Note that we use the term gene silencer to describe a factor that completely blocks the synthesis of any target gene transcript to distinguish it from a repressor that results in lower numbers of detectable transcripts but nevertheless still some.
Address correspondence and reprint requests to Alexander Bruce, Transcription Regulation Group, The Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridgeshire CB10 1SA, UK. E-mail: email@example.com
Release of distinct cellular cargoes in response to specific stimuli is a process fundamental to all higher eukaryotes and controlled by the regulated secretory pathway (RSP). However, the mechanism by which genes involved in the RSP are selectively expressed, leading to the establishment and appropriate functioning of regulated secretion remaining largely unknown. Using the rat pheochromocytoma cell line PC12, we provide evidence that, by controlling expression of many genes involved in the RSP, the transcriptional repressor REST can regulate this pathway and hence the neurosecretory phenotype. Introduction of REST transgenes into PC12 cells leads to the repression of many genes, the products of which are involved in regulated secretion. Moreover, chromatin immunoprecipitation assays show that many of the repressed genes recruit the recombinant REST protein to RE1 sites within their promoters and abrogation of REST function leads to reactivation of these transcripts. In addition to the observed transcriptional effects, PC12 cells expressing REST have fewer secretory granules and a reduction in the ability to store and release noradrenaline. Furthermore, an important trigger for synaptic release, influx of calcium through voltage-operated calcium channels, is compromised. This is the first demonstration of a transcription factor that directly controls expression of many major components of the RSP and provides further insight into the function of REST.
repressor element 1-silencing transcription factor/neuronal restrictive silencing factor
regulated secretory pathway
vesicular mono-amine transporter
voltage-operated calcium channel
The repressor element 1-silencing transcription factor (REST, also known as the neuron restrictive silencing factor or NRSF) was first identified as a protein that binds to a 21-bp DNA sequence element [known as repressor element 1 (RE1), also known as neuron restrictive silencing element (NRSE)] resulting in transcriptional repression of the neural-specificvoltage-gated type II sodium channel (nav1.2) and superior cervical ganglion 10 genes (Chong et al. 1995; Schoenherr and Anderson 1995). Subsequently, functional RE1s have been identified in about 30 genes [e.g. m4muscarinic receptor (Wood et al. 1996) and brain derived neurotrophic factor (Timmusk et al. 1999)]. We recently described a bioinformatics sequence analysis of the human, mouse and Fugu rubripes genomes and identified 1892, 1894 and 554 candidate RE1s, respectively, with proximity to genes with roles as diverse as transcriptional regulation to metabolism and various aspects of neuronal function (Bruce et al. 2004; database available at: http://www.bioinformatics.leeds.ac.uk/cgi-bin/RE1db/nrse.cgi).
The originally proposed role for REST was that of a factor responsible for restricting neuronal gene expression to the nervous system by silencing expression of these genes in non-neuronal cells. Given that the majority of known REST target genes had neuronally restricted patterns of expression, and REST was originally thought not to be expressed within neurones (Chong et al. 1995; Schoenherr and Anderson 1995), this was a reasonable hypothesis. However, subsequent studies suggest other additional and more complex roles for REST and suggest that it may act more like a classical transcriptional repressor rather than a gene silencer in both neural and non-neural cells.
For example, REST expression has been detected within hippocampal neurones and expression levels have been shown to rise in response to epileptic and ischaemic insults (Palm et al. 1998; Calderone et al. 2003), supporting an important repressor role for REST within neurones. Additionally, REST also plays a pivotal role during neuronal degeneration in Huntington's disease (Zuccato et al. 2003). Whilst outside of the nervous system, REST is critically important for directing the appropriate expression of the fetal cardiovascular gene programme in heart and vascular smooth muscle (Kuwahara et al. 2003; Cheong et al. 2005) and acting in a classical tumour suppressor gene (Westbrook et al. 2005). Currently, a complete understanding of the true roles played by REST is lacking.
The regulated secretory pathway (RSP) is a fundamental process essential to all higher eukaryotic organisms. The processes and mechanisms by which discrete cargoes are released in a tightly controlled manner in response to defined cellular cues has been the topic of intense research. Indeed, there is a large knowledge about the mechanisms underpinning the functioning of the RSP as exemplified by output of many laboratories relating to the synaptic vesicle cycle (reviewed in Sudhof 2004). However, a knowledge of the molecular processes governing the attainment of neurosecretory competence and how the RSP is established and dynamically regulated have been crucially lacking. It is against this backdrop that we have investigated the role employed by REST in regulating gene expression and consequent function within the RSP.
Materials and methods
PC12, PC12 REST and PC12 HZ4 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 5% horse serum, 2 mm l-glutamine and 5 units/mL of penicillin/streptomycin and JTC19 cells in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mm l-glutamine and 5 units/mL of penicillin/streptomycin in 5% CO2 at 37°C.
