Multiple conserved regulatory domains promote Fezf2 expression in the developing cerebral cortex

Background The genetic programs required for development of the cerebral cortex are under intense investigation. However, non-coding DNA elements that control the expression of developmentally important genes remain poorly defined. Here we investigate the regulation of Fezf2, a transcription factor that is necessary for the generation of deep-layer cortical projection neurons. Results Using a combination of chromatin immunoprecipitation followed by high throughput sequencing (ChIP-seq) we mapped the binding of four deep-layer-enriched transcription factors previously shown to be important for cortical development. Building upon this we characterized the activity of three regulatory regions around the Fezf2 locus at multiple stages throughout corticogenesis. We identified a promoter that was sufficient for expression in the cerebral cortex, and enhancers that drove reporter gene expression in distinct forebrain domains, including progenitor cells and cortical projection neurons. Conclusions These results provide insight into the regulatory logic controlling Fezf2 expression and further the understanding of how multiple non-coding regulatory domains can collaborate to control gene expression in vivo.


Multiple conserved regulatory domains promote
Fezf2 expression in the developing cerebral cortex Eckler et al.

Background
Current estimates suggest that greater than half of evolutionarily conserved sequences in the human genome do not correspond to protein coding regions [1][2][3][4]. Transcriptional enhancers, non-coding DNA sequences that promote gene expression in a spatial and temporal manner, constitute a substantial portion of these regions [5]. Genome-wide association studies (GWAS) have identified sequence variations within putative enhancers that are associated with a wide range of neurological disorders such as autism, epilepsy, and schizophrenia [6][7][8][9][10]. However, the function of these DNA elements during normal brain development remains poorly understood [11].
The cerebral cortex contains six layers of projection neurons that are generated in a stereotyped manner such that deep-layer neurons, layers 5 and 6 (L5 and L6), are born before upper-layer neurons [12]. It has previously been shown that the transcription factor Fezf2 is necessary and sufficient for the generation of deep-layer subcerebral projection neurons (SCPNs) [13][14][15][16]. Collectively, these studies suggest that Fezf2 functions high in a transcriptional hierarchy that regulates SCPN development and fate specification [14,[17][18][19]. Despite the essential role of Fezf2 in neural development, little is known about how its transcription is precisely regulated.
To investigate the regulatory mechanisms controlling Fezf2 transcription, we utilized chromatin immunoprecipitation combined with high throughput DNA sequencing (ChIP-seq) to identify a transcription factor-binding signature around the Fezf2 locus. Guided by our ChIPseq data, we mapped a promoter that was sufficient for reporter gene expression in the cerebral cortex throughout cortical development, and investigated the activity of two putative Fezf2 enhancers. We demonstrate that a downstream enhancer, 434, is sufficient to drive expression in cortical progenitor cells while a putative upstream enhancer, 1316, is strictly active in deep-layer neurons within the cortex. Taken together, these results provide insight into the developmental programs that promote Fezf2 expression during development of the cerebral cortex.

Results
Fezf2 is dynamically expressed during development of the cerebral cortex Towards understanding how Fezf2 transcription is regulated, we first examined its expression during cortical neurogenesis ( Figure 1A-C'). Consistent with previous reports [20,21], at E12.5 Fezf2 expression was detected in ventricular zone (VZ) progenitors and deep-layer cortical neurons ( Figure 1A-A'). At E15.5 Fezf2 expression began to decline in L6, but remained high in L5 ( Figure 1B-B'). Expression at P0 was similar to E15.5 ( Figure 1C-C'). Thus, at late stages of cortical development, expression of Fezf2 in the cerebral cortex was restricted to three distinct domains: higher expression in L5 neurons, and lower expression in L6 neurons and progenitors ( Figure 1C-C'). This dynamic spatial and temporal expression pattern suggests that Fezf2 transcription is under the control of a complex regulatory program.

