SEARCH

SEARCH BY CITATION

Keywords:

  • RE1 silencing transcription factor;
  • Chromatin;
  • Histones;
  • Neural stem cell;
  • Epigenetics;
  • Transcription

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

The control of gene expression in neural stem cells is key to understanding their developmental and therapeutic potential, yet we know little of the transcriptional mechanisms that underlie their differentiation. Recent evidence has implicated the RE1 silencing transcription factor (REST) in neuronal differentiation. However, the means by which REST regulates transcription in neural stem cells remain unclear. Here, we show that REST recruits distinct corepressor platforms in neural stem cells. REST is able to both silence and repress neuronal genes in embryonic hippocampal neural stem cells by creating a chromatin environment that contains both repressive local epigenetic signature (characterized by low levels of histones H4 and H3K9 acetylation and elevated dimethylation of H3K9) and H3K4 methylation, which are characteristic of gene activation. Furthermore, inhibition of REST function leads to activation of several neuron-specific genes but does not lead to overt formation of mature neurons, supporting the notion that REST regulates part, but not all, of the neuronal differentiation program.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Recent years have seen an intense interest in neural stem cells (NSCs) fueled by the promise of their therapeutic potential (reviewed in [1]) and their use as tools to unravel the molecular mechanisms underlying neurogenesis [2]. Understanding the mechanisms by which NSCs differentiate into neurons is key to our unlocking this potential. The transition from NSC to neuron is accompanied by wholesale changes in gene-expression patterns ultimately resulting in transcriptional activation of cohorts of pan-neuronal and subtype-specific genes. Many genes associated with differentiated neuronal function such as those encoding neurotransmitters and their receptors, molecules required for vesicle trafficking and release, guidance cues involved in axonal outgrowth, and synaptogenesis are either expressed at low levels or not at all in neuronal progenitor cells and are expressed only upon terminal differentiation [3, 4]. How these transcriptional programs are established and maintained is largely unknown. Although numerous transcription factors have been identified that play key roles in neurogenesis [5, [6], [7], [8]9], the identity of their target genes remains unknown for the most part. An understanding of these complex transcriptional programs requires identification of the targets of these transcription factors and insight into how the interaction of a transcription factor with its targets changes as a function of cell type and developmental stage.

One transcription factor that has been proposed to play a pivotal role in NSC differentiation is the RE1 silencing transcription factor (REST) (aka neuron restrictive silencer factor, NRSF). REST is a Krüppel-type zinc finger transcription factor that interacts with a 21-bp RE1 cis-element found in the regulatory regions of numerous neuron-specific genes, including those encoding neurotransmitter receptors, SNAREs (soluble NSF [N-ethylmaleimide-sensitive factor] attachment protein receptor), and associated proteins, ion channels, and neuronal trophic factors [8, 9]. Originally, REST was proposed as a silencer of neuronal gene expression in non-neural tissue [10, 11]. However, it is becoming increasingly clear that REST has multiple actions that vary according to cell type and developmental stage. During development of the neural tube, REST is expressed widely throughout the ventricular zone of the neuroepithelium [10, 11], whereas in the adult central nervous system, expression is more heterogeneous and REST is most evident in the hippocampus, midbrain, and hindbrain [12, [13], [14]15].

A corollary to this supposition is that this context dependence is a reflection of differential cofactor recruitment and/or endogenous chromatin structure operative at distinct developmental stages. This latter view is substantiated by our earlier studies in which we demonstrated that differential corepressor recruitment occurs even in cell lines of similar embryological origin [8, 16–18, [17], [18]] and that distinct chromatin environments are associated with individual target genes [16, 18].

There is ample reason to believe that NSCs are heterogeneous with respect to positional identity and developmental stage [19, [20]21]. Likewise, neurons generated either from different NSCs or by use of alternative differentiation paradigms reflect this heterogeneity. So for instance, retinoic acid differentiation of embryoid bodies gives rise to populations of neural progenitors and neurons, with very low or undetectable levels of REST protein [22], whereas NSCs and differentiated neurons derived from adult hippocampus both contain REST protein. Indeed, in the latter case, REST is required for subsequent differentiation of adult hippocampal stem cells into neurons [14]. Collectively, these studies clearly indicate that REST plays distinct functional roles in different NSC backgrounds. Here, we have investigated these ideas by examining REST interactions with its target genes in MHP36 NSCs derived from embryonic hippocampus. MHP36 NSCs were originally isolated from the transgenic mouse, H-2Kb-tsA58 [23]. They are able to differentiate in vitro [24] and in vivo where they have been demonstrated to restore function in the ischemia-damaged brain of rats [25] and marmosets [26].