Nuclear protein extracts were prepared from 80% confluent 10-cm diameter tissue culture plates of PC12, PC12 REST, PC12 HZ4 and JTC19 cells using previously described techniques (Wood et al. 2003). Nuclear protein extracts (10 μg per well) were diluted in a volume of 30 μL containing 4 × sample buffer and 1 × NuPAGE-sample reducing buffer (Invitrogen, Carlsbad, CA, USA). The samples were then loaded onto pre-cast NuPAGE 4–12% Bis-Tris-Glycine acrylamide gels (Invitrogen), without prior boiling (REST protein is heat labile) and resolved by electrophoresis (200 V constant) in 1 × [3-(N-Morpholino)propanesulfonic acid] (MOPS) running buffer (Invitrogen) for 1.5 h alongside SeeBlue Plus2 molecular weight standards (Invitrogen). The resolved proteins were transferred onto Immobilon-P polyvinylidene difluoride membranes (Sigma, St Louis, MO, USA) using the Xcell II blot module from Invitrogen and 1 × NuPAGE transfer buffer (Invitrogen) according to the manufacturers' protocols. Blotted membranes were then blocked by incubation in blocking buffer [1 × Tris-buffered-saline (TBS), 0.1% Tween20 and 5% non-fat milk powder] for 1 h at room temperature (21–23°C) on an orbital shaker. Blocked membranes were then incubated with either anti-REST (Upstate Biotechnology, Lake Placid, NY, USA; 07–579, 1 : 2000), anti-myc (Sigma; C3956, 1 : 1000) or anti-CstF64 (Santa-Cruz Biotechnology, Santa Cruz, CA, USA; sc-16473, 1 : 500) antibodies diluted in 15 mL of blocking buffer on an orbital shaker overnight at 4°C. The membranes were next washed (at room temperature, 21–23°C) four times in 1 × TBS plus 0.1% Tween20 (TBS-T) for 15 min each and then incubated in blocking buffer containing the appropriate HRP-conjugated secondary antibodies [goat anti-rabbit-HRP (Santa Cruz Biotechnology; sc-2004, 1 : 10 000) for anti-REST/anti-myc antibodies or donkey anti-goat horseraddish peroxidase (Santa Cruz Biotechnology; sc-2056, 1 : 10 000) for anti-CstF64 antibody] for 1 h at room temperature (21-23°C) with shaking. After four 15-min washes in TBS-T, the membranes were rinsed in HPLC water and the immune complexes detected using the SuperSignal Chemiluminescent kit (Pierce, Rockville, MD, USA) as described by manufacturers' protocols.
RT-PCR and quantitative RT-PCR
Conducted as previously described (Wood et al. 2003). Deoxyoligonucleotide primer sequences are available as supplementary data.
Affymetrix transcript profiling microarrays
Extracted total RNA (as above) was used to prepare cRNA probes that were hybridized on Affymetrix rat genome U34A microarrays (according to manufacturers' protocols). A GeneArray scanner (Hewlett-Packard, Palo Alto, CA, USA) was used for data collection and analysis used Microarray Suite-5 program (Affymetrix, Santa Clara, CA, USA) to identify genes with consistent expression level changes greater than two-fold between cell types in each of two biologically independent replicates.
Chromatin immunoprecipitation (ChIP) assay
Antibodies raised to the N-terminus of REST (H290 Santa-Cruz Biotechnolgoy), myc epitope tag (Sigma), C-terminus of histone H3 (Abcam, Cambridge, UK) and a normal rabbit IgG control (Sigma) were used for ChIP analysis of PC12, PC12 REST and PC12 HZ4 cells using a previously described protocol (Bruce et al. 2004). DNA fragments were used as template (2 μL) in real-time PCR reactions using Bio-Rad iQ cycler and iQ PCR mix (Bio-Rad Laboratories, Hercules, CA, USA) containing 300 nm deoxyoligonucletide primers designed to amplify genomic DNA sequences adjacent to putative RE1 sites in the rat genome (see supplementary data for sequences). Fold enrichments of specific antibodies over normal rabbit IgG control were determined by reference to standard curve generated using rat genomic DNA.
