The development of a neuron from a precursor cell comprises a complex set of steps ranging from regulation of the proliferative cycle through the acquisition of distinct morphology and functionality. How these processes are orchestrated is largely unknown. Using in utero manipulation of gene expression in the mouse embryonic cerebral cortex, we found that the transition between multipolar and bipolar stages of newborn cortical pyramidal neurons is markedly delayed by depletion of CoREST, a corepressor component of chromatin remodeling complexes. This profoundly affects the onset of their radial migration. The loss of CoREST function also perturbs the dynamics of neuronal precursor cell populations, transiently increasing the fraction of cells remaining in progenitor states, but not the acquisition of the neuronal glutamatergic fate of pyramidal cells. The function of CoREST in these processes appears to be independent of its best-known interactor, the RE-1 silencer of transcription/neural restrictive silencing factor, and requires the histone demethylase LSD1. This reveals the importance of epigenetic control in the execution of neural development programs, specifically in the cerebral cortex.
The mature cerebral cortex arises from the proliferation and differentiation of common precursor cells into a vast array of different neuronal and glial lineages. Newborn cortical neurons migrate from their place of birth in the proliferative zones to their target destination in the cortex; cortical pyramidal (glutamatergic) neurons are generated by precursor cells located in the ventricular zone (VZ) and subventricular zone (SVZ) of the embryonic dorsal telencephalon. During the stages of proliferation and differentiation, precursors of pyramidal neurons in the VZ and SVZ display complex behaviors and exist in distinct transient populations characterized by their morphology, behavior, and gene expression patterns (Noctor et al. 2004; Englund et al. 2005; Noctor et al. 2008; Costa et al. 2009; Stancik et al. 2010). Newborn pyramidal neurons migrate radially to occupy specific layers in the cortical plate and acquire distinct morphological and functional identities (Rakic et al. 2009). How the behaviors of the precursors of pyramidal neurons are controlled and how they are linked to the fate of neurons that originate from them is not clearly understood.
Transcriptional repression is central in the control of developmental programs, including neural differentiation. Corepressors are key for the regulation of gene expression, playing pathway-specific roles (Perissi et al. 2010; Riccio 2010). CoREST (CoREST1, Rcor1) is a corepressor discovered by its association with the transcriptional repressor REST/NRSF (Andrés et al. 1999). RE-1 silencer of transcription/neural restrictive silencing factor (REST/NRSF) has been proposed as a key regulator of neural gene expression during development (Ballas et al. 2005; JØrgensen et al. 2009). Most corepressors complexes, including those containing REST/NRSF and CoREST, act through epigenetic mechanisms promoting covalent modifications of chromatin (Ooi and Wood 2007). Studies on CoREST function indicate that it controls chromatin configurations that allow fine regulation of neural gene expression (Ballas et al. 2005). CoREST and REST/NRSF associate with hundreds of genes, including many encoding neuronal proteins such as ion channels and synaptic proteins. They also bind to genes encoding components of signal transduction pathways and regulatory networks, including some involved in the specification of neural cell types (Johnson et al. 2008; Abrajano et al. 2010; Qureshi et al. 2010). Although the set of genes to which CoREST and REST/NRSF bind show significant overlapping, as expected for components of a common complex, it has become evident that CoREST also binds to genes that are not associated with REST/NRSF (Abrajano et al. 2009). These data support the idea that CoREST can participate in the control of neural differentiation both in association with REST/NRSF or in a REST/NRSF-independent manner. We directly addressed the role of CoREST in the development of the cortical pyramidal neurons in vivo. We found that CoREST function is required for the appropriate timing of onset of neuronal migration as well as for the acquisition of mature morphological features but not for the acquisition of the glutamatergic neuron fate.