Identification of putative regulatory regions around the Fezf2 locus
Recent work has identified several transcription factors that are essential for cortical neuron fate specification and differentiation. Among these, SOX5 and TBR1 appear to directly regulate Fezf2 transcription [22][23][24]. Using chromatin isolated from E15.5 mouse cortices, we mapped the binding of transcription factors expressed in L5 and L6 that were previously shown to be important for cortical development. These included FOXG1, NFIB, SOX5, and TBR1 [22][23][24][25][26][27][28][29][30][31][32][33]. We found that FOXG1, SOX5, and TBR1 bound to the evolutionarily conserved region immediately upstream of Fezf2 ( Figure 2B), suggesting that this region may contain the Fezf2 promoter. Interestingly, all of these factors bound to a previously identified enhancer downstream of Fezf2, enhancer 434 (also known as E4) ( Figure 2B). In addition to the upstream region and enhancer 434, FOXG1 and TBR1 bound a region in the neighboring Cadps gene, near the 5' end of Fezf2 ( Figure 2C). Taken together, our ChIPseq results indicate that multiple transcription factors that are expressed in deep-layer neurons and are critical for cortical development bind at or near the Fezf2 locus.
To determine the significance of our ChIP-seq data we searched the VISTA Enhancer Browser (enhancer.lbl. gov) for sequences in and near Fezf2 that may regulate its expression [34]. This approach uncovered several enhancers around Fezf2. These included an enhancer at the 5' side of Fezf2, enhancer 1316, and the previously studied enhancer at the 3' side, enhancer 434 ( Figure 2A). Our ChIP-seq data demonstrated binding within or near these enhancers, suggesting they are likely important for regulating Fezf2 transcription within the cerebral cortex ( Figure 2B,C).

The 2.7 kb promoter is active across all Fezf2 expression domains
Previous work identified the 2.7 kb region upstream of and including the Fezf2 start codon as sufficient to drive forebrain expression before the onset of cortical neurogenesis [35]. This region was bound by multiple transcription factors in our ChIP-seq dataset ( Figure 2B). We generated stable transgenic lines expressing the LacZ reporter under the control of this 2.7 kb sequence to study its activity throughout cortical development. When the earliest-born cortical projection neurons were being generated at E11.5, we observed LacZ activity throughout the cortex that mimicked endogenous Fezf2 expression (compare Figure 3A with Figure 1A). We observed ß-gal expression in SOX2 + cortical progenitor cells and in TBR1 + and Tuj1 + postmitotic neurons ( Figure 3B  We next examined 2.7 kb promoter activity at P0. Whole mount X-Gal staining revealed expression throughout the cerebral cortex in a high anterior-medial to low posteriorlateral pattern ( Figure 3E). However, unlike endogenous Fezf2, which is expressed in the VZ and in deep cortical layers (L5 and L6), the 2.7 kb-LacZ transgene was active in all cortical layers (compare Figure 3F with Figure 1C). LacZ was expressed in SOX2 + progenitor cells in the VZ and in the CTIP2 + or SATB2 + projection neurons ( Figure 3G-I'). LacZ expression at P12 was similar to that observed at P0 (Additional file 1). Taken together, these results indicate that the 2.7 kb promoter is active throughout cortical development and is sufficient to drive transcription across all endogenous Fezf2 expression domains (progenitor cells, L5 and L6 neurons). However, its expression in upper cortical layers indicates that additional cis-elements are required to properly coordinate Fezf2 expression.