Noradrenaline uptake and stimulated release experiments
PC12, PC12 HZ4/REST cells grown to 80% confluency in 10-cm diameter plates were washed with phosphate-buffered saline (PBS) and scraped into 3 mL of incubation buffer (final concentrations in mM: NaCl 138, KCl 4, CaCl2 1.3, MgCl2 1.2, NaH2PO4 1.2, glucose 10, ascorbic acid 1, HEPES 10; monoamine oxidase inhibitor pargyline 0.01, pH 7.4). Cells were sedimented at 357 g for 2 min, re-suspended in 500 μL of fresh incubation buffer containing noradrenaline ([3H]NA; 100 nm, 36 Ci/mmol; Amersham Pharmacia Biotech, Piscataway, NJ, USA) and incubated at 37°C for 60 min to allow uptake of the [3H]NA. In some experiments, plasma membrane (noradrenaline transporter; NAT) or vesicular mono-amine transporter (VMAT) inhibitors desipramine (1 μm) and reserpine (10 μm) were added to estimate their effects on the accumulation of [3H]NA. Although, in reserpine experiments, loading time was 15 min. Cells for release experiments were pelleted, washed with fresh pro-release buffer (incubation buffer with 1 μm desipramine to block reverse transport via NAT), re-suspended in the same buffer, and 100 μL aliquots of cell suspension were used. For potassium depolarization-evoked release of noradrenaline, four aliquots of cell suspensions were pre-incubated for 5 min at 37°C and then, to two of the aliquots, 100 μL of pre-warmed pro-release buffer was added (negative, passive release control), and to the remaining two aliquots 100 μL of pre-warmed release buffer (containing 142 mm KCl at the expense of NaCl to maintain isoosmolarity) was added. Incubation at 37°C continued for an additional 5 min, after which time 1 mL of ice-cold stop buffer (as pre-release buffer but containing no calcium and 5 mm EDTA) was added. After sedimentation, 500 μL aliquots of supernatant were counted by liquid scintillation to assay the potassium evoked [3H]NA release. The cell pellets were dissolved in 100 μL of 1 m NaOH and 50 μL aliquots counted in duplicate to enable the fractional release to be calculated. Release experiments were repeated whereby cells were incubated in a modified pre-release medium containing no calcium and 0.1 mm EDTA in the presence of 1 μm ionomycin (plasma membrane calcium ionophore) for 2 min at 37°C and then incubated for additional 5 min with or without addition of CaCl2 (1.3 mm final concentration) to initiate the release. Calcium-evoked [3H]NA released was calculated in an analogous way to that for potassium-evoked release.
Cells were washed in 0.1 m phosphate buffer pH 7.4 for 10 min, post-fixed in 0.5% osmium tetroxide for 45 min, and washed again with the same buffer. The cells were then dehydrated through a series of ethanols followed by two 10-min propylene oxide (Fisher Scientific, Hampton, NH, USA) washes, immersed in Durcupan ACM resin (Fluka, Buchs, Switzerland) for 12–20 h, mounted on glass slides, and placed in an oven at 60°C for 48 h to polymerize the resin. After examination by light microscopy, areas with suitable staining for electron microscopy were selected. Suitable areas were cut out and glued to the flat surface of a resin block. Once trimmed, serial ultra-thin sections (70 nm) were cut from the block using a Leica Ultra Cut S ultra-microtome (Leica Microsystems GmBH, Wetzlar, Germany) and collected on Formvar-coated 1-mm slot grids. Sections were stained with lead citrate before viewing on a CM10 transmission electron microscope (Phillips, Amsterdam, the Netherlands). Negatives were digitized using an Agfa Duoscan Scanner (Agfa, Mortsel, Belgium) and processed in Corel Draw 12 to adjust contrast and brightness. The approximate volume of cytoplasm occupied by dense core secretory granules (DCGs) in each of the PC12 cell lines was quantified according to described protocols (Howard et al. 1998).
Intracellular calcium imaging
PC12 and PC12 REST cells were seeded on round glass coverslips and grown to 80% confluency. They were washed in Krebs-HEPES buffer (138 mm NaCl, 4 mm KCl, 1.3 mm CaCl2, 1.2 mm MgCl2, 1.2 mm NaH2PO4, 10 mm Na-HEPES pH 7.4 and 10 mm glucose) and incubated at 37°C for 60 min in 1 mL of buffer supplemented with 10 μm fura-2AM. Cell-coated coverslips were then washed again and installed in a superfusion chamber on an inverted fluorescence microscope (Olympus, Tokyo, Japan). The cells were superfused at room temperature (22–25°C) at a rate of 0.5 mL/min using an application system enabling fast switching of the superfusion media. The cells were stimulated twice by exposures to a 73 mm KCl medium (KCl was increased isoosmotically at the expense of NaCl) and once by 5 mm ATP. Fluorescence measurements were taken at 510 nm after alternate excitation at 340 and 380 nm wavelengths. Data was collected every 1.3 s and processed with Metafluor software (Universal Imaging Corporation, West Chester, PA, USA).
Adenoviral infection of PC12 HZ4 and PC12 REST cells
PC12 HZ4 and PC12 REST cells were infected with previously derived adenoviral vectors harbouring transgenes for either green fluorescent protein (GFP) alone or GFP and dominant negative REST (DN : REST), referred to as Ad and Ad : DN, respectively, according to published protocols (Wood et al. 2003). Total RNA was prepared 24 h post-infection and cDNA prepared for real-time PCR analysis using deoxyoligonucleotide primers directed against RSP-related genes (for primer sequences see supplementary data). Normalized gene transcript levels were determined in each infection condition as described (Wood et al. 2003) and the fold change in expression observed in Ad : DN over Ad infection calculated for either PC12 HZ4 or PC12 REST cells. An inverted fluorescence microscope (Olympus) was used to photograph infected cells expressing the GFP marker gene in culture and compared with similar photographs taken under light microscopy conditions, enabling the success of transgene delivery to be assessed.