In these NSCs, we find that REST is present at both silent and actively transcribed loci and acts as both a repressor and silencer. REST recruitment results in modification of the local chromatin structure around the RE1 sites, and although inhibition of its function can lead to early stages of neuronal differentiation, this is not sufficient to cause differentiation to mature neurons.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

MHP36 Cell Culture

Cells were cultured on fibronectin-coated tissue-culture dishes at 37°C in Dulbecco's modified Eagle's medium/F-12 without l-glutamine, with pyridoxine (Invitrogen Corporation, Carlsbad, CA, http://www.invitrogen.com) with the following additives: 0.03% bovine serum albumin (BSA) pathocyte (Valeant Pharmaceuticals, Costa Mesa, CA), 94 μg/ml human apotransferrin, 15 μg/ml human putrescine dihydrochloride, 4.7 μg/ml bovine pancreatic insulin, 380 ng/ml l-thyroxine, 317 ng/ml tri-iodo-l-thyronine, 58 ng/ml progesterone, 1.9 mM l-glutamine, 38 ng/ml sodium selenite, heparin sodium salt (9.4 units per milliliter) (all from Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), 47 μg/ml gentamicin and geneticin G418 (200 μg/ml) (both from Invitrogen Corporation), and 10 ng/ml human basic fibroblast growth factor and 10 ng/ml murine interferon (both PeproTech, Rocky Hill, NJ, http://www.peprotech.com). Cells were treated with 100 nM trichostatin A (TSA) for 24 hours or an equivalent volume of 100% methanol as a control. In separate experiments, cells were infected with adenovirus-expressing green fluorescent protein (GFP) or adenovirus-expressing GFP and the DNA-binding domain of REST (DN:REST) for 48 hours at which point cells were approximately 80% confluent and 70% expressing the adenovirus as judged by GFP fluorescence.

Transfection of MHP36 Cells

Cells were transfected using nucleofector kit (amaxa Inc., Gaithersburg, MD, http://www.amaxa.com). The transfection protocol was optimized using the optimization kit and carried out according to the manufacturer's instructions. One × 106 cells were resuspended in 100 μl of solution V. Two micrograms of DNA purified using an endo-free maxi-prep (Qiagen Inc., Valencia, CA, http://www1.qiagen.com) was used in the ratio 5:1 pMTDN:REST [27] to pmaxGFP. After 24 hours, media were replaced and transfection efficiency assessed using GFP fluorescence. For pmaxGFP-alone controls, an efficiency of 80% was routinely achieved. For pmaxGFP cotransfected with pMTDN:REST, an efficiency of 50%–60% was achieved. GFP was visible up to 9 days after transfection.

Immunocytochemistry

Cells were fixed in 3% paraformaldehyde and permeabilized with 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 1 hour. Nuclear staining was visualized with 4,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). The following primary antibodies were used: goat anti-REST p18 (1:100), rabbit anti-Sin3a (1:200), and rabbit anti-Sin3b (1:200) (all Santa Cruz Biotechnology, Inc., Santa Cruz, CA, http://www.scbt.com), rabbit anti-CoREST (1:200) (Upstate/Chemicon, U.K.), rabbit anti-Tau (1:200), chicken anti-MAP2 (1:400), and rabbit anti-glial fibrillary acidic protein (GFAP) (1:500) (all Abcam, Cambridge, U.K., http://www.abcam.com), mouse monoclonal anti-nestin immunoglobulin G1 (IgG1) (1:4) (Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww), mouse monoclonal amti-NeuN (1:500) (Chemicon International, Temecula, CA, http://www.chemicon.com), and mouse monoclonal anti-β III tubulin IgG2a-Tuj1 (1:50) (Covance, Princeton, NJ, http://www.covance.com). Secondary antibodies used were donkey anti-goat IgG-cyanine 3 (cy3) (1:1,000) and donkey anti-rabbit IgG-fluorescein isothiocyanate (FITC) (1:200) (both Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, http://www.jacksonimmuno.com), goat anti-chicken IgY-cy3 (1:800) (Abcam), and goat anti-mouse IgG1-FITC (1:133) and goat anti-mouse IgG2a-TRITC (tetramethylrhodamine B isothiocyanate) (1:133) (both SouthernBiotech, Birmingham, AL, http://www.southernbiotech.com). For colocalization, images were viewed using a Delta Vision confocal microscope (Applied Precision, Issaquah, WA, http://www.api.com/index.html). After washing with PBS, fluorescence was examined using an Olympus IX-70 inverted microscope (×40 oil immersion lens) (Olympus, Tokyo, http://www.olympus-global.com), and three-dimensional data sets were collected using Delta Vision System, subsequently deconvolved, and projected onto a single plate.