Enhancer 434 activity is strongest in cortical progenitor cells
Previous work has shown that enhancer 434 is highly conserved across vertebrate species and is necessary for Fezf2 expression in the cortex [36]. Deletion of this  enhancer from either a bacterial artificial chromosome (BAC) containing a Fezf2-EGFP reporter or from its endogenous locus resulted in a loss of EGFP or Fezf2 expression in the cerebral cortex [36]. However, it remains unknown whether enhancer 434 alone is sufficient to promote Fezf2 expression in subcerebral projection neurons.
We generated stable transgenic lines expressing LacZ under the control of enhancer 434 and the hsp68 minimal promoter [37]. At E11.5, ß-gal was expressed in SOX2 + and PAX6 + cortical progenitor cells ( Figure 4A-B), including the metaphase cells that were labeled by PHH3 antibody at the ventricular surface ( Figure 4C). In addition, Tuj1 + neurons located in the preplate expressed ß-gal ( Figure 4D). Whole mount staining of P0 brains revealed strong LacZ activity throughout multiple brain regions ( Figure 4E). However, the majority of enhancer 434 activity was in the VZ of the cortex and basal ganglia ( Figure 4F). Immunohistochemistry with antibodies for LacZ and the radial glial cell (RGC) marker BLBP demonstrated significant co-localization ( Figure 4G). Additionally, some LacZ + cells in the VZ expressed the RGC marker PAX6 ( Figure 4H). Some cells in the cortical plate expressed LacZ, however a minority of these LacZ + cells expressed the projection neuron markers SOX5 and SATB2 (Additional file 2). Thus, at P0 enhancer 434 activity is strongest in cortical progenitor cells. This strong LacZ signal observed in the VZ is consistent with the activity of enhancer 434 reported by the VISTA Enhancer Browser (enhancer.lbl.gov).
In adult mice, LacZ expression was observed in the cerebral cortex, white matter, hippocampus, and striatum ( Figure 4I-L). Most of the LacZ + cells exhibited typical astrocyte-like morphology and expressed the astrocyte marker GFAP ( Figure 4K, L). A minority of LacZ + cells in the cerebral cortex expressed the cortical projection neuron marker SATB2 (Figure 4J'). This shift in enhancer 434 activity suggests that it may have divergent functions during development versus in adult animals. Taken together, these data demonstrate that during cortical development enhancer 434 activity is strongest in progenitor cells, one of the endogenous expression domains of Fezf2.

Enhancer 1316 is active in deep-layer projection neurons in the cerebral cortex
We next investigated the activity of enhancer 1316. MultiZ alignment of this region showed strong vertebrate conservation, especially within the middle 500 bp (Additional file 3). We generated stable transgenic lines expressing LacZ under control of enhancer 1316 and the hsp68 minimal promoter. At E11.5, ß-gal was present throughout the newly generated preplate ( Figure 5A), in Tuj1 + neurons ( Figure 5D). LacZ + cells did not express the progenitor cell markers PAX6 or SOX2 ( Figure 5B, C).
Whole mount staining at P0 revealed LacZ activity in deep layers of the cerebral cortex ( Figure 5E). LacZ was expressed in the subplate and in some deep-layer neurons, but not in cortical progenitors or astrocytes ( Figure 5F). Some LacZ + cells expressed CTIP2 and SATB2 ( Figure 5H-I'). 'None of the LacZ + cells expressed SOX2 ( Figure 5G, G'). Analysis at P12 revealed a similar LacZ expression pattern to that observed at P0 (Additional file 4). Taken together, these results indicate that within the cerebral cortex, enhancer 1316 is strictly active in deep-layer neurons.

