We assayed gene expression by both protein expression for markers for which specific antibodies are available, and by qPCR for detection of mRNAs. As summarized in Figure 1d, ESCs express pluripotency markers (Nanog, Foxd3, Oct4), but not NSC (Nestin), NPC (N-Cad; Zic1) or neuronal (βIII-Tubulin) markers. Pluripotency marker expression is initially maintained in EBs, but is reduced upon neural induction by RA treatment, when NSC markers begin to be expressed, followed by NPC and neuronal markers. Foxd4 expression commenced upon RA treatment, coinciding with the decline of pluripotency markers, but before the up-regulation of NSC, NPC and neuronal markers. This pattern of expression in the ESC/EB culture paradigm is consistent with the time course of its expression in both mouse and Xenopus embryos (Kaestner et al., 1995; Sullivan et al., 2001). Consistent with findings in the Xenopus embryo that foxd4l1 promotes the maintenance of a proliferative neural stem cell state (Sullivan et al., 2001), the percentage of proliferative cells (phospho-Histone 3-positive) is highest when Foxd4 is maximally expressed (Figure 1b–c). These assays demonstrate that in a mouse ESC/EB neural differentiation protocol, Foxd4 is expressed maximally during the transition between a pluripotent cell state and a neurally committed state (Figure 1d).

Figure 1.

Foxd4 is expressed in the period between pluripotency markers and neural stem cell markers in a mouse ESC neural differentiation protocol (modified from Chatzi et al., 2009). In this protocol, ESC are dissociated, removed from LIF and placed in rotation culture to form embryoid bodies (EBs). A pulse of RA exposure occurs on D3–4 of EB culture to promote neuronal differentiation. (a) EBs of each stage (as illustrated by brightfield microscopy, left column) were cryosectioned and assayed by immunofluorescence microscopy to reveal markers associated with pluripotency through neural differentiation, including Nanog, phospho-histone 3 (PH3), Nestin, and βIII-Tubulin. (b) Immunopositive cells were counted in EBs collected from each time point and expressed as a percentage of the total cell number. Sample sizes are reported in Supplemental Data Table 1 (c) EBs were collected at each time point and assayed for mRNA levels of Foxd4, Foxd3, Oct4, N-cadherin (Cdh2), and Zic1 by qPCR. Expression of each transcript is reported as a relative value normalized to whole E10.5 embryo expression. (d) Summary of culture methodology and expression of pluripotency/differentiation markers, including Foxd4.

2.2 Foxd4 is required for the transition from pluripotent to neural stem state

The time course of Foxd4 expression in the ESC/EB protocol suggests that it acts to limit pluripotency and to promote a neural cell fate. Since pluripotency genes (FoxD3, Sox2, Oct4, Nanog) are required for mouse ESCs to grow and form new colonies when subcloned (Liang et al., 2008; Xu et al., 2009), we tested whether Foxd4 is required for maintaining pluripotency. ESCs were transfected with plasmid vectors encoding two different shRNAs for Foxd4 (referred to as shRNA “A” and “B”, see Materials and Methods) or two different scrambled sequences as controls, plated, selected for transfection by hygromycin, and assayed for colony size and ability to form secondary colonies (Figure 2a). There was no significant difference in the size of the primary ESC colonies (Figure 2b) or their ability to generate secondary colonies (Figure 2c), indicating that Foxd4 expression is not required for the maintenance of pluripotent ESCs. Both knockdown constructs were shown to be effective in reducing the expression of a Foxd4-Luciferase construct when co-expressed in HEK 293T cells (26% and 20% of scrambled control levels, respectively; data not shown).

Figure 2.

Foxd4 is not necessary for maintaining pluripotency. (a) Schematic of ESC assay for assessing pluripotency. ESCs were electroporated with one of two different Foxd4 knock-down (KD) vectors (A, B) or their control constructs that co-express a GFP/Hygromycin resistance cassette. Transfected cells were plated under ES cell maintenance conditions (on a STO feeder layer with LIF) and assessed for the ability to produce viable ES cell colonies. (b) Initial formation of ES cell colonies from transfected cells was not significantly different between knockdown and control constructs (p >0.1 by one-way ANOVA; n=6 colonies/condition). (c) Knockdown and control-transfected ES cells were not different in their ability to generate secondary colonies, as assessed by dissociating primary ES cell colonies and plating 10,000 cells onto a new ES cell culture (p >0.1 by one-way ANOVA; n=6 samples/condition, quantified at two fields each).