RE1s are observed within the regulatory regions of many genes related to regulated secretion
Previously, we described a bioinformatics search of the human, mouse and Fugu genomes for RE1s (RE1db) (Bruce et al. 2004). Interrogation of the RE1db detailed that, from the 1892 RE1s found in the human genome, greater than 50 were associated with genes with ascribed functions in vesicular trafficking and fusion and were conserved between the species [e.g. synaptosomal-associated protein, 25 kDa (snap25), synaptotagmins II, V, VII, X, XIV, N-ethylmaleimide-sensitive factor attachment protein nsf, rabphillin 3A and l-type voltage-gated calcium channelβ2 (cab2; Bruce et al. 2004)]. In addition, four previously described REST target genes have predetermined roles within the RSP [synapsin 1 (Schoch et al. 1996), synaptophysin and synaptotagmin IV (Schoenherr et al. 1996) and dynamin I (Yoo et al. 2001)]. Furthermore, it is known that REST expression is high in non-neuronal cells and low or absent in cell types that rely on active and extensive RSPs [i.e. neuronal or neuroendocrine cells (Chong et al. 1995; Schoenherr and Anderson 1995; Palm et al. 1998)]. Therefore, taking these observations together, we hypothesized that one of the roles employed by REST may be to directly contribute to the establishment and regulation of the neurosecretory phenotype.
Functional recombinant REST protein expression in PC12 cells
Our hypothesis predicted that introduction of REST into neuroendocrine cells would attenuate regulated secretion. To test this, we worked with the rat pheochromocytoma cell line PC12 and two previously derived stable PC12 cell clones, PC12 REST and PC12 HZ4, that each express integrated rest transgenes (Fig. 1; Wood et al. 2003). The PC12 cell line was suited to our study because it is a well-utilized model system of a neuroendocrine cell used to study the RSP (Corradi et al. 1996; Pance et al. 1999; Grundschober et al. 2002) and has also been previously exploited to investigate REST target gene regulation by introduction of recombinant rest genes [given that PC12 cells do not express endogenous rest (Ballas et al. 2001)]. The cell line designated PC12 REST expresses full-length REST protein, whilst PC12 HZ4 expresses an N-terminal myc-epitope-tagged version of the HZ4 fragment of REST (Fig. 1a; Wood et al. 2003). Despite HZ4 constituting a truncated protein (comprising the entire N-terminal repression domain and DNA binding domain but lacking the C-terminal repression domain; Fig. 1a) it has clearly been shown to have robust repressor activity (Schoenherr and Anderson 1995; Roopra et al. 2000; Wood et al. 2003). Furthermore, it remains unclear why the REST protein retains two independent repression domains, and we anticipated that inclusion of the PC12 HZ4 cell line in this study may provide insight into potential functional relationships between the two repression domains. Although we have previously reported a thorough functional characterization of these clones at both the mRNA and protein levels (Wood et al. 2003), it is nonetheless noteworthy that the recombinant rest expression in PC12 HZ4 is 4.5 times greater than that in PC12 REST at the transcript level (Fig. 1b). Western blotting analyses clearly demonstrate an absence of detectable REST protein within regular PC12 cells and confirms recombinant REST protein expression within PC12 REST and PC12 HZ4 cells (Fig. 1c). As the anti-REST antibody we used in this study was raised to an epitope within the C-terminus of REST (lacking in the HZ4 REST protein), we confirmed expression of the HZ4 REST protein within PC12 HZ4 cells using an antibody raised to the N-terminal myc-epitope tag of HZ4 REST. This antibody also detects endogenous myc protein expression within PC12, PC12 REST and PC12 HZ4 cells and showed it to be equivalent across the three cell lines (Fig. 1cii). The level at which the recombinant REST protein is expressed within PC12 REST cells is equivalent to endogenous REST expression within the rat lung fibroblast cell line JTC19 (Fig. 1ci). In conclusion, the recombinant PC12 cell lines express detectable levels of REST protein and, based on the transcript analysis, this expression is higher in PC12 HZ4 than PC12 REST. Unmodified PC12 cells do not display levels of detectable REST protein.
REST mediated transcriptional repression of genes involved in the regulated secretory pathway
Having established a model system, high-density deoxyoligonucleotide microarrays (rat genome U34A; Affymetrix) were utilized to profile the expression of 7000 full-length sequences and 1000 EST clusters in the three cell lines. Given that REST has a very well-characterized role as a repressor (Chong et al. 1995; Schoenherr and Anderson 1995), transcript levels that showed greater than two-fold repression in the recombinant REST-expressing cells were identified (see supplementary data). Interrogation of these genes showed many previously identified and functionally characterized REST target genes. For example, the synaptotagmin IV and nmdz1 (Schoenherr et al. 1996), dynamin I (Yoo et al. 2001), glutamate receptor subunit 2 (GRIA2; Myers et al. 1998), pax4 (Kemp et al. 2003), choline acetyl-transferase (Lonnerberg et al. 1996) and corticotrophin releasing factor (Seth and Majzoub 2001) genes were repressed in either or both the PC12 HZ4 and PC12 REST cells, thus supporting our previous characterizations of these cell lines confirming functional REST expression (Wood et al. 2003). To summarize, a total 596 probes reported greater than two-fold repression in PC12 HZ4 (including 200 EST sequences) and 547 probes were repressed in PC12 REST (including 206 ESTs). However, given that the recombinant PC12 cell lines express REST in a constitutive manner, it was difficult to ascertain whether all the observed transcript level changes were as a result of direct REST regulation or attributable to secondary changes (potentially arising from regulation of other transcription factors by REST). In order to address this concern, we sought to identify whether the down-regulated genes were associated with any RE1 sites. We determined whether the mouse homologues of repressed rat genes were proximal to putative RE1 sequences identified in our previous bioinformatics analyses (Bruce et al. 2004). Using this approach, we identified 135 and 141 genes from the PC12 HZ4 and PC12 REST data sets, respectively, that are linked to putative RE1 sequences (Table 1). Further analysis detailed the presence of genes with specific roles (according to official Gene Ontology annotation) within the RSP (highlighted as bold text in Table 1).