RNA Extraction and cDNA Synthesis

RNA was extracted using Tri-reagent (Sigma-Aldrich) from 1 × 107 cells or 50–100 mg of tissue. RNA was treated with DNAse free (Ambion, Austin, TX, http://www.ambion.com) and cDNA made from 2 μg of RNA using Moloney murine leukemia virus (H-point mutant) reverse transcriptase and oligo(dT) and random hexamers as primers. cDNA was then purified through spin columns (Qiagen Inc.) using ultracentrifugation and eluted in 50 μl of elution buffer. Two-microliter aliquots were analyzed in quantitative polymerase chain reaction (PCR).

Chromatin Immunoprecipitation

One × 108 cells were fixed for 10 minutes in 0.37% formaldehyde while 100 mg of tissue was roughly chopped in and fixed for 15 minutes in 1% formaldehyde. The fixation reaction was quenched with 0.125 M glycine. Tissue was then homogenized in fresh PBS. Nuclei were isolated by incubation on ice for 10 minutes in cell lysis buffer (10 mM Tris-HCl pH 8.0, 10 mM NaCl, 0.2% NP-40, 10 mM sodium butyrate, and proteinase inhibitors) and lysed in nuclei lysis buffer (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 1% SDS, 10 mM sodium butyrate, and proteinase inhibitors) for 10 minutes on ice. The lysate was sonicated to shear the chromatin, cell debris removed by centrifugation, and immunoprecipitation (IP) dilution buffer (20 mM Tris-HCl, pH 8.1, 150 mM NaCl, 2 mM EDTA, 1% Triton X-100, 0.01% SDS, 10 mM sodium butyrate, and proteinase inhibitors) added to bring the ratio of nuclei lysis buffer/dilution buffer to 1:4. Chromatin was precleared for 3 hours at 4°C with 200 μl of protein-G-Sepharose blocked with 5% BSA. Whole-cell extract was divided into aliquots: 270 μl kept as input chromatin and 1.35-ml aliquots incubated with antibodies against proteins of interest. Immune complexes were collected by rotation at 4°C for 3 hours with 100 μl of blocked protein-G-Sepharose. Beads were washed twice with wash buffer 1 (20 mM Tris-HCl pH 8.1, 50 mM NaCl, 2 mM EDTA, 1% Triton X-100, and 0.1% SDS), once with wash buffer 2 (10 mm Tris-HCl pH 8.1, 250 mM LiCl, 1 mM EDTA, 1% NP-40, and 1% deoxycholic acid), and twice with TE (10 mM Tris/HCl, 1 mM EDTA) pH 8.0. Immune complexes were eluted with elution buffer (100 mM NaHCO3, 1% SDS). Cross-links were reversed by incubation at 65°C for 6 hours in the presence of 5 μg of RNase A and 0.3 M NaCl. Ninety micrograms of proteinase K was added, and samples were incubated overnight at 45°C. DNA was purified using phenol/chloroform extraction and ethanol precipitation. Input pellets were resuspended in 200 μl of water, other samples in 100 μl of water. Two-microliter aliquots were analyzed in quantitative PCR using primers designed adjacent to the RE1 region. The amount of DNA pulled down with antibodies to the proteins of interest was compared with that pulled down with the same amount of a nonspecific antibody (rabbit or goat polyclonal HA antibody; Santa Cruz Biotechnology, Inc.) or, in the case of antisera, to the same volume of normal rabbit serum (Santa Cruz Biotechnology, Inc.). The following antibodies were used: goat polyclonal P18 to REST, rabbit polyclonal to Sin3a, rabbit polyclonal to Sin3b, rabbit polyclonal to HDAC1, rabbit polyclonal to HDAC2, and rabbit polyclonal to c-myc (all Santa Cruz Biotechnology, Inc.), rabbit antiserum to CoREST, rabbit antiserum to H4Ac, rabbit polyclonal to H3K9Ac, rabbit antiserum to H3K4diMe, and rabbit antiserum to H3K4triMe (all Upstate Biotechnology, Lake Placid, NY, http://www.upstatebiotech.com), and rabbit polyclonal to H3K9diMe and rabbit polyclonal to H3 (all Abcam).

Quantitative Real-Time PCR

Primers were designed using the primer3 (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www_slow.cgi) and “mfold” (http://www.bioinfo.rpi.edu/ applications/mfold/old/dna) programs to obtain amplicons using the following parameters: melting temperature (Tm) between 58°C and 62°C, devoid of secondary structure at Tm 60°C and amplicon size of 50–150 bp. Chromatin IP (ChIP) primers were derived from sequence within 500 bp of the RE1. Primer sequences are available on request. Real-time PCR was carried out using SYBR Green containing Supermix (Bio-Rad, Hercules, CA, http://www.bio-rad.com) on a Bio-Rad iCycler. All reactions were performed in duplicate. PCR cycle parameters were 95°C for 3 minutes, followed by 45 cycles of 95°C for 30 seconds, 60°C for 30 seconds, and 72°C for 30 seconds. At the end of the program, the temperature was reduced to 50°C and then gradually increased by 1°C for 10 seconds up to 96°C to produce a melt curve. Gene expression was normalized to cyclophilin using the formula 1/2ΔCt, where ΔCt = average threshold cycle gene − average threshold cycle cyclophilin. All expression was also normalized to mitochondrial 16s rRNA to control for constant cyclophilin levels (data not shown). PCR was carried out for 45 cycles, and a lack of any product by this cycle number was taken as evidence that the gene was silent.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Expression of REST and Its Corepressors and Target Genes in MHP36 Cells