To assess the role of Foxd4 in the differentiation of ESCs, we were able to utilize the fact that Foxd4 expression is not required for ESC maintenance in order to generate selected clonal lines of ESCs that were stably transfected with the Foxd4 KD (shRNA “A”) and scrambled control shRNA constructs. Both constructs generated proliferative ESC lines, and there was no apparent difference in the morphology of the scrambled (Figure 3a) and Foxd4-depleted colonies (Figure 3b). To assess whether Foxd4 is required for suppressing pluripotency and/or initiating neural fate, we first assayed gene expression in differentiating EBs derived from the Foxd4-depleted ESC colonies in D4 EBs (e.g. 2 days after RA treatment). As expected, the knock-down construct (“A”) reduced Foxd4 expression by 70% relative to control EBs (Figure 3c), as assessed by qPCR. This expression analysis suggested that knock-down of Foxd4 expression leaves the embryoid body cultures in a relatively undifferentiated state, as there is an increase in the expression of pluripotency genes (Foxd3, Oct4), along with a relative decrease of two markers of neural differentiation (Zic1, N-Cad/Cadh2). To further assess the state of differentiation, we assessed pluripotency markers at the cellular level by sectioning and staining Foxd4-depleted and scrambled-control EBs at an even later stage of differentiation, D9 (Figure 3d–l). For Nanog, we counted the percentage of cells that expressed high or low levels separately, as previous studies showed a correspondence of high levels with robust pluripotency and low levels with “unstable” stem cells that are beginning to differentiate (Kalmar et al., 2009). Scrambled-control EBs contained many low-Nanog expressing cells but very few high-Nanog expressing cells, whereas Foxd4-depleted EBs contained significantly fewer low-Nanog expressing cells and more high-Nanog expressing cells (Interaction p<0.0001 by 2-way ANOVA; Figure 3d–f). Thus, reducing Foxd4 decreases the frequency of low-Nanog-expressing cells associated with differentiation, and increases the frequency of high-Nanog expressing cells associated with pluripotency. Foxd4-depletion likewise increased the number of cells that co-expressed other markers of pluripotency: Sox2, Nanog, and Oct4 (Figure 3g–l). As single-labeling for individual markers might leave doubt as to pluripotency status, we focused on the number of cells expressing two markers simultaneously. Cells double labeled for Sox2/Nanog and Sox2/Oct4 were virtually absent in controls, but readily apparent in Foxd4 depleted EBs (p<0.02 and p<0.0004 by t-test, respectively). Thus, depleting Foxd4 expression permits the continued expression of pluripotency markers in differentiating EB cultures, suggesting that Foxd4 depleted cells remain undifferentiated.

Figure 3.

Foxd4 is required for differentiation of ES cells. (a–b) Stably-selected control and Foxd4-depleted ESCs proliferate and generate morphologically-normal ESC colonies. (c) EB cultures generated from Foxd4-KD ESCs show decreased expression of Foxd4 (p<0.001 by two-way ANOVA with Sidak’s multiple comparison analysis), increased expression of Foxd3 (p<0.0001) and Oct4 (p<0.05), and decreased expression of Zic1 (p<0.001) and N-Cad/Cadh2 (p<0.001). (d–l) Immunofluorescence analysis of cellular expression of pluripotency markers in D9 EBs. (d–f) Quantification of high- and low-levels of Nanog expression. (g–i) Quantification of cells expressing both Sox2 and high levels of Nanog. (j–l) Quantification of cells expressing both Sox2 and Oct4. (m–n) Knockdown of Foxd4 prevents neural differentiation, as assessed by an adherent cell culture protocol. Knockdown and control ESCs were cultured in neural culture media for 15 days; the knockdown cells failed to differentiate into neurons, as assessed by neurofilament or βIII Tubulin staining (not shown) staining, as compared to abundant neuronal differentiation observed in cultures of control cells (n=4 cultures for each condition and label). Scale bars = 100 μm.

Finally, to determine whether Foxd4 is required for neuronal differentiation, stably-transfected knock-down and scrambled-control ES cells were cultured using an alternative differentiation protocol that promotes neurogenesis (Ying et al., 2003) by plating approximately 7,500 control or Foxd4-KD ESCs directly on an adhesive substrate in the presence of neural-promoting growth media. After 15 days of culture, the scrambled-control ESCs generated large numbers of NF- or βIII Tubulin-positive neurons characterized by elaborate arbors of processes (Figure 3m), as expected. In contrast, neither NF- or βIII Tubulin-positive neurons were visible in the Foxd4-depleted cultures (Figure 3n). Thus, it appears that in the absence of Foxd4 expression, ESCs are unable to generate neurons even under favorable conditions.