Materials and Methods
In Utero Electroporation
Protocols were approved by the Bioethics Committee of the Faculty of Medicine, Universidad de Chile and performed in accordance with institutional and NIH guidelines. In utero electroporation was carried out as described (Saito 2006). Timed pregnant CF1 mice at embryonic day 14.5 (E14.5) were anesthetized using isofluorane and the uterus was exposed through a laparotomy. One microliter of plasmid solution (1–2 μg/μL) mixed with 0.05% Fast Green in phosphate-buffered saline (PBS) was injected through the uterine wall into the lateral ventricle of the embryos. Five 40-V pulses of 50-ms duration at 1-s intervals were delivered across the embryonic head using 1-cm-diameter electrodes (Harvard Apparatus) and a square wave electroporator (Bio-Rad GenePulse Xcell). The laparotomy was closed and development was allowed to continue until E17.5 or postnatal (P) day 0 or 7. At these stages, brains were fixed by transcardial perfusion of 4% paraformaldehyde in PBS. Brains were embedded into agarose and sectioned using a vibratome (VT1000S, Leica) and mounted for direct visualization of enhanced green fluorescent protein (EGFP) fluorescence or for immunofluorescence analysis. The primary antibodies were obtained from Chemicon, unless stated otherwise. Secondary antibodies were from the Molecular Probes Alexa series. Images were obtained using a Disk Spinning Unit microscope (Olympus BX61WIDSU), analyzed using the Image J package, and assembled for display using Photoshop software.
For bromodeoxyuridine (BrdU) labeling, pregnant mice were intraperitoneally injected with BrdU (Sigma-Aldrich, 50 mg/kg in sterile saline) at stage E16.5. Mice were anesthetized 4 h after this injection and embryonic brains were processed as described above. Brain sections were blocked with 4% donkey serum for 2 h at room temperature, incubated overnight with anti-GFP antibody (1:500; Invitrogen), washed, and then incubated with anti-rabbit FC-FITC-conjugated antibody (1:250; Jackson Immunoresearch Laboratories). Sections were then treated with 0.1 N HCl for 6 min at 37 °C, followed by neutralization in PBS for 10 min and blocking, after which immunohistochemistry was performed using a rat anti-BrdU antibody (1:500; Abcam) and donkey antirat FC-Cy3-conjugated secondary antibody (1:500; Jackson Immunoresearch Laboratories). The number of BrdU-positive cells was counted in 3 different fields from the lateral ventricle zones of E16.5 brains with 2 sections per embryo from 5 control and 5 CoREST shRNA embryos. The percent of cells was calculated dividing the number of GFP-positive/BrdU-positive cells by the total number of GFP-positive cells. Student's t-test was used to determine the statistical significance. All comparisons were made between littermates.
Analysis of Neuronal Migration
For quantification of neuronal migration along the radial axis of the cerebral cortex, images were divided into 4 bins by lines tangential to the ventricular surface, which represented the closest possible approximation to the histological compartmentalization of the developing cortex at a given stage. These bins are termed cortical plate (CP, bin 4), intermediate zone (IZ, bin 3), subventricular zone (SVZ, bin 2), and ventricular zone (VZ, bin 1). The proportion of GFP-expressing neurons in each bin was counted. In some case, images were digitally segmented to better define cell contours. In all cases, n represents the number of sections considered for the analysis, each section representing an independent brain; brains of at least 3 different experiments (litters) were analyzed. For analysis of neuronal morphology, high magnification images were used and cells were ascribed to 1 of 3 morphological categories (multipolar, unipolar/bipolar, and rounded) at E17.5 or the number of processes (P7) was counted. At least 20 cells, from at least 3 different brains were counted in each condition. Results are indicated as mean ± standard error of the mean . P values were obtained by using 2-tailed Student's t-test.