First, we established that Rest and the corepressors Sin3a, Sin3b, and CoREST (Rcor1) are all expressed in MHP36 NSCs (Fig. 1A). Second, we examined expression of eight REST target genes (Fig. 1B), comprising those encoding synaptosomal-associated protein 25 (Snap25), voltage-gated sodium channel alpha subunit Nav1.2 (Scn2a), muscarinic acetylcholine receptor M4 (Chrm4), superior cervical ganglion-10 protein (Scg10) or stathmin2, brain-derived neurotrophic factor (Bdnf), L1 cell adhesion molecule (L1cam), synapsin-1 (Syn1), and synaptotagmin-7 (Syt7). Collectively, these represent the major ontological classifications of neuronal REST target genes, including those involved in cytoskeleton, vesicular trafficking and release, neurotransmission, and survival [8]. Only the Bdnf, Syn1, and Syt7 genes were expressed by MHP36 cells (Fig. 1B). All others were silent.

thumbnail image

Figure Figure 1.. Quantitative reverse transcription-polymerase chain reaction analysis of gene expression in MHP36 neural stem cells. (A): Expression levels of RE1 silencing transcription factor (REST) and the corepressors Sin3a, Sin3b, and CoREST. (B): Expression levels of neuronal REST target genes. All expression is shown normalized to cyclophilin × 10−5. Error bars represent the standard error of the mean from a minimum of three independent RNA samples.

Download figure to PowerPoint

Characterization of the Corepressor Complexes Recruited by REST in MHP36 Cells

Having established expression of REST, its corepressors, and target genes in MHP36 cells, we next examined the occupancy by REST, CoREST, Sin3a, Sin3b, histone deacetylase 1 (Hdac1), and histone deacetylase 2 (Hdac2) at the RE1 sites of the endogenous target genes using ChIP. As can be seen in Figure 2A, REST is present at all the RE1 sites in MHP36 cells. Furthermore, REST is bound, irrespective of the expression level of the associated gene (i.e., REST is present at both transcribed and silent genes in MHP36 cells).

thumbnail image

Figure Figure 2.. Chromatin immunoprecipitation (ChIP) analysis showing REST (A), CoREST (B), Sin3a (C), Sin3b (D), and Hdac1 and Hdac2 (E) enrichments at the RE1 elements of neuronal genes. All enrichments are shown relative to a sham antibody immunoprecipitation. Error bars represent the standard error of the mean from a minimum of three independent ChIPs. Abbreviation: REST, RE1 silencing transcription factor.

Download figure to PowerPoint

REST interacts with numerous cofactors principally via recruitment of two distinct corepressor platforms containing Sin3a/Sin3b and/or CoREST. Because cofactor recruitment can have a profound effect on transcriptional activity [28, 29], we used ChIP to look for the presence of these corepressors at its target genes in MHP36 cells. Specificities of the antibodies against the corepressors were confirmed by Western blot analysis (supplemental online Fig. 5). CoREST and Sin3b are enriched at seven of the eight RE1s, but Sin3a could not be detected at any RE1 (Figs. 2 and 3). Both Hdac1 and Hdac2 could be detected at most, but not all, loci. Notably, Hdac1 and Hdac2 appear to be absent from the Scn2a and L1cam genes (see below).

thumbnail image

Figure Figure 3.. REST complexes at RE1 sites showing an overview of the results in Figure 2, showing the different REST complexes found at different RE1 elements in MHP36 cells. Asterisk represents those genes with detectable transcript levels (Fig. 1). Abbreviation: REST, RE1 silencing transcription factor.

Download figure to PowerPoint

REST and all three corepressors are found in both the nucleus and the cytoplasm of MHP36 cells, indicating that the lack of recruitment of Sin3a is not due to lack of nuclear expression (Fig. 4). Little is known of the functional differences between Sin3a and Sin3b, but both isoforms have been shown to have separate physiological functions [30]. Further work is required to see whether the recruitment of the different isoforms by REST has any effect on REST function.

thumbnail image

Figure Figure 4.. REST and corepressor colocalization in MHP36 cells. REST staining is in red, and staining of corepressors is in green. DAPI staining of the nuclei is shown in blue. Top panel is CoREST, middle panel is Sin3b, and bottom panel is Sin3a. REST and all three corepressors exhibit predominant nuclear staining with some cytoplasmic staining. Scale bar = 20 μm. Abbreviations: DAPI, 4,6-diamidino-2-phenylindole; REST, RE1 silencing transcription factor.