2.3 Foxd4 expression promotes ESC differentiation

We next tested whether increasing levels of Foxd4 can perturb ESC pluripotency. ESCs were electroporated with a Foxd4-IRES-hygromycin expression vector and plated under standard ESC growth conditions including hygromycin selection (Figure 4a). As compared to empty vector-control ESCs, which produce numerous morphologically normal colonies, Foxd4-positive ESCs did not generate significant numbers of viable colonies – only 3 colonies were detected for cultures transfected with Foxd4 overexpression plasmid across all experiments (less than 1 per transfection), compared to 208 (average of 34 per transfection) for cultures transfected with vector-only control plasmid (p=0.025 by t-test; Figure 4b). Thus, Foxd4 overexpression is not compatible with viability of pluripotent ESC populations.

Figure 4.

Foxd4 expression promotes neuronal differentiation of ES cells. (a) Schematic of ESC assay for assessing pluripotency. ESCs were electroporated with either Foxd4-IRES-Hygromycin expression vector or a control Hygromycin-only construct. Transfected cells were plated onto ESC maintenance conditions (on a STO feeder layer with LIF) under Hygromycin selection and assessed for the ability to produce viable ESC colonies. (b) Foxd4-overexpressing ESCs do not make significant numbers of viable ESC colonies, unlike control ESCs (p=0.025 by t-test for yield from 6 cultures each condition). (c–h) Chimeric ESC colonies were generated by transfecting Foxd4-IRES-EGFP and control constructs, and culturing in ESC maintainance conditions without selection. Foxd4-EGFP expressing cells were observed infrequently in ESC colonies; these GFP-positive cells did not express Nanog (c–e), but some GFP-positive cells at the edges of colonies did express Nestin (f–h). (i–k) ESCs were transfected with a Foxd4-IRES-EGFP construct and allowed to differentiate in neuronal media for 14 days. Foxd4-IRES-EGFP positive cells differentiated into neurons, similar to untransfected cells in the same cultures.

To address whether the loss of viable colonies was due to acute lethality caused by Foxd4 overexpression, we investigated the fate of Foxd4-overexpressing cells by transiently transfecting ESCs with a Foxd4-IRES-EGFP plasmid to generate chimeric ESC colonies. As predicted by the inability of Foxd4-ires-hygromycin transfected cells to generate viable colonies, the GFP-positive cells were outliers in the colonies: across eight transfections, containing several hundred colonies, we only identified five colonies that contained GFP-positive cells. We investigated this small cohort of chimeric colonies by immunostaining with markers characteristic of ESCs (Nanog) and neuroectodermal stem cells (Nestin). Consistent with the loss of colony formation, the GFP-positive cells did not express Nanog whereas the non-transfected, GFP-negative cells did (n=2 chimeric colonies, example in Figure 4c–e), however, some GFP-positive cells were found within colonies, or in the “halo” of cells at the base of colonies, that co-expressed Nestin (n=3 chimeric colonies, example in Figure 4f–h). When transfected ESCs were instead plated under neural differentiation inducing conditions (as in Figure 3m–n), numerous IRES-EGFP cells were observed, forming GFP-positive clusters of βIII-Tubulin-expressing neurons, scattered among similar clusters of non-GFP-expressing, βIII-Tubulin-positive untransfected cells (observed in 4/4 cultures, example in Figure 4i–k).Together, these results indicate that that Foxd4 expression is not acutely lethal, but instead Foxd4 likely acts by suppressing pluripotency and/or promoting neuronal differentiation.

2.4 Mouse Foxd4 may regulate neuroectodermal development at sites of secondary induction

The earliest described embryonic expression of Foxd4 in the mouse is in the neural plate/neural ectoderm at E7.5–8.0 (Kaestner et al., 1995). To determine whether Foxd4 also is expressed at other sites undergoing neural induction, we examined its expression in the E9 mouse embryo by in situ hybridization. Similar to reports from Xenopus embryos (Fetka et al., 2000; Solter et al., 1999; Sullivan et al., 2001), we observed robust expression in the most caudal aspect of the tail, which is a site of secondary neural induction (Figure 5a). We also observed expression in the nascent olfactory placode region (Figure 5a). To confirm that this labeling was in the olfactory sensory ectoderm (OE), and not in the underlying mesenchyme (which includes a significant population of neural crest-derived cells; LaMantia et al., 2000), we assessed its expression by qPCR in microdissected samples of OE, mesenchyme, and the underlying forebrain neural ectoderm. The highest levels of Foxd4 were observed in both isolated forebrain and OE, whose identities were confirmed by assessing Pax6 levels (Figure 5b) which is expressed in both compartments (Grindley et al., 1995).