Discussion
We set out to understand how expression of Fezf2 is spatially and temporally controlled during development of the cerebral cortex. Previous work demonstrates that Fezf2 transcription levels are critically important for specification of cortical projection neuron fates [13][14][15][16][17][22][23][24]31]. Our ChIP-seq data indicate that multiple transcription factors that are expressed in deep-layer neurons bind to evolutionarily conserved sequences around the Fezf2 locus. The large number of transcription factors that bound enhancer 434 underscores its developmental importance. Recent deletion analysis demonstrated that this enhancer is critical for subcerebral projection neuron identity and connectivity [36]. In the current study, the majority of enhancer 434 activity during cortical neurogenesis was in progenitor cells, suggesting that this element also functions to promote Fezf2 expression in the VZ, one of the endogenous Fezf2 expression domains ( Figure 6).
An alternative interpretation of this result is that different experimental strategies may account for the somewhat distinct conclusions regarding the activity of enhancer 434. In the study by Shim et al. [36], enhancer 434 was deleted from either a bacterial artificial chromosome containing a Fezf2-EGFP reporter or from its endogenous genomic locus. In both cases, deletion of enhancer 434 resulted in a loss of GFP or Fezf2 expression, respectively, and failed extension of subcerebral projections. In the present study, we assayed the activity of enhancer 434 in isolation and found that it drove LacZ activity strongest in progenitor cells with lower levels of reporter activity in postmitotic neurons. These data indicate that when taken out of its endogenous locus, enhancer 434 is sufficient to drive reporter expression in the progenitor cells. It is important to note however that both the Shim et al. study [36] and the data presented here indicate a critical role for enhancer 434 in the regulation of Fezf2 transcription.
Although the present study identifies conserved regulatory elements around the Fezf2 locus, functional enhancerpromoter interactions cannot be established based upon transgenic reporter assays alone. Specifically, whether enhancer 1316 functionally interacts with the Fezf2 promoter to regulate transcription remains unknown. Recently published high throughput chromosome conformation capture (Hi-C) suggests an interaction between enhancer 1316 and the Fezf2 promoter [38]. However, this enhancer does not appear to be essential for Fezf2 expression during development. Transgenic mice containing a modified BAC (RP23-141E17) that express EGFP or CRE from the Fezf2 open reading frame have been generated. Although enhancer 1316 is not contained within this BAC, GFP, or CRE expression in these transgenic mice recapitulates endogenous Fezf2 gene expression [36,39,40], suggesting that this enhancer is not essential for Fezf2 expression during development. Thus, it is possible that additional enhancer (s) present on the RP23-141E17 BAC may function redundantly with enhancer 1316 to promote Fezf2 expression in deep layer neurons.
Multiple cis-regulatory sequences around Fezf2 exhibited distinct yet overlapping activities during cortical development. These results are similar to the regulation of the eve gene locus in Drosophila where multiple, partially redundant enhancers control expression in individual stripes along the anterior-posterior axis of the embryo [41,42]. Ultimately, combined activity of these enhancers is required to drive the precise spatio-temporal expression of eve. Similar regulatory paradigms have also been uncovered in mammals through analysis of the interferon-beta [43] or Fgf8 [44] loci. In our study, neither of the identified enhancers nor the 2.7 kb promoter, when assayed in isolation, were sufficient to drive LacZ reporter expression in an identical pattern to that of endogenous Fezf2 expression. This suggests the presence of an additional enhancer (enhancers) that promotes high levels of expression specifically in L5 neurons. Moreover, the strong activity of the 2.7 kb promoter in upper cortical layers indicates that repressor sequences are necessary to restrict its activity to cortical progenitor cells and deep-layer neurons. Identification of these regulatory sequences combined with an increased understanding of the chromatin environment surrounding the Fezf2 locus should help to shed light on the mechanisms that precisely regulate its transcription.
We recently reported that in addition to deep-layer cortical neurons, Fezf2 is expressed in progenitor cells throughout cortical neurogenesis [39]. Phenotypic analysis of Fezf2 −/− mice indicates that it is required for both the patterning and maintenance of forebrain progenitors [45], and the fate specification of deep-layer cortical projection neurons [13][14][15][16]. In Fezf2 −/− mutant cortices, subcerebral projection neurons switch their identity to become callosal or corticothalamic projection neurons. However, it remains unclear whether Fezf2 functions within cortical progenitor cells, postmitotic neurons or a combination of both in order to properly specify the fate of deeplayer cortical projection neurons. Identifying distinct promoter and enhancer sequences that regulate Fezf2 transcription may help provide answers to this question.

Conclusions
Given recent progress identifying putative mammalian enhancers [37], it is critical that we understand the spatial and temporal activities of these sequences in the regulation of individual genes, in order to appreciate the deleterious effects of mutations in non-coding sequences identified through GWAS. Focusing on Fezf2, a key regulator of cortical development and deep-layer projection neuron fate, our study suggests that multiple enhancer and promoter activities can coordinate gene expression during brain development. Recent work indicates that Fezf2 is expressed in neocortical progenitors throughout development. Although mice lacking Fezf2 exhibit developmental defects largely within deep-layer neurons (L5 and L6), its role in cortical progenitors versus postmitotic neurons remains unclear. Identifying the regulatory mechanisms that control Fezf2 transcription will enable the functional dissection of its role within these distinct cellular populations. We propose that the combined activities of the 2.7 kb promoter, enhancer 434, and possibly enhancer 1316 constitute a cis-regulatory module that controls Fezf2 in distinct domains of the cerebral cortex, ultimately leading to its dynamic expression during cortical development ( Figure 6).