cDNAs designed to encode short-hairpin RNAs (shRNAs) targeting CoREST, REST/NRSF, and LSD1 were cloned into the pZoff plasmid, which uses the human H1 RNA promoter for the expression of the shRNAs. H1-shRNAs expression cassettes were subcloned into the pFUXH1 plasmid, which expresses EGFP under the control of human ubiquitin C promoter. The resulting plasmids are termed p-shCoREST, p-shREST/NRSF, and p-shLSD1. In all cases, the efficacy of several interference constructs was assessed in cultured cell by western analysis and the most effective was chosen for subsequent work. The sequences of the cDNAs encoding the shRNAs are:
A pFUXH1 plasmid encoding a nonsense combination of the bases contained in the cDNA encoding shCoREST was used as control. Sequence of this construction was:
A p-shCoREST-insensitive cDNA encoding CoREST was obtained by polymerase chain reaction (PCR)-mediated silent mutagenesis of mouse CoREST with the primer gGGaATGTTctTaagcCAgGAAGATGTGGAGGCTGTGTCT. Mutated nucleotides are in lowercase and were corroborated by sequencing. This construction was cloned into the pCAGIG expression plasmid and is denoted as pCoREST△. Subsequent mutagenesis on this background was performed by PCR. Deletions carried out on the full length p-shCoREST-insensitive CoREST comprised aa 1–153 (pCoREST△-△N), aa243–382 (pCoREST△-△C), aa 193 (K193) (pCoREST△-△S1), and aa 154–174 (pCoREST△-△S2/3). S1 and S2/3 stand for sumoylation by SUMO1 and interaction with SUMO2/3, respectively. The design was based in published studies of CoREST structure–function relationship (Ballas et al. 2005; Shi et al. 2005; Muraoka et al. 2008; Ouyang et al. 2009). The expression of all these constructs was verified by transfection into HEK293 cells followed by western blot using and anti-HA antibody, which was included in all the mutant forms.
To address the role of CoREST in the developing mouse cerebral cortex, we first investigated its expression pattern. In order to examine the time course of CoREST expression in the developing mouse brain, we prepared whole-cortex extracts at different ages that were subject to western analysis. CoREST was detectable at E14.5 as a band of 66 kDa (Fig. 1A). This band reached its highest intensity between P0 and P15 and declined thereafter. We conclude that CoREST is expressed in the developing brain and that its expression spans the period of active embryonic neurogenesis. To examine the spatial expression pattern of CoREST in the developing neocortex, we probed coronal sections of mouse embryonic brains at different stages of development. CoREST immunostaining displayed a nuclear pattern apparent in all cortical layers (Fig. 1B). CoREST signal was detectable in Tuj1-positive neurons of the embryonic cortical plate (Fig. 1C). CoREST signal was also detectable in proliferative zones. In the adult mouse brain, CoREST is expressed in all neurons of the cortex (Fig. 1D). Together, these results indicate that CoREST is present in both developing neurons and mature cortical neurons.
In order to determine the role CoREST plays in neurogenesis, we constructed and screened small hairpin expression plasmids targeting the CoREST transcript. We selected a construct that efficiently knocked down CoREST expression in vitro in cultured cortical neurons (not shown) and N2A cells (Fig. 2A). This construct (p-shCoREST) was delivered through in utero electroporation into dorsal cortical progenitors at E14.5. This approach targets the precursors of pyramidal neurons that populate layers 2–3 of the mature cerebral cortex. A missense shRNA was used as a control. We confirmed the efficiency of p-shCoREST in vivo by immunofluorescence analysis of electroporated embryonic mouse brains (Fig. 2B). We observed a robust decrease of the expression of CoREST throughout the period extending from E17.5 (Fig. 2B) throughout postnatal day 7 (P7) (Supplementary Fig. S1). Examination of the developing cortex at E17.5 revealed that depletion of CoREST produced a marked impairment of cell migration. Thus, at E17.5, 14.3 ± 2.2% of control cells were found in the VZ and 36.2 ± 3.3% had reached the CP. In contrast, 33.2 ± 4.3 of the CoREST-depleted cells remained in the VZ and only 14.0 ± 1.7% were in the CP (n = 7, P < 0.01, Fig. 3). It must be noticed that CoREST-depleted neurons do not restrict to the VZ/SVZ but are also found in the IZ. The differences in cell migration were not apparent at E16.5, stage at which all cells are found at the VZ/SVZ in both control and CoREST-depleted cells (data not shown). In spite of the sustained depletion of CoREST, neurons ultimately reached the CP, as observed at P0 and P7. However, these neurons displayed morphological defects, including rounded soma instead of the characteristic pyramidal contour and a significant reduction in the number of basal dendrites (7.4 ± 2.1 vs. 5.1 ± 2.1, P < 0.001, Fig. 5C; see also Supplementary Fig. S1). The defects of migration were rescued by coelctroporation of an shRNA-insensitive form of CoREST, pCoREST△ as shown in Figure 4. The sole overexpression of CoREST, by IUEP of pCoREST△ did not associate to differences in the migration of newborn neurons (Supplementary Fig. S2). In addition to the migratory defect, CoREST-depleted cells markedly retained the multipolar shape that is characteristic of the intermediate progenitor cells and newborn neurons in the SVZ and did not acquire the bipolar shape associated with the radial migration of neurons from the SVZ to the CP. At E17.5, 64.6 ± 4.6% of control cells acquired a bipolar appearance versus 33.8 ± 4.6% of CoREST-depleted cells (n = 4, P < 0.05, Fig. 5A,B).