Figure 5.

Foxd4 is expressed in the nascent olfactory placode, and modulates olfactory neuron differentiation. (a) In situ hybridization of E9.5 mouse embryo shows significant expression of Foxd4 in the olfactory ectoderm (OE) and in the tail bud. (b) qPCR quantification of Foxd4 and Pax6 in microdissected and separated OE, frontonasal mesenchyme (Mes) and frontonasal neuroepithelium (FNE). (c) Summary of approach for electroporation and culture of frontonasal neuroepithelium, as described previously (Tucker et al., 2010). (d) Explant cultures transfected with Foxd4 overexpression vector (OvExp), shRNA #A (KD), or control (empty) expression vector, immunostained with βIII-Tubulin to reveal olfactory neurons.

To determine whether mouse Foxd4 is required for neuronal differentiation in the OE, we used an explant culture approach we previously employed (Tucker et al., 2010) to electroporate tdTomato-tagged versions of Foxd4 shRNA knock-down and over-expression vectors into the surface ectoderm of E9 embryos. The head was then hemisected and cultured for 48 hours to allow the OE to differentiate (Figure 5c). tdTomato-positive explants were then assessed by immunostaining for βIII-Tubulin. Cultures transfected with shRNA to deplete OE cells of endogenous Foxd4 contained virtually no βIII-Tubulin-positive cells (6/6 cultures). In contrast, OE cultures that were transfected with Foxd4-expressing vectors showed clusters of βIII-Tubulin-positive cells (3/3 cultures) that were distinctly larger than those observed in empty-vector controls (6/6 cultures; Figure 5d). When coupled with expression data, these experiments suggest that Foxd4 has an additional role in establishing neuroectodermal cell fate in other embryonic sites, including the OE.

2.5 Mouse Foxd4 has homologous activity to Xenopus foxd4l1

Xenopus foxd4l1 is sufficient to induce the ectopic expression of neural plate stem cell genes in epidermal precursor cells (Yan et al., 2009). To determine whether mouse Foxd4 has the same in vivo activity, we microinjected mouse Foxd4 mRNA (plus βgal mRNA as a lineage tracer) into a defined precursor of the epidermis (blastomere V1.1; Moody, 1987) on one side of a 16-cell Xenopus embryo. Pink nuclear βGal histochemical staining identified the clone of cells derived from the injected blastomere, and neural gene expression was assessed by in situ hybridization (Figure 6a). Both mouse and frog Foxd4 homologues ectopically induced at high frequencies three genes (gmnn, sox11, zic2) that promote proliferative neural precursors upstream of other Sox B1-family genes (Yan et al., 2009); βgal mRNA alone did not cause this induction (Figure 6a–c). Previous work showed that this ectopic induction requires both transcriptional activation and repression, and identified two domains in Xenopus Foxd4l1 that are required for these activities (Klein et al., 2013; Neilson et al., 2012). Analysis of the mouse protein demonstrates that it contains these same two domains: an acidic blob (AB) for activation, and an Engrailed Homology-1 domain (Eh-1) for Groucho mediated repression (Figure 6d). To determine if these two domains are functional in mouse Foxd4, we expressed either Xenopus or mouse mRNAs in a defined 16-cell precursor of the neural plate (blastomere D1.1; Moody, 1987). This results in the βGal-labeled clone occupying approximately 50% of the neural plate on only the experimental side of the embryo (Figure 6e). The expression of two neural genes that are activated by Foxd4l1 (gmnn, zic2) and two neural genes that are repressed by Foxd4l1 (sox11; zic1) were assessed by in situ hybridization (Figure 6e). In embryos injected with βgal mRNA alone, the intensity of the ISH reaction product (blue) for each neural gene was the same within the clone (pink nuclei) and outside the clone (e=endogenous expression). However, for both Xenopus Foxd4l1 and mouse Foxd4, gmnn and zic2 expression was increased (bluer) within the clone compared to endogenous expression outside the clone (Figure 6f, g). Similarly, sox11 and zic1 expression was decreased (less blue) within the clone compared to endogenous expression outside the clone (Figure 6f, g). Since previous work showed that the AB is required for up-regulation of gmnn and zic2, and the Eh-1 domain is required for down-regulation of sox11 and zic1 in the neural ectoderm (Klein et al., 2013; Neilson et al., 2012), these results demonstrate that in an intact embryo environment, mouse Foxd4 can up-regulate genes that maintain a proliferative, immature neural ectoderm and down-regulate genes that promote the transition to differentiating NPCs.