Download figure to PowerPoint

REST Acts Both as a Silencer and a Repressor in NSCs

When REST was first identified, it was believed to act as a silencer of neuronal genes in non-neuronal cells and neuronal progenitors based largely upon the reciprocal expression of REST outside of the adult nervous system and its target genes within the nervous system. However, there is abundant evidence that shows that REST represses, but does not silence, gene expression in non-neural and neural cells [8, 16, 18].

To investigate the function of REST in NSCs, we examined changes in gene-expression levels consequent to infection with an adenovirus construct comprising the myc-tagged DNA-binding domain of REST (Fig. 5A), which effectively acts as a dominant-negative form of REST (DN:REST) [8, 16, 18, 27, 31]. Infection with DN:REST led to a decrease in endogenous REST-binding at the RE1 element and replacement with DN:REST (Fig. 6F, 6G). The enrichments of corepressors Sin3b, CoREST, and Hdac1/Hdac2 were also decreased or abolished, demonstrating that their presence at the RE1 sites is due to their recruitment by endogenous REST. Using primers that detect both endogenous and DN:REST, an approximately 17-fold increase in REST transcripts is seen, indicating that DN:REST is present in an excess (inset, Fig. 5A). All gene-expression levels are shown relative to the housekeeping gene cyclophilin (the same results are obtained if expression levels are normalized to 16S rRNA; data not shown).

thumbnail image

Figure Figure 5.. Regulation of REST target gene expression levels in MHP36 cells. (A): Response to infection with adenovirus (control bars, white) or adenovirus overexpressing the DN:REST (black bars). (B): Response to TSA treatment (black bars) or control cells (white bars). Inset of (A): A 17-fold increase is found in REST transcripts in DN:REST-expressing cells. Inset of (B): No change in REST expression is observed with TSA. All expression levels are shown normalized to cyclophilin × 10−5. The presence/absence of Hdac1/2 as demonstrated by chromatin immunoprecipitation (Fig. 2) is indicated by +/−, respectively. Error bars represent the standard error of the mean from a minimum of three independent RNA samples (∗, .05 <p < .1; ∗∗, p < .05 using a paired t test). Abbreviations: DN, dominant-negative; REST, RE1 silencing transcription factor; TSA, trichostatin A.

Download figure to PowerPoint

thumbnail image

Figure Figure 6.. Chromatin immunoprecipitation (ChIP) analysis of the histone modifications around the RE1 elements in response to overexpression of the dominant-negative form of RE1 silencing transcription factor (DN:REST) in MHP36 cells. White bars represent control cells (infected with adenovirus), and black bars represent cells infected with adenovirus DN:REST. (A) H4Ac, (B) H3K9Ac, (C) H3K9diMe, (D) H3K4diMe (E) H3K4triMe, (F) REST, (G) myc, (H) Sin3b, (I) CoREST, and (J) Hdac1. Error bars represent the standard error of the mean from a minimum of three independent ChIPs (∗, .05< p < .1; ∗∗, p < .05 using an unpaired t test).

Download figure to PowerPoint

It can be seen that Snap25, Chrm4, Scg10, and L1cam are all silent genes in cells infected with empty adenovirus and in noninfected control cells, but when DN:REST is overexpressed, these genes are induced, implying that endogenous REST is silencing these genes. Syn1 and Syt7 are expressed at low levels in cells infected with empty adenovirus and noninfected cells, and upon overexpression of DN:REST, their expression levels are increased sevenfold and fourfold, respectively, indicating that endogenous REST is repressing these genes (Fig. 5A). Therefore, REST has a dual role in MHP36 NSCs and is capable of acting as both a silencer and a repressor of neuronal genes.

In contrast, Scn2a and Bdnf both have REST and attendant corepressors enriched at their RE1s (Figs. 2 and 3), but their expression levels are not altered upon overexpression of DN:REST (Fig. 5A). A similar situation for Scn2a has also been reported in cortical progenitors [22]. Therefore, the presence of a REST complex at an RE1 site does not necessarily imply that it is regulating the expression of the gene. These results underline the diversity of actions driven by REST recruitment at different loci within NSCs.

Histone Deacetylase Is Required for Repression/Silencing of Some, but Not All, REST Target Genes in NSCs

Our ChIP data showed that Hdac1 and Hdac2 are present at some, but not all, RE1 elements in the MHP36 cells (Figs. 2 and 3). To test the functional requirement for histone deacetylase activity in mediating silencing and repression, changes in gene-expression levels were measured in MHP36 cells after treatment with the histone deacetylase inhibitor trichostatin A (TSA) for 24 hours (Fig. 5B). Expressions of REST (inset graph, Fig. 5B) and corepressors (supplemental online Fig. 1) were unaffected by TSA. Importantly, their presence/absence at RE1s was also unaffected by TSA treatment (supplemental online Fig. 2). Several studies have reported TSA-sensitive [17, 32, 33] and -insensitive [16] REST repression, but few have examined Hdac1/2 recruitment. It is important to bear in mind that sensitivity of any given locus to TSA inhibition of histone deacetylase is, by necessity, hostage to the total sum of histone acetylase and deacetylase activities recruited to that promoter [34].