Table 1. RE1db genes repressed in (a) PC12 REST and (b) PC12 HZ4 versus parental PC12 cells
Gene names for which mRNA levels were equal or greater than two-fold repressed in either of the recombinant REST-expressing PC12 cell lines (PC12 REST and PC12 HZ4) when compared with parental (non-REST expressing) PC12 cells in microarray transcript profiling experiments. All genes are found proximal to REST binding sites (RE1s) within a database of human, mouse and Fugu genomes [RE1db (10)]. **Denotes those genes commonly repressed in PC12 REST and PC12 HZ4. Underlined text denotes previously characterized REST target genes and bold type genes with known roles in vesicular trafficking and regulated secretion.
Thirty-one genes known to function in the RSP are repressed in PC12 HZ4. The products of these genes are involved in many steps in the pathway including; granule and vesicle biogenesis, transport and membrane fusion. These include the known REST target gene synaptotagmin IV and other putative REST target genes such as syntaxins 2, 3, 4 and 5, synaptotagmin VII, synaptotagmin-binding zyginl protein, pre-nylated SNARE Ytk6p, nsf, synaptic vesicle protein 2, phosphatidylinositol 3-kinase, phosphatidylinositol 4-kinase, sec7, munc13-2, mint1, doc2, snap25, vesl-2, rabphillin3A and rab3C (reviewed in Sudhof 2004). Membrane fusion is crucial for RSP function and the dataset includes a preponderance of fusion apparatus genes, as well as genes that perform other roles within the RSP. These include genes with functions involved in vesicle budding/membrane recycling [clathrin light chain, assembly protein 180 (snap91) and dynamin 1], vesicular motor proteins (kinesin-related proteins 1 and 1D), initiation of vesicle fusion (cab2, secretogogue receptor 1a and galanin receptor 2) and DCG biosynthesis and function (chromogranin A and neurexin III). Additionally, some RSP-related genes (genes that do not perform overt RSP functions yet are necessary for establishment and appropriate functioning of discrete RSPs) were also down-regulated in PC12 HZ4. For example, neurotransmitter receptors (for acetylcholine, dopamine, glutamate, opiods, purines, serotonin and vasopressin), neurotransmitter transporters (glutamate and monocarboxylate transporters) and neurotransmitter metabolic enzymes (acetyl-cholinesterase T subunit) genes are all repressed. In PC12 REST cells there are 39 RE1 proximal genes that are also repressed in PC12 HZ4 (Table 1), including RSP-related genes (e.g. chromogranin A, kinesin-related protein 4, doc2b, cab2, syntaxin binding protein munc18-2, secretin and secretogogue receptor 1a).
Transcriptionally repressed RSP related genes recruit REST protein to identified RE1 sites
The observation that genes with known functions in the RSP were shown to be repressed in PC12 REST and PC12 HZ4 cells, and that these genes were proximal to putative RE1 sequences, led us to hypothesize that they were direct targets of REST. Accordingly, we conducted ChIP analyses on PC12, PC12 REST and PC12 HZ4 cells using antibodies directed against REST or the myc epitope. The promoter regions harbouring RE1s from three previously characterized REST target genes that were repressed in the microarray study [nmdz1 (Schoenherr et al. 1996), gria2 (Myers et al. 1998) and synaptotagmin IV/syt4 (Schoenherr et al. 1996), i.e. positive controls] and six putative REST repressed target genes [chromogranin A/chga, snap25, synaptotagmin VII/syt7, rabphillin 3A/rp3a, neuronal munc18-1 interacting protein/mint1/apba1 and secretin/secr; all genes involved in the membrane fusion apparatus (Sudhof 2004)] were analysed for presence of REST protein (Fig. 2). The ChIP analysis shows that each of the three positive control gene promoters are highly enriched for REST binding in PC12 REST and PC12 HZ4 cells (using either an anti-REST antibody raised to the N-terminus of REST or the anti-myc antibody) when compared with PC12 cells alone. Interrogation of the remaining RSP-related genes detail REST occupancy at four of the six genes in PC12 REST cells and five of the six genes in PC12 HZ4 cells. The fact that no enrichment was seen at the secr gene locus in any cell line demonstrates that the observed enrichments in PC12 REST and PC12 HZ4 cells are not attributable to artefacts arising from increased non-specific interactions in the ChIP assay with these cell lines. No enrichment was observed in any of the genomic loci examined for the regular PC12 cell line when using anti-REST antibodies. This is consistent with the lack of detectable REST expression within these cells (Fig. 1). However, to exclude the possibility that PC12 chromatin was refractory to ChIP with any antibodies, a control precipitation using an anti-histone H3 antibody was undertaken and shown to efficiently precipitate chromatin surrounding the snap25 RE1 (Fig. 2, grey bar). Thus, the observed promoter occupancy by REST in PC12 HZ4 and PC12 REST cells confirms the majority of transcriptionally repressed RSP genes as novel and direct targets of REST, although some genes may be repressed by secondary mechanisms (e.g. secr).