Figure 6.

Mouse Foxd4 has a similar ability as frog foxd4l1 to regulate expression of neuroectodermal genes. (a) Xenopus 16-cell epidermal progenitors (blastomere V1.1) were microinjected with indicated mRNAs and embryos examined at stages 10.5–11.5 for the presence of the clonal progeny of the injected cell (pink nuclear labeling) and for mRNA expression by in situ hybridization (diffuse blue labeling). As an example, gmnn was ectopically induced in the ventral epidermis by mouse Foxd4 but not by the lineage tracer (nβgal). (b) High magnification image of injected embryos. Ectopic expression of either frog foxd4l1 or mouse Foxd4 leads to induction of gmnn, zic2 and sox11 expression (blue label) in the ventral epidermis within the injected cell clone (pink nuclei). (c) The percentage of embryos that show ectopic induction of each neural gene in the ventral epidermis after injection of frog foxd4l1 (green bars) or mouse Foxd4 (blue bars; n for each experiment noted on graph). (d) There is a high degree of homology between frog foxd4l1 and mouse Foxd4, particularly within the N-terminal “acidic blob” (AB) domain and the C-terminal Eh1 domain, which have previously been identified as key activator and repressor domains in frog foxd4l1. (e) Xenopus 16-cell neural ectoderm progenitors (blastomere D1.1) were microinjected with indicated mRNAs and examined at stages 11.5–12.5 as above. As examples, gmnn expression in the neural ectoderm was increased by mouse Foxd4 but not by the lineage tracer (nβgal), and sox11 expression was repressed (arrows). (f) High magnification image of injected embryos, illustrating modulation of expression in injected clones. Injection of either frog foxd4l1 or mouse Foxd4 lead to increased expression of gmnn and zic2 within the injected clone (asterisk), relative to the level in neighboring uninjected cells that are also in the neural ectoderm (e). Similarly, injection of either frog foxd4l1 or mouse Foxd4 lead to decreased expression of sox11 and zic1 relative to adjacent uninjected neural ectoderm (e). (g) The percentage of embryos that show either increased expression (left) or decreased expression (right) of each neural gene in the neural ectoderm after injection of frog foxd4l1 (green bars) or mouse Foxd4 (blue bars). The number of replicates for each experiment is noted within each bar on the graphs (c, g).

3 Discussion

The mechanisms by which embryonic cells acquire a neural fate have been of central interest for over a century (Moody et al., 2013). Although there has been tremendous progress in elucidating the signaling pathways and transcription factors that regulate aspects of mammalian neurogenesis (Florio and Huttner, 2014; Imayoshi and Kageyama, 2014), such as the network of transcription factors that regulate later stages of neural cell specification, much less is known about the regulation of the earliest stages of neurogenesis. A key step in neural development that is difficult to assess in mammalian embryos is the transition of the nascent neural ectoderm from pluripotency to the neurally-committed stem cells of the neural plate. Gene expression, knock-down and ectopic expression studies in the more accessible Xenopus embryo have provided considerable evidence that the forkhead transcription factor, Foxd4l1 plays an important role in this transition (Moody et al., 2013). Since there is no similar evidence available in other vertebrates, we determined whether mouse Foxd4 is required for neural cell fate and neuronal differentiation by taking advantage of ESC/EB culture paradigms as well as the intact mouse embryo.