The apparently reduced migration of CoREST-depleted neurons could be an indirect consequence of changes in the rates of cell proliferation or cell death. To gain insight into the mechanism accounting for the observed difference in cell positioning caused by CoREST knockdown, we injected the cell proliferation marker BrdU into pregnant mice at E16.5, 4 h prior to collection of the brains which had been electroporated at E14.5. We performed immunostaining for BrdU and counted the percent of GFP/BrdU-labeled cells with respect to the total number of GFP-expressing cells. No differences were observed between cells carrying the control plasmid and CoREST-depleted cells (21.45 ± 2.51% of in the control vs. 18.93 ± 1.83 of CoREST-depleted cells, data from 5 brains in each condition; at least 100 cells per brain were counted; Supplementary Fig. S3). In addition, very few cells were immunostained by the cell death marker activated caspase-3, and no difference was seen between control and CoREST shRNA (Supplementary Fig. S4). Together, these results do not support the view that the defect in neuronal migration caused by depletion of CoREST arises from changes in cell division or apoptosis. Additionally, and since radial migration depends of the interaction of newborn neurons with radial glia cells, we examined the gross architecture of this migration scaffold by detecting the expression of the intermediate filament protein vimentin. Depletion of CoREST did not alter this scaffold (Supplementary Fig. S5).
We examined the CoREST-depleted cells that were delayed in the VZ/SVZ at E17.5, since this was the earliest appreciable defect associated with decreased CoREST expression and asked if they retained a precursor cell identity, defined by the expression of the transcription factors Sox2 and Tbr2; these characterize neuroepithelial cells and radial glia cells (Sox2) and intermediate progenitors (Tbr2). We found that 32.4 ± 2.4% of cells electroporated with p-shCoREST expressed Sox2 at E17.5 versus 14 ± 2.9% of the controls (n = 5, P < 0.01, Fig. 6, left panels). Additionally, 36.1 ± 2.8% of cells carrying p-shCoREST expressed Tbr2 versus 15.0 ± 2.7% of the controls (n = 5, P < 0.05, Fig. 6, right panels). These results suggest that decreased CoREST levels alter early stages of neuronal development, including the transitions between populations of precursor cells.
CoREST function has been linked to the specification of neural lineages (Qureshi et al. 2010). We investigated the acquisition of the terminal fate of CoREST-depleted cells as they exited the SVZ and entered the IZ and CP. We found that all CoREST-depleted cells at these stages expressed β-tubulin (Supplementary Fig. S6) and MAP2 (data not shown), whereas none expressed glial markers (data not shown). This indicates that CoREST, in our experimental conditions, does not influence the neuronal fate. Furthermore, we evaluated the expression of the vesicular glutamate transporter and found an expression pattern indistinguishable from that observed in control cells (Supplementary Fig. S6). Furthermore, neurons reach the expected cortical layers according to their date of birth. The identity of the layers was determined by the enriched expression of Cux1 (layer II/II) and Tbr1 (deeper layers, Supplementary Fig. S7). Thus, cells with decreased levels of CoREST acquire their normal identity as glutamatergic neurons and reach the appropriate cortical layers. These results are consistent with the idea that epigenetic regulation, and particularly, CoREST function is required for the correct timing of execution of developmental programs of cortical pyramidal neurons but not for the specification of fate.