Snap25, Chrm4, and Scg10 are all silent genes that are activated by DN:REST and have Hdac1 and Hdac2 present at their RE1s. Treatment with TSA activates these genes, indicating that they are silenced by REST through the actions of histone deacetylases. Interestingly, a more than 100-fold greater level of induction is seen at Scg10 in TSA-treated cells than in DN:REST-expressing cells. Presumably, this reflects the widespread nature of TSA inhibition that may also affect other histone deacetylases that are recruited independently of REST. L1cam is also induced by DN:REST, but in this case neither Hdac1 nor Hdac2 is detected around the RE1 site and TSA treatment does not lead to induction of the gene, consistent with the notion that REST silences L1cam in a histone deacetylase-independent manner. Scn2a is also silent in these cells, and like L1cam, no Hdac1 or Hdac2 is detectable at the RE1. However, in contrast to L1cam, DN:REST had no effect on Scn2a expression, and in keeping with this, TSA also has no effect, indicating that REST is present at the silent Scn2a gene but is not required to maintain its silence.

Unlike the Snap25, Chrm4, Scn2a, Scg10, and L1cam genes, Bdnf, Syn1, and Syt7 are all expressed, albeit at low levels, in MHP36 cells and all recruit Hdac1 and Hdac2 to their RE1s. In the case of Bdnf, DN:REST has no effect on gene-expression levels, and even though Hdac1 and Hdac2 are present, TSA was still without effect on Bdnf expression, indicating that REST does not necessarily repress gene expression even when it is present together with a corepressor platform (although not necessarily all the possible corepressors). In contrast, expressions of Syn1 and Syt7 are increased sevenfold and fourfold, respectively, in the presence of DN:REST, and TSA also leads to an increase in their expression (12- and 6-fold, respectively), supporting the idea that REST is acting as a histone deacetylase-dependent repressor of Syn1 and Syt7 expression. As was observed for Scg10, TSA produces an almost twofold greater increase than DN:REST on expression of Syn1, again suggesting that REST-independent histone deacetylases are also regulating this gene. Therefore, REST acts to silence and repress genes in NSCs in both a histone deacetylase-dependent and -independent manner.

Changes in REST Occupancy Lead to Epigenetic Changes in Histones Around RE1 Elements

To try to gain a deeper understanding of the mechanisms of REST silencing and repression in NSCs, we looked at histone modifications around the RE1s in response to changing REST occupancy using DN:REST. We chose to focus on acetylation of H4 (H4Ac), acetylation of H3K9 (H3K9Ac), and dimethylation and trimethylation of H3K4 (H3K4diMe and H3K4triMe) as epigenetic marks associated with open or active chromatin [35, 36] and on dimethylation of H3K9 (H3K9diMe) as a mark associated with heterochromatin and silenced/repressed euchromatin [37, 38].

Histone acetylation has long been associated with euchromatin [39, 40], and more recent whole genome and whole chromosome studies have shown that H3 and H4 hyperacetylations are correlated with actively transcribed loci [34, 41]. More specifically, hyperacetylation of H3K9 localizes to promoters and other regulatory elements but not to the entire transcriptional unit [42]. Both dimethylation and trimethylation of H3K4 have been linked with actively transcribed or transcriptionally poised genes [34, 36, 43]. This comapping of H3K4diMe and H3K4triMe is also evident at the RE1 sites where a fairly constant ratio of di- to trimethylation is evident, irrespective of absolute levels (Fig. 6D, 6E). Schneider et al. [36] also report that high levels of H3K4 methylation are also found at inactive genes within the β-globin locus, indicating that H3K4 methylation may maintain or mark chromatin that is “poised” but not actively transcribed. It should be pointed out that changes in H3K4triMe should be evident even if those changes are restricted to the promoter, at least in the cases of the Snap25, Chrm4, Bdnf, and Syn1 RE1s, all of which lie within 1 kb of their promoters (supplemental online Fig. 4) and therefore within the resolution of ChIP.