3.1 Foxd4 represses pluripotency and promotes neural cell fate

The time course of Foxd4 expression in both embryos and ESC/EB culture suggests that it plays an important role is restricting pluripotent stem cell fate to neural fate, and thus would be expected to down-regulate pluripotency genes. We provide three lines of evidence that this is the case. First, Foxd4 knock-down does not interfere with ESC colony size or propagation, indicating that it acts downstream of pluripotency genes. Second, pluripotency gene expression is enhanced when Foxd4 is depleted. Third, increasing the level of Foxd4 down-regulates pluripotency genes. Thus, an important role of Foxd4 is to direct pluripotent cells to a more restricted fate. The observation that in the ESC/EB differentiation time course multiple NSC, NPC and neuronal markers are detected after the onset of Foxd4 expression indicates that the restricted state promoted by Foxd4 is neural. These markers are lost when Foxd4 is depleted, and they are increased when Foxd4 levels are experimentally increased. It is interesting that in both the ESC/EB system and in the mouse OE the number of differentiated neurons were increased by Foxd4 over-expression. At first glance, this would seem to contradict the conclusions draw from Xenopus, in which Foxd4l1 represses neural differentiation genes (Yan et al., 2009). However, this repression appears to be transient, resulting in a prolonged period during which neural stem cell genes are expressed allowing the expansion of a neurally-committed NPC population. When Foxd4 levels are subsequently reduced, there is then a larger pool of neural stem cells available to undergo differentiation (Sullivan et al., 2001).

3.2 Foxd4 is required for the transition from pluripotency to neural stem cell state

Evidence from Xenopus and mouse embryos, as well as mouse ESC/EB cultures concur that Foxd4 is expressed at the transition from a pluripotent cell to a neurally committed stem cell. In frog, foxd4l1 is expressed in the blastula cells that will give rise to the neural ectoderm, in the nascent neural ectoderm, and then is down-regulated as the neural plate forms and primary neurons begin to differentiate (Klein and Moody, 2015; Sullivan et al., 2001). In mouse, foxd4 also is transiently expressed throughout the early neural ectoderm (Kaestner et al., 1995), and subsequently in the anterior forebrain, caudal-most neural tube and olfactory ectoderm. Monitoring Foxd4 expression in an ESC/EB neural differentiation culture system is consistent with the in vivo time course: Foxd4 is expressed after neural induction by RA, peaks in advance of NSC markers and declines as NPC and neuronal markers are expressed. This time course suggests that Foxd4 is transiently expressed in response to neural inductive signaling to promote an immature, proliferative neural state. Consistent with this idea, we found that there is more proliferation during the peak period of Foxd4 expression.

Foxd4 likely acts at this transition in cooperation with other nuclear factors. For example, Gmnn, which also promotes a neural fate in Xenopus (Kroll et al., 1998; Seo et al., 2005), is required for mouse ESCs to undergo neural fate commitment (Yellajoshyula et al., 2011). Like Foxd4, when Gmnn is depleted ESCs still proliferate and form colonies, and over-expression upregulates NSC and NPC genes (Yellajoshyula et al., 2011). Loss of Gmnn in mouse embryos can reduce the numbers of NPCs and either reduce or increase differentiated neurons depending on the marker assayed (Emmett and O’Shea, 2012; Patterson et al., 2014; Schultz et al., 2011; Spella et al., 2011). These differential effects are likely due to Gmnn’s ability to affect the epigenetic state of NSCs (Caronna et al., 2013; Seo et al., 2007; Yellajoshyula et al., 2012). As both frog and mouse Foxd4 affect gmnn expression in Xenopus neural ectoderm, it will be important to investigate their specific molecular relationships during neural fate acquisition.

In mouse, a “primitive neural stem cell” population has been identified that is intermediate between ESCs and multipotent NSCs (Li et al., 2015; Tropepe et al., 2001; Zhang et al., 2010). Like the classical work on neural induction in Xenopus (reviewed in Moody et al., 2013; Ozair et al., 2013), these cells will express neural markers when BMP signaling is inhibited, and require FGF signaling to attain a multipotent NSC state. From the work presented herein, we predict that these cells are equivalent to the Foxd4-positive cells in our ESC/EB protocol. Other factors likely to be involved in this transition include Ars2 (Andreu-Agullo et al., 2012), Zfhx1b (Dang and Tropepe, 2010; Dang et al., 2012) and Otx2 (Rhinn et al., 1998; Tian et al., 2002). Otx2 is expressed in the mouse Organizer in response to Foxa2 (Tamplin et al., 2008), and loss of Otx2 expression results in Foxd4 down-regulation (Rhinn et al., 1998; Tamplin et al., 2008). Therefore, it will be very important to investigate the interacting relationships between these transcription factors and Foxd4 during the transition from ESCs and multipotent NSCs in vitro and in the embryo.