We sought to advance our understanding of the molecular mechanisms by which CoREST participates in the control of pyramidal cell development. It is known that CoREST interacts in different contexts with transcription factors as well as with chromatin modifying enzymes, although the occurrence and physiological relevance of these interactions during neural development are not known. We used the defect of the timing of onset of neuronal migration, expressed as percent of cells remaining in the VZ/SVZ at E17.5 after electroporation at E14.5, as an easily quantifiable parameter to test the ability of different maneuvers to rescue or mimick the effect of CoREST depletion. We made deletion constructions aimed to test the relevance of these interactions in the background of pCoREST△ and tested their ability to revert the phenotype caused by p-shCoREST. pCoREST△, which in its full length form completely rescues the migration defect caused by p-shCoREST (see Fig. 4), encodes the p-shCoREST-insensitive full length form of CoREST.
The best-known interactor of CoREST is REST/NRSF. REST/NRSF is expressed in nonneural cells and in neural progenitors and is expressed during cortical development (Ballas et al. 2005, and data not shown). We tested the ability pCoREST△-△N, which lacks the domain of association with REST/NRSF to rescue the migration defect caused by depletion of CoREST. We generated pCoREST△-△N and the other constructs mentioned below, using pCoREST△ as the backbone, which render the respective transcripts insensitive to the interference by p-shCoREST; all constructs include an HA epitope, which was used to probe the expression of these deletion mutants before testing them in vivo (Supplementary Fig. S8). We coelectroporated p-shCoREST and pCoREST△-△N at E14.5 and focused on neuronal migration at E17.5. We observed that pCoREST△-△N was able to sustain the migration of newborn cells in a way undistinguishable from the control condition, which differs markedly from the CoREST depletion condition (Fig. 7A, i–iii). This result suggests that REST/NRSF or at least its interaction with CoREST is not critical for pyramidal neuron development. To directly test this idea, we used p-shREST/NRSF, which effectively decreases REST/NRSF expression (Supplementary Fig. S9) to electroporate progenitors at E14.5 and assessed the migration at E17.5. We did not observe differences with respect to the control (Fig. 7A, iv).
CoREST bridges transcription factors to proteins that modify chromatin structure, such as LSD1, HDAC1, and HDAC2. LSD1 and CoREST are mutually required to sustain histone demethylation (Shi et al. 2005; Foster et al. 2010). These data led us to propose that LSD1 functioning in association with CoREST could be involved in the control of pyramidal neuron development. First, we tested the expression of LSD1 in the developing brain cortex and found it to be ubiquitously distributed (Supplementary Fig. S10). Then, we asked if pCoREST△-△C, which lacks the domain of association with LSD1 was able to revert the effect of CoREST depletion when coelectroporated with p-shCoREST at E14.5. As shown in Figure 7A, v, cells depleted of CoREST and containing pCoREST△-△C were not distinguishable from CoREST-depleted cells. We then directly decreased LSD1 expression by electroporation of p-shLSD1, which encodes a tested shRNA targeting LSD1 (Sun et al. 2010). This mimicked the migration phenotype caused by p-shCoREST (Fig. 7A, vi). Taken together, these results support the idea that CoREST/LSD1-mediated chromatin modification is required for early cortical development. Additionally, modification by SUMO1 has been proposed to modulate CoREST participation in REST/NRSF-dependent transcriptional repression (Muraoka et al. 2008). Deletion of the sumoylation motif does not impair the ability of CoREST to sustain a normal pattern of cortical development (Fig. 7A, vii). Furthermore, the interaction of CoREST with SUMO2/3 also appears as important for CoREST function in other contexts (Ouyang et al. 2009). However, the deletion of the motif involved in this interaction did not impair the ability of CoREST to sustain the normal migratory pattern (Fig. 7A, viii).