Figure 6A–6E shows that although the amount of neither H3K4diMe nor H3K4triMe changes in the presence of DN:REST, there is a clear trend showing a loss of H3K9diMe and a gain of H3K9Ac, as well as a gain of H4Ac, at the majority of RE1s. This is not due to a change in the amount of histones in the ChIP fragments; H3 enrichments remained unchanged in both DN:REST and control cells (supplemental online Fig. 3). Although CoREST has recently been reported to interact with the demethylase LSD1 [44, 45], the fact that no increase in H3K4diMe is seen in DN:REST-infected cells suggests that this mechanism is not used at these genes in MHP36 cells (however, a recent report suggests that LSD1 can also demethylate H3K9diMe [46]). This apparent redundancy of REST in conferring H3K4 methylation is consonant with a recent cross-species survey of histone methylation sites across chromosomes 21 and 22, the results of which imply that the bulk of H3K4 methylation is not conferred by proximal DNA-binding proteins [47].

Inhibition of REST Function in MHP36 Cells Results in Expression of β-Tubulin III but Is Insufficient to Drive Neuronal Differentiation

Having shown that REST can repress or silence genes in MHP36 cells, and in light of a recent study demonstrating a role for REST in adult NSC differentiation [14], we next asked whether attenuation of REST function was sufficient to cause neuronal differentiation. We therefore transfected MHP36 cells with DN:REST and looked for expression of pan-neuronal markers β-tubulin III, MAP2, and Tau over an 8- to 9-day time period. Expressions of nestin and GFAP were also monitored as markers of neural progenitors and glial cells, respectively. In clear contrast to cells transfected with empty plasmid, β-tubulin III-positive cells became visible by 2 days after DN:REST transfection, and this remains 9 days after transfection (Fig. 7). Because Tubb3 contains an RE1 site, induction of this gene could be a direct consequence of loss of REST at its regulatory site. Indeed, no mature neurons were generated by attenuation of REST function, as judged by staining with antibodies against MAP2, Tau, and NeuN; furthermore, we could not detect clear changes in nestin expression (data now shown). All cells were GFAP-negative over the time period. Interestingly, the effect of DN:REST expression on TuJ1 expression could not be recapitulated by TSA treatment, indicating that induction of Tubb3 by REST was independent of HDAC activity (data not shown). Therefore, these results indicate that the attenuation of REST function induces only a part of the neuronal transcriptional program but is insufficient to drive neuronal differentiation of MHP36 NSCs.

thumbnail image

Figure Figure 7.. Effect of the dominant-negative form of RE1 silencing transcription factor (DN:REST) on neural gene expression. MHP36 cell transfected with DN:REST (GFP-positive, green) expressed β-III tubulin (red) 2 days after transfection, in clear contrast to a control cell (GFP-positive, green). Expression of β-III tubulin (red) remained evident 8 days after DN:REST transfection; however, no change in nestin expression (green) was observed. Abbreviation: GFP, green fluorescent protein.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

The perception of REST has changed dramatically over recent years from that of a silencer of neuronal gene expression in non-neural tissue to include that of a regulator of NSC differentiation [14, 22], a mediator of cell death in neurodegeneration [48] and latterly, a tumor suppressor [49]. There are more than 1,300 RE1 sites in the mouse genome that lie proximally to approximately 1,000 known or predicted genes [50]. Although we anticipate that the specific cohort of REST-regulated genes will vary across different cell types and developmental stages, we have focused in this study upon REST occupancy, corepressor recruitment, gene expression, and epigenetic signature at its target genes in embryonic hippocampal stem cells. We show further that attenuation of REST function is sufficient to activate several RE1-containing genes but is insufficient to drive full neuronal differentiation.

REST Interaction with Target Genes in Embryonic Hippocampal NSCs

Recently, REST has been strongly implicated in neuronal differentiation. A recent study [22] reported that REST protein was post-translationally downregulated in neural progenitors derived from embryoid bodies and in cortical progenitors although its presence could still be detected within chromatin. In contrast, it has also been demonstrated that REST is present and interacts with its target genes in both adult NSCs and differentiated neurons [14]. In the latter study, REST appears to be transformed into a transcriptional activator via interaction with a double-stranded RE1 small modulatory dsRNA; furthermore, this interaction appears to be obligate for differentiation of NSC into neurons. Because both REST and its corepressors are expressed in MHP36 cells, this further highlights heterogeneity of NSCs, which in turn may reflect their different positional or developmental origins.

We show that REST is present at RE1 sites and represses or silences genes in MHP36 cells. Interestingly, derepressed transcript levels are still very low compared with levels in adult neurons (data not shown), perhaps reflecting the presence of further repressors and/or the absence of appropriate activators. Such low-level gene expression has been observed in several multipotent cells, including hematopoietic stem cells [51, 52], and may reflect a relatively open chromatin configuration prior to differentiation.

Here, we have shown that in MHP36 cells REST is able to recruit distinct corepressors to different target genes. At present, the mechanisms behind this are unknown, but ultimately such selective recruitment of corepressors by a single transcription factor may be responsible for the differential effects on gene regulation that are observed.