Animals
All experiments in this study were carried out in accordance with protocols approved by the IACUC at the University of California, Santa Cruz and were performed in accordance with institutional and federal guidelines. The day of vaginal plug detection was designated as embryonic day 0.5 (E0.5). The day of birth was designated as postnatal day 0 (P0). Transgenic mice were generated using standard protocols [46]. To generate stable transgenic lines, founders were outcrossed to CD1 wild-type females for a minimum of three generations before analysis. Number of founder lines generated: 2.7 kb-LacZ (2), enhancer 434-LacZ (3), enhancer 1316-LacZ (4).

Plasmids
Generation of the 2.7 kb-lacZ plasmid was previously described [35]. For enhancer 434-lacZ, the genomic region was obtained from mouse genomic DNA by PCR using primers: ATCGCTCGAGCAGGCTGTAGGATGG GCAGCAGGAGTTTC and ATCGAAGCTTGTAACA AGTCAGGTGAGCAGGCGGTA. The product was then cloned into the hsp68-lacZ expression vector using Gateway cloning according to the manufacturer's protocol [11]. Enhancer 1316-lacZ was generated from human genomic DNA using primers AAACCACACAGCTGGTTTCC and TTTCCCGATAGATCGTCAGC and cloned into the hsp68-lacZ expression vector [37].

Microscopy
Bright field and epiflourescence images were captured on an Olympus BX51 microscope using a Q Imaging Retiga EXj camera or a Keyence BZ-9000 microscope. Confocal images were captured on a Leica TCS SP5 confocal microscope. Images were processed using Adobe Photoshop CS5 to adjust brightness and contrast.

ChIP-Seq
Chromatin immunoprecipation was performed as previously described [24]. Antibodies used were Rabbit anti FOXG1 (Cell Signaling), Rabbit anti NFIB (Active Motif), Rabbit anti SOX5 (Abcam), and Rabbit anti TBR1 (Abcam). Sequencing libraries were generated using the Illumina TruSeq kit according to the manufacturer's protocol. Sequencing was performed on an Illumina Hiseq 2000 at the UCSC Genome Technology Center. Input DNA was sequenced as control. Sequencing reads were mapped to the mouse genome (mm9) using the Bowtie mapping algorithm. Non-overlapping reads and PCR duplicates were removed. Peaks were called using the MACS algorithm [48]. Basewise conservation is represented as greyscale darkness with higher conservation corresponding to darker values. Gap Annotation: single line, no bases in the aligned species; double line, aligning species has one or more un-alignable bases in the gap region; red coloring, aligning species has Ns in the gap region. Genomic Breaks: green square brackets, enclose shorter alignments consisting of DNA from one genomic context in the aligned species nested inside a larger chain of alignments from a different genomic context.

Competing interests
The authors declare that they have no competing interests.
Authors' contributions MJE: conception and design, data collection and analysis, manuscript writing, and final approval of the manuscript. KAL: data collection and analysis, final approval of the manuscript. WLM: data collection and analysis, manuscript writing, and final approval of the manuscript. CG: conception and design, data collection, manuscript writing, and final approval of the manuscript. RR: data collection and final approval of the manuscript. SK: analyzing sequencing data and final approval of the manuscript. AV: generating the enhancer 1316-LacZ reporter construct, manuscript editing, and final approval of the manuscript. JLRR: generating the enhancer 1316-LacZ reporter construct, manuscript editing, and final approval of the manuscript. BC: conception and design, data analysis, manuscript writing, and final approval of the manuscript. All authors read and approved the final manuscript.