Epigenetic regulation is emerging as a key mechanism orchestrating the time- and space-regulated patterns of gene expression that sustain neural development and function (Riccio 2010). Most advances in the field come from studies leading to detailed molecular characterization of the structure and function of epigenetic regulatory complexes and to the identification of genes regulated by them. However, the ubiquity of expression of many components of these complexes and their diversity of functions in different cellular contexts pose a challenge to test and translate the models emerging from molecular studies into the context of physiological development. In the case of the developing cerebral cortex, the diversity of precursor cells, the onset of neuronal activity, and changing microenvironmental cues should be considered to contribute to the functional outcome and significance of epigenetic regulation. CoREST is a component of corepressor complexes that act epigenetically; it has been studied mainly in association with the transcriptional repressor REST/NRSF. As REST/NRSF is proposed to regulate key steps of early neural development, it is highly relevant to understand the physiological role of CoREST. Recent studies have demonstrated that in addition to REST/NRSF, CoREST binds with other transcription factors and regulates processes as diverse such as control of inflammatory process in nervous tissue (Saijo et al. 2009), hematopoietic differentiation (Saleque et al. 2007; Hu et al. 2009), and heat shock response (Gómez et al. 2008). Many of genes have been identified as potentially regulated by CoREST; their analysis suggests an important role in controlling of pluripotency of progenitor cells and the lineage determination in neural progenitor cells (Qureshi et al. 2010). Our work addresses for the first time the role of CoREST in the in vivo development of the nervous system and highlight the emerging role that epigenetic control plays in cortical development.
The most striking defect produced by the loss of CoREST function in the developing cortex is the delay in the onset of radial migration of newborn pyramidal, with cells remaining for a longer than normal period in the proliferative zones. More specifically, our observations indicate that regulation by CoREST is required for the timing of the transition from the multipolar to the bipolar stage (Noctor et al. 2004). Mutation of several human genes isolated on the basis of its relationship with neurological development diseases affect pyramidal neuron migration and loss of function of some of the ortholog genes in the mouse embryonic cortex leads to an impairment of the exit from the multipolar stage toward the bipolar stage (Bai et al. 2003; Tsai et al. 2005; LoTurco and Bai 2006). Deletion of Cdk5 also leads to a marked disruption of cortical architecture in the mouse brain, which appears to originate in the disruption of the multipolar to bipolar transition (Ohshima et al. 2007). Disregulation of gene expression caused by depletion of CoREST could impact upon one or more or the pathways already characterized; nevertheless, it is worth mentioning that the exit from the multipolar stages appears to be “a point of vulnerability to disruption” (LoTurco and Bai 2006) and therefore, the disruption of still undiscovered pathways could account for the defects we observe. In this respect, the comprehensive identification of the genes disregulated by loss of CoREST specifically in neural progenitor cells and newborn neurons should prove useful to identify determinants of the transition from the multipolar to bipolar stages of pyramidal neuron migration. It is important to note that, in contrast to the definitive arrest of migration associated with mutation of human genes which leads to histological abnormalities such as heterotopias, or the case with Cdk5, most or all CoREST knockdown cells ultimately reach the cortical plate, albeit later than normal, despite sustained depletion of CoREST and supposedly sustained disruption of gene expression.
In addition to the perturbation of the timing of multipolar to bipolar transition, changes in the proportion of cells at different stages of migration could also originate in variations in the numbers of cells that are born or die within the temporal frame investigated. We did not find evidence supporting this view, as the proportion of CoREST-depleted cells labeled with BrdU at E16.5 was indistinguishable from the controls. Equally, the number of episodes of apoptotic cell death did not differ. Additionally, we observed that depletion of CoREST is associated with an increased proportion of cells expressing Sox2 or Tbr2 at E17.5. It is interesting to note that changes in cell cycle length, caused by overexpression of Cdk4/CyClinD1 associate to changes in the proportion of proliferative to neurogenic divisions (Lange et al. 2009), particularly increasing the number or intermediate progenitors. This change results in a transient phenotype of retention of cells in the VZ/SVZ, which is comparable to the one observed when cells are depleted of CoREST. Therefore, CoREST could also be regulating the expression of genes involved in controlling cell cycle length. It has been reported that CoREST associates with genes involved in the specification of neuronal identity (Abrajano et al. 2010). Nevertheless, disruption of CoREST expression throughout the stages from neural progenitor cells to mature neurons does not appear to affect the neuronal identity in vivo, as CoREST-depleted cells acquire their expected fate as glutamatergic neurons. Although, after exiting the VZ/SVZ, the general process of migration and positioning in their cortical plate appears conserved, CoREST-depleted neurons display conspicuous morphological defects, particularly a reduced number of dendrites. Our experimental design does not allow discerning if the defects we observed in postnatal pyramidal neuron morphology are a consequence or early defects in progenitor development, early stages after neuronal birth or if they arise from the perturbation of later process independently regulated by CoREST.