REST expression inhibits the regulated secretion of noradrenaline
To test if REST expression was sufficient to attenuate the RSP, we assayed the uptake and regulated release of tritiated [3H]NA. Apparent [3H]NA uptake was found to be less efficient in the PC12 HZ4 and PC12 REST cells compared with PC12 cells (Fig. 3a, filled bars). This reduction was not as a result of reduced expression of nat, as shown by RT-PCR analysis (Fig. 3b) and this gene was not found proximal to any putative RE1 sites in our previous bioinformatics analysis (Bruce et al. 2004). Furthermore, in the presence of a NAT inhibitor (desipramine), apparent loading in all three cell lines was severely attenuated, indicating that functional NAT is present in all cell lines (Fig. 3a, grey bars). The measured levels of [3H]NA, passively released under non-stimulatory conditions (normalized to cell content) was greater in the PC12 HZ4 and PC12 REST cells (particularly PC12 HZ4) than the PC12 cells (Fig. 3c). This constitutive release can, in part, explain why PC12 HZ4 and PC12 REST appear to take up less [3H]NA, as these cells are less able to accumulate the [3H]NA. Stimulated release measurements described below have been expressed as a percentage of loaded [3H]NA. The stimulated release of [3H]NA (initiated by potassium-induced plasma membrane depolarization) is shown in Fig. 4. As can be seen, PC12 cells release 10% of their stored [3H]NA, whilst PC12 HZ4 and PC12 REST cells do not release any stored [3H]NA. This complete block of [3H]NA release in the PC12 HZ4 and PC12 REST cells is an indicator of defective RSP function and thus neurosecretory competence.
PC12 HZ4 cells have attenuated noradrenaline storage capacity and lack DCGs; PC12 REST cells have fewer DCGs than PC12 cells
The passive release of [3H]NA from PC12 HZ4 cells was greater than that from PC12 REST and PC12 cells (Fig. 3c). As these experiments were performed in the presence of ascorbic acid (to prevent uncatalysed oxidative degradation of [3H]NA) and pargyline (monoamine oxidase inhibitor that prevents enzymatic degradation of [3H]NA), it was considered unlikely that this effect was because of the release of [3H]NA metabolites through the plasma membrane. To test if the passive release was as a result of compromised [3H]NA storage capacity in PC12 HZ4 and subsequent release by a constitutive pathway, cell loading experiments were conducted in the presence and absence of the VMAT inhibitor reserpine (Fig. 5a). As shown, the level of [3H]NA accumulated in the PC12 and PC12 REST cell lines was severely reduced in the presence of reserpine. As reserpine is highly selective for VMAT and does not inhibit NAT, this result indicates that the majority of accumulated [3H]NA in these cells (in the absence of reserpine) was efficiently stored within secretory/storage organelles. However, in PC12 HZ4 cells [3H]NA accumulation occurred independently of VMAT and secretory/storage organelles as reserpine had little effect on its uptake. This result suggests that secretory/storage organelles are absent from PC12 HZ4. Therefore, we examined regular PC12, PC12 HZ4 and PC12 REST cell sections by electron microscopy for the presence of DCGs (Figs 5b and c). Normal PC12 cells have plentiful and recognizable electron dense structures, with dimensions consistent with DCGs that occupied approximately 0.6% of the cellular cytoplasmic space. The presence of these structures is consistent with the reserpine experiments that suggested [3H]NA storage organelles were intact. We could find no evidence for DCGs in any PC12 HZ4 micrographs we examined. This was again consistent with the lack of reserpine sensitivity in cell loading experiments. These data demonstrate that PC12 HZ4 cells are completely compromised in secretory organelle biogenesis and neurotransmitter storage capacity in relation to DCG formation when compared with regular PC12 cells. The examination of PC12 REST cell sections did detail the presence of some DCGs. However, they were fewer in number to those identified in normal PC12 cells and only accounted for approximately 0.1% of the cytoplasmic volume. This result was also consistent with the reserpine sensitivity data, given that there was some reserpine-sensitive [3H]NA accumulation in PC12 REST cells when compared with PC12 cells (Fig. 5a).