At the majority of RE1 sites, histone deacetylase activity appears to be responsible for maintaining a silent or repressed state. Notably, TSA tended to induce higher levels of expression than DN:REST, consistent with the view that other histone deacetylase-dependent mechanisms are operative. TSA was without effect on expression of Scn2a1 and L1cam, a result consistent with the absence of Hdac1 and Hdac2 at these sites. Sin3b and CoREST are both recruited to these RE1 sites, indicating that the presence of the two major corepressor platforms is not always sufficient to confer histone deacetylase regulation and that a histone deacetylase-independent repression pathway is also operative. Taken together, these data indicate that REST occupies RE1 sites in NSCs where it establishes and/or maintains very low levels of transcription prior to recruitment or activation of appropriate activators.

REST Confers Repressive Epigenetic Marks onto the Chromatin of Target Genes

So how does REST recruitment affect the epigenetic signature at RE1 sites? REST clearly maintains a low degree of H4 and H3K9 acetylation at the majority of RE1 sites examined and thus appears to be responsible for maintaining the RE1 in a hypoacetylated state. It is intriguing that REST recruitment at the Scn2a and Bdnf RE1s causes a decrease in acetylation, consistent with a transition from “repressed” to “active” chromatin, even though attenuation of REST occupancy does not lead to any change in gene expression, possibly indicating that no transcriptional activators are present.

In a reciprocal manner to H3K9Ac, levels of dimethylated H3K9 are increased in the presence of REST at all RE1s examined, consistent with the reported recruitment of the G9a histone methyltransferase by REST [53]. Thus, REST effects the balance of acetylation and methylation of H3K9 at RE1 sites toward a more repressive signature, but this is not necessarily sufficient to lead to a change in transcription. The presence of methylated H3K4 at RE1 sites of genes that are either silent or transcribed very weakly implies that these genes could be in open chromatin, a situation that may well apply to many neuronal genes in NSCs [54, 55]. However, unlike acetylation and methylation of H3K9, the K4 methylation signature does not appear to be controlled by REST, implying that the lysine-specific demethylase, LSD1 [44, 45], is not operative at these sites.

These data raise the prospect that, in NSCs, neuronal genes are marked by methylation of H3K4 in addition to methylation of H3K9 and low levels of H3 and H4 acetylation and that REST is recruited to maintain local transcriptional repression, in part by controlling local H3K9 methylation/acetylation. Such coexpression of repressive and active epigenetic marks has also been observed at silent loci in pluripotent embryonic stem cells [56].

REST as a Regulator of Neuronal Differentiation

REST can induce neuronal differentiation in adult NSCs [14], and REST mutants containing a heterologous activation domain can also induce neuronal gene expression, even in myoblasts [57, 58]. Clearly, inhibition of REST function in MHP36 cells leads to elevated expression levels of several neuronal target genes, but our experiments showed that transient inhibition of REST function was not sufficient to cause the cells to differentiate into mature neurons. Cells did express TuJ1 immunoreactivity, but given that βIII-tubulin is a direct REST target gene, this could not be taken as a readout of neuronal differentiation. Indeed, we could detect no induction of other pan-neuronal markers such as MAP2, NeuN, and doublecortin. These observations are supported by studies that have used a dominant-positive REST-VP16 construct [57] in which REST-VP16 was able to activate direct REST target genes and produce TuJ1-positive cells but was unable to produce mature neurons without the addition of retinoic acid. Perhaps it is not surprising that mature neurons did not develop, given that some target genes such as Scn2a and Bdnf were not activated in the presence of DN:REST, even though their chromatin structure was altered. Therefore, although REST may play a role in the early stages of differentiation, it would appear that the presence of appropriate activators is crucial to produce functional neurons.

Summary

The picture that emerges is of REST as a player in the regulation of neuronal gene expression and neuronal differentiation in NSCs and in differentiated neurons. The exact role of REST varies according to the source and regional identity of the cells and/or their developmental stage. Here, we demonstrate that differences exist in the corepressor platforms recruited in MHP36 embryonic hippocampal stem cells at different loci. Furthermore, we have shown that REST may precipitate changes in the epigenetic signature without necessarily effecting gene expression. Weaving together changes in occupancy, epigenetic signatures, and gene expression will ultimately provide mechanistic insight into the manifold transcriptional programs underwritten by REST in different stem cell backgrounds.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

This work was supported by the Wellcome Trust and the Biotechnology and Biological Sciences Research Council.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
Supp_Fig_1.pdf72KSupplemental Figure 1
Supp_Fig_2.pdf528KSupplemental Figure 2
Supp_Fig_3.pdf530KSupplemental Figure 3
Supp_Fig_4.pdf264KSupplemental Figure 4
Supp_Fig_5.pdf20KSupplemental Figure 5
Buckley_Supp_Legends.pdf35KSupplemental Legends

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.