Our analysis of Foxd4 provides, for the first time, characterization of a transcriptional regulator essential for the early transition between undifferentiated ectodermal stem cells to neurogenic-biased stem cells. This transition in the embryo is key for establishing neural plate and neurogenic placode domains that will acquire the capacity to generate committed neural precursors in response to downstream transcriptional regulators. Thus, we have placed Foxd4 in a unique and heretofore uncharacterized position in the gene regulatory network in mammals that regulates neural stem cell progression from undifferentiated ectoderm. The conservation of Foxd4 function in mouse and in Xenopus indicates that the regulation of this critical step is a highly specified molecular event shared across distinct vertebrate species. Our identification of Foxd4 as a regulator of acquisition of neurogenic competence in pluripotent ectodermal stem cells provides an important tool in modulating the transition between ectoderm and neural stem cells both in vitro and in vivo.

4 Materials and Methods

4.1 ES cell cultures

Mouse ES cells (E14Tg2a) and feeder cells (STO) were obtained from ATCC. ES cell media consisted of Dulbecco’s Minimum Essential Medium (DMEM, Invitrogen), supplemented with 15% Fetal Calf Serum (FCS, HyClone), 0.1 mM 2-mercaptoethanol, 1x antibiotic (Penicillin/Streptomycin/Amphotercin-B, Invitrogen), and 100 U/ml LIF (Enzo). Feeder cell media consisted of DMEM supplemented with 10% FCS and 1x antibiotic. Feeder cell layers were obtained by mitotically inactivating confluent plates of STO cells by a 2 h treatment with 10 μg/ml mitomycin C, then passaging the cells at a 1:2 ratio onto gelatin-coated tissue culture plastic dishes. Stock ES cells were routinely maintained by trypsinizing and passaging confluent plates of ES cells at a 1:4 subculture ratio every 3–4 days. To generate embryoid bodies, 5 x 10⁶ ES cells were plated in EB medium in a 100 mm petri dish and cultured in a shaking culture incubator at approximately 50 RPM to facilitate formation and differentiation of embryoid bodies (Chatzi et al., 2009). EB media consists of DMEM with 10% FCS, 1x MEM Non-essential amino acids (Gibco/Life Technologies) and 1x antibiotic. EBs were treated with retinoic acid (10 μM) on day 3 and 4 to promote neural differentiation; with media changed on day 5 to terminate RA treatment. As a second method to generate neuronal cultures, we adapted previous protocols (Ying et al., 2003) by dissociating, counting, and plating undifferentiated ES cells at approximately 7,500 cells per well on an 8-well gelatin-coated chamber slide in N2B27 medium. N2B27 medium is a 1:1 mixture of DMEM/F12 (Gibco/Life Technologies) supplemented with N2 (Gibco) and Neurobasal medium supplemented with B27 (Gibco). Culture media was renewed every 2 days for all experiments.

4.2 Plasmid vector creation and transfection

The coding sequence of mouse Foxd4 was cloned by PCR from an EST plasmid (IMAGE # 8861454, Open Biosystems), using oligonucleotide linkers to add restriction sites and a translation consensus sequence (5′-CCACC-3′) to the 5′ end, and an HA-epitope tag (DPYPYDVPDYA) to the 3′ end. This coding frame was subcloned into several variants of a pCAGG-derived expression vector, containing the composite CMV-Chicken β-Actin promoter to express the coding frame and an EMCV-IRES reporter at the 3′ end to co-express either EGFP, tdTomato, or Hygromycin, as described. To knock-down endogenous Foxd4 mRNA, we generated a shRNA-expression vector by modifying the above expression vectors to include a H1-RNA promoter cassette (derived from pSilencer 3.1, Life Technologies), and using the pCAGG promoter to drive expression of either a GFP or Hygromycin selection cassette. shRNA sequences targeted two unique 19mer sites (“A”: 5′-GAGATCAGACGGAAGAAGA-3′ and “B”: 5′-GGACCTTGCCCGAATTTTA-3′) within the coding sequence. To transfect ES cells with plasmid, cells were dissociated with trypsin, washed and resuspended in OptiMem (Gibco/Life Technologies), and 1 x 10⁷ cells were placed in a 4 mm electroporation cuvette with 5 μg of plasmid DNA, electroporated with 2 pulses (1 ms, 500 V) in a square pulse electroporator (BTX-830, Harvard Apparatus), and plated as described.