The fact that DCGs were visible and [3H]NA accumulation was reserpine sensitive (Fig. 5) suggested that the main contributing factor to the RSP-deficient phenotype in PC12 REST was because of a block in the sensing/triggering mechanism for membrane fusion (cf. PC12 HZ4 cells where processes preceding fusion are affected, i.e. neurotransmitter storage capacity). Indeed, the microarray experiments (Table 1) showed that RE1 proximal genes with recognized roles in the initiation of fusion (e.g. secretogogue receptor 1a), ion channel genes capable of contributing to the resting potential difference across the plasma membrane (e.g. four potassium channel subunits) and calcium ion channels genes with integral roles during the triggering of secretory organelle membrane fusion (e.g. cab2) were transcriptionally repressed. To test this hypothesis, the release of [3H]NA from PC12 REST and regular PC12 cells was recorded in the presence of a plasma membrane calcium ionophore (ionomycin) and in the presence or absence of calcium ions (Ca2+). Given that it has been demonstrated before that a cytoplasmic influx of Ca2+ is essential to initiate synaptic vesicle/granule fusion (Sudhof 2004), it was predicted that pre-incubation of cells with ionomycin would uncouple this process from the requirement for membrane depolarization and thus restore the RSP-deficient phenotype in PC12 REST cells. When PC12 cells were pre-incubated with ionomycin and transferred to Ca2+-containing buffer, they release 20% of their intracellular [3H]NA (Fig. 6a). As predicted, PC12 REST cells released similar levels of [3H]NA, in effect reversing their RSP-deficient phenotype. Because vesicle/granule release is triggered by an influx of Ca2+ ions through VOCCs, and many of these calcium channel genes are potential targets for REST regulation (see RE1db; Bruce et al. 2004), we examined calcium influx in response to plasma membrane depolarization. PC12 and PC12 REST cells were loaded with a calcium-sensitive fluorescent dye and intracellular calcium recordings taken before, during and after repeated potassium ion induced membrane depolarization in a superfusion chamber (Fig. 6b). The PC12 cells produced a well-defined transient peak of fluorescence, consistent with an influx of Ca2+ via plasma membrane VOCCs. This decayed as the potassium was washed from the cells, but the response was reproduced when the cells were depolarized for a second time. However, the PC12 REST cells showed a much reduced fluorescence when initially depolarized that could not be reproduced when depolarized for a second time. A lack of increased intracellular Ca2+ in response to membrane depolarization suggested a deficiency in VOCC activity, consistent with a lack of VOCC gene expression in PC12 REST cells. However, it is important to note that other sources of Ca2+ increase, such as release from intracellular stores, remained unaffected, as treatment with ATP resulted in a robust increase in intracellular Ca2+ in PC12 REST cells. These data are consistent with a sensing/triggering deficiency contributing to the RSP phenotype in PC12 REST cells resulting from a lack of voltage-dependent calcium influx.
Transcription of repressed genes is induced by abrogating REST function
The above data demonstrates that the ectopic expression of rest in PC12 cells, a model neuroendocrine system extensively utilized in RSP studies, results in the reduced expression of many RE1-containing genes with characterized roles within the RSP. Furthermore, these changes in the transcriptome are sufficient to ablate neurosecretory competence with respect to noradrenaline secretion. These deficiencies are manifest on multiple levels, from secretory organelle biogenesis to sensing changes in plasma membrane potential difference and consequent triggering of stimulated secretion mechanisms. We therefore sought to examine whether we could rescue this RSP-deficient phenotype by abrogating REST function in PC12 HZ4 and PC12 REST cells. To this end, we utilized previously derived adenoviral vectors harbouring either gfp alone (Ad) or gfp and dominant negative rest (Ad:DN; Wood et al. 2003). Figure 7(a) shows fluorescence and light microscopy images of PC12 HZ4 and PC12 REST cells in culture 24 h post-infection. gfp expression confirms successful transgene delivery in each of the four experimental conditions. Figure 7(b) details the fold activation in mRNA expression in PC12 HZ4 and PC12 REST cells when infected by Ad:DN over that with the control Ad vector. As can be seen, abrogation of REST function by dominant negative REST results in increased expression of cga and snap25 in PC12 HZ4 and PC12 REST. The magnitude of the effect is greater in PC12 HZ4 than PC12 REST and possibly reflects the more severe nature of the RSP-deficient phenotype in PC12 HZ4. These results substantiate our data demonstrating that REST exerts its negative effect on the RSP via a direct transcriptional regulatory mechanism on genes with RSP function. Interestingly, we also assayed the expression of the chromogenin Bcgb gene transcript (which is very closely related to cga; Fig. 7b) and found that it could be induced in both PC12 HZ4 and PC12 REST cells, although the effect was very robust within PC12 HZ4 cells. Given that the cgb gene was not represented on the Affymetrix microarray chips we were using, we initially overlooked this gene as a potential REST target. It is therefore likely that the genes identified in this study represent only a subset of RSP-related genes that are regulated by REST.