The current level of understanding of the cell signaling networks that CoREST integrates is not sufficient to propose definite mechanisms to explain how it participates in the regulation of cortical development. Nevertheless, it is interesting to note that the action of CoREST in this process appears to be independent of REST/NRSF. This is supported by our finding that cortical neurons migrate normally when transfected with a form of CoREST lacking its domain of interaction with REST/NRSF and also when transfected with an shRNA that directly knock down REST/NRSF. This is in agreement with data obtained from in vitro and molecular studies suggesting that CoREST may act in a REST/NRSF-independent way, regulating different sets of genes and processes. Our study did not address a role for CoREST or REST/NRSF at earlier stages, in which they may function together. Additionally, the deletion of the domain of interaction with SUMO2/3, which has been shown to be necessary for CoREST to act as a corepressor in a REST/NRSF-independent manner (Ouyang et al. 2009) was without effect. Taking into account our understanding of CoREST function, it is reasonable to propose that its role in the cortex may be in association with still unidentified transcriptional repressors; their isolation, as well as of the genes that they regulate should prove useful to advance our understanding of early cortical development. Our work identifies LSD1 and thus histone demethylation as a main mediator of CoREST function in pyramidal neuron development, which is consistent with the view of their close functional association (Shi et al. 2005; Foster et al. 2010) and the increasingly apparent relevance of LSD1 in neural development control (Sun et al. 2010; Zibetti et al. 2010). It must be noticed that, although the delay in the onset of migration is mimicked by the decrease of LSD1, the defects in dendrite morphology are not observed in the latter case (data not shown). This suggests that different mechanisms lie downstream this complex. Importantly, this observation supports the idea that the defects in dendritic morphology specifically originate from the deficit of CoREST rather than being secondary to a delayed migration.
It is known that CoREST also binds to the histone deacetylases HDAC1 and HDAC2 (Perissi et al. 2010), and therefore histone deacetylation could also be a mechanism linking CoREST to regulation of gene expression relevant for early pyramidal neuron development. The role of HDAC1 was not directly explored in our study; nevertheless the ability to rescue CoREST depletion defects with a form of CoREST lacking the amino terminus argues against an important role of CoREST/HDAC1 association; in contrast, we cannot discard that HDAC2 could contribute as an effector.
Our findings reveal for the first time the relevance of epigenetic regulation in the development of the cerebral cortex and more broadly in neural development. Furthermore, they constitute the first study aimed to unravel the in vivo physiological role of CoREST and associated complexes. In spite of the great number of genes and pathways predicted to be regulated by CoREST and which could, when disrupted, catastrophically alter the viability or fate of pyramidal neurons, the defects caused by depletion of CoREST appears as restricted to specific aspects of neuronal development. This should allow using this regulatory pathway to identify transcription factors associated to CoREST and genes regulated by them in the control of the multipolar to bipolar transition and the choice of type of division of neuronal progenitor cells. Thus, our findings open a new avenue to understand the molecular mechanisms underlying the architectural complexity of the cerebral cortex.
ACT-47 (Chile); FONDECYT 1090281 (Chile); ICM-PP07-048-F (Chile); CONICYT (Chile) Fellowships (to P.F. and J.C.).
We thank Dr Pedro Zamorano (U. Antofagasta) for the silencing vector, Dr A.R. Kriegstein's lab (UCSF), particularly Dr D. Castañeda-Castellanos for advice on IUEP procedures, and Dr J. Sierralta and Dr A. Couve for comments and support. Conflict of Interest : None declared.