4.3 Marker analysis

Analysis by quantitative PCR and immunofluorescence was performed as described previously (Paronett et al., 2015). For qPCR analysis, tissue samples were homogenized in TRIzol (Invitrogen, Carlsbad, CA), total RNA was isolated, and residual genomic DNA was removed using DNAse (Turbo DNA-free, Invitrogen). cDNA synthesis was performed using random-hexamer priming and ImPromp-II reverse transcriptase (Promega, Madison WI). Reactions were assembled using a EpMotion 5070 liquid handling system (Eppendorf, Hauppauge, NY) that combine forward and reverse gene-specific primers (0.3 μM final concentration, Integrated DNA Technologies, Coralville, IA), with 7.5 μl of SsoFast EvaGreen Supermix (Bio-Rad, Hercules, CA) in a 14 μl reaction. Oligonucleotide sequences for qPCR are included as Supplemental Table 3.

Immunofluorescent analysis was performed on cultured tissue by fixing tissue in 4% paraformaldehyde in PBS for 1h, washing thoroughly in PBS, then incubating in a blocking solution (3% BSA, 10% normal goat serum, 0.3% Triton X-100 and 0.1% sodium azide in PBS) for 1 hour. Primary antibodies were diluted in the same blocking solution and incubated overnight, washed for 1 hour in PBS, then incubated in species appropriate Alexa-fluor series secondary antibodies (Invitrogen, diluted to 1:2,000 in block solution) for 1 hour. After washing, specimens were counterstained with the nuclear dye bis-benzamide (Sigma), and coverslipped in a MOWIOL solution with paraphenylene diamine (PPDA) to prevent photobleaching. Antibodies used in this study along with their respective dilutions are included as Supplemental Table 4.

In situ hybridization on whole E9.5 embryos was performed using a 532-bp fragment of Foxd4 (Accession number NM_008022.2, base 1412-1943), chosen for low homology to other forkhead family members. This fragment was subcloned into a modified Bluescript vector and digoxigenin-labeled RNA (Roche) was synthesized for antisense and sense control probes using T7 and T3 RNA polymerase (Promega), respectively. In situ hybridization on whole E9.5 and 10.5 embryos was performed as described previously (Maynard et al., 2002). Sense control embryos did not show detectable signal. Hybridized embryos were cleared in glycerol for photography.

4.4 Olfactory epithelia electroporation assays

Explant electroporation and culture was carried out using established protocols (Tucker et al., 2010). Briefly, E9.5 CD-1 mouse embryos (Charles River Laboratories) were dissected, and head regions were bisected and placed in an electroporation chamber (CUY520-P5, Protech International) with plasmid vector. After a brief pulse (30V, 50 ms pulse, 950 ms delay x 5 pulses; BTX-830, BTX/Harvard Apparatus), embryos were placed on membrane filters floating on standard (feeder) culture media for 48 hours to allow neuronal differentiation. Explants were stained with β–III Tubulin antibody to reveal neuronal differentiation. Animal housing and experimental procedures met GWU Institutional Animal Care and Use Committee requirements in accordance with NIH guidelines.

4.5 Xenopus assays

Mouse Foxd4 (mFoxd4) and Xenopus foxd4l1 cDNAs were constructed as previously described (Gaur et al., 2016; Sullivan et al., 2001). Fertilized Xenopus laevis eggs were obtained by gonadotropin-induced natural mating of adult frogs (Moody, 2000). The eggs were dejellied with 2% cysteine solution and selected at the 2-cell stage if the first cleavage furrow bisected the lightly pigmented region of the animal hemisphere to accurately identify the dorsal-ventral axis (Klein, 1987; Miyata et al., 1987). These selected embryos were cultured in 100% Steinberg’s solution until the 16-cell stage, when each blastomere was microinjected with 1nl of mRNA according to standard methods (Moody, 2000). mRNAs encoding Xenopus foxd4l1.1a (100pg/nl; Sullivan et al., 2001), mFoxd4 (75pg/nl), and a nuclear-localized β-galactosidase (nβgal; 100pg/nl) were synthesized in vitro (Ambion, mMessage mMachine kits). Each transcription factor mRNA was mixed with nβgal mRNA and microinjected into either a dorsal-animal blastomere (D11), which is the major precursor of neural ectoderm or a ventral-animal blastomere (V11), which is the major precursor of the epidermis (Moody, 1987). When embryos reached early neural ectoderm stages they were fixed in 4% paraformaldehyde in MEM, stained for nβGal histochemistry, processed for ISH and bleached according to standard procedures (Yan et al., 2009). Anti-sense RNA probes for early neural ectoderm genes (gmnn, zic2, zic1, sox2, sox11,) were synthesized in vitro (Ambion MEGAscript kit) as previously described (Yan et al., 2009).