We have shown that by regulating the expression of many genes that are involved in discrete steps of the RSP, the levels of one transcription factor (REST) can have profound effects on the functioning of the RSP. This is especially important when one considers cell types in which neurosecretory competence is an important component of cellular phenotype. These include neurones for which the RSP is important in synaptic transmission, and pancreatic islet beta cells which secrete insulin in a glucose-responsive manner. Both cell types express very low or undetectable levels of REST (Chong et al. 1995; Schoenherr and Anderson 1995; Atouf et al. 1997; Palm et al. 1998; Abderrahmani et al. 2001), a fact not surprising given the data presented herein. Given the inverse correlation between rest expression and RSP functioning, it is perhaps prescient to consider possible mechanistic roles for REST in various disease states in which RSP activity is perturbed. For example, REST levels are known to increase in pyramidal neurones of the CA1 region of the hippocampus in response to ischaemia (Calderone et al. 2003), and across the hippocampus in response to kainate-induced seizures (Palm et al. 1998). Under these conditions, it is possible that down-regulation of RSP components could attenuate regulated secretion and, in epilepsy, contribute to refractory periods in which neurones are protected from re-entering seizure. It is also possible that increased rest expression in pancreatic islet beta cells could contribute to diabetes by preventing the glucose-dependent release of insulin. Indeed, a study in which a rest transgene was expressed within beta cells detailed such an impairment (Abderrahmani et al. 2004). It is therefore likely that, in addition to contributing to the establishment of neurosecretory competence, REST harbours the potential to dynamically regulate RSPs in diseased or stressed states.
Using the rat pheochromocytoma cell line PC12, we have shown that rest expression is associated with the transcriptional repression of genes involved in the RSP, and that these genes are found proximal to putative REST binding sites in the genome (Bruce et al. 2004). Crucially, this transcriptional repression results in a complete block in regulated secretion. We have noted that more RE1-containing genes were found to be repressed in PC12 HZ4 than in PC12 REST cells and that correlated with a more severe attenuation of RSP function (whereby not only genes relating to vesicular/granule membrane fusion were repressed but DCG storage/secretory organelle biogenesis was also abrogated). Given that the rest transgene in PC12 REST cells is expressed at a much lower level than that in PC12 HZ4 (Fig. 1), it is possible the observed differences are as a result of differing levels of REST in the two cell lines. Another possible explanation is provided by the differing repressor domain make-up of full-length REST and HZ4 REST (Fig. 1). The reason why REST requires and retains two independent repression domains is unknown. Given that the PC12 REST cell phenotype is less severe than that observed in PC12 HZ4, it is possible that the C-terminal domain, or more specifically the co-factors recruited, can act to antagonize the functioning of the N-terminal domain. The existence of such a mechanism represents an intriguing new avenue for future research. Despite these speculations, it is important to note that, despite the severity of the phenotype, both PC12 HZ4 and PC12 REST cells are defective with respect to the RSP.
It could be argued that our model predicts that inhibition of REST function in cells lacking functioning RSPs (e.g. fibroblasts) could initiate attainment of neurosecretory competence. However, this seems improbable, as expression of other required factors are likely to be lacking. Indeed, we noted that rest expression in PC12 cells had no effect on the expression of the nat gene (Fig. 3). Given that this gene is central to the noradrenergic secretory phenotype, it is unlikely to be reactivated in cell lineages lacking RSPs when REST function is inhibited. However, there is precedence for similar neurosecretory incompetent PC12 cells; PC12-27 (Corradi et al. 1996) and PC12 A35C (Pance et al. 1999). In PC12 A35C, the expression of many genes involved in the RSP is decreased and, similarly to PC12 HZ4, they lack DCGs. Cell fusion experiments between PC12 A35C and regular PC12 cells lack a functioning RSP, indicating that the phenotype is the result of an inhibitory factor within PC12 A35C (Pance et al. 1999). Data presented here implicate REST as that factor and Pance and co-workers have now identified REST as an important component of the A35C phenotype (A. Pance and A. Jackson, personal communication). A second neurosecretory incompetent PC12 cell line, PC12-27, has also been described (Corradi et al. 1996; Grundschober et al. 2002). These cells also show elevated REST levels, although the role of REST in this phenotype is less clear, as the authors were unable to mimic the phenotype by ectopically expressing REST (Grundschober et al. 2002). However, the level at which ectopic REST was expressed was not reported and, given the data presented here, the severity of the phenotype may be linked to the level of REST expression. A further interesting observation to arise from this study is that REST is able to regulate the transcriptional activity of both chromogranin A and chromogranin B genes. In the field of DCG biogenesis research there have been conflicting reports suggesting that either chromogranin A (Kim et al. 2001) or chromogranin B (Huh et al. 2003) represents the crucial ‘on/off’ master regulator of DCG formation. It could be argued that REST is a better candidate to fulfil this contentious role.
Our data suggest that REST, by directly regulating the transcription of genes involved in the RSP, functionally regulates the RSP and hence contributes to the neurosecretory phenotype. This is the first time that major components of the RSP have been shown to be regulated by a single transcription factor and offers novel insight into disease states in which RSP functioning is abrogated.
This work was supported by the UK Biotechnology and Biological Sciences Research Council (BBSRC), the Wellcome Trust and Czech Academy of Sciences; GAAV5011206 (AV0Z50110509).