Tomoko Kuwabara ^{1, 2}, Jenny Hsieh ¹, Kinichi Nakashima ^{1, 3}, Kazunari Taira ², and Fred H. Gage ^1, *

¹ Laboratory of Genetics, The Salk Institute, 10010 North Torrey Pines Road, La Jolla, CA 92037 USA
² Gene Function Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Central 4, 1-1-4 Higashi, Tsukuba Science City 305-8562, Japan
³ Department of Cell Fate Modulation, Institute of Molecular Embryology and Genetics, Kumamoto University, Kumamoto, 860-0811, Japan

*Correspondence:  Fred H. Gage:   858-453-4100 (phone):   858-597-0824 (fax)
E-mail:  gage@salk.edu

Discovering the molecular mechanisms that regulate neuron-specific gene expression remains a central challenge for CNS research. Here, we report that small, noncoding double-stranded (ds) RNAs play a critical role in mediating neuronal differentiation. The sequence defined by this dsRNA is NRSE/RE1, which is recognized by NRSF/REST, known primarily as a negative transcriptional regulator that restricts neuronal gene expression to neurons. The NRSE dsRNA can trigger gene expression of neuron-specific genes through interaction with NRSF/REST transcriptional machinery, resulting in the transition from neural stem cells with neuron-specific genes silenced by NRSF/REST into cells with neuronal identity that can express neuronal genes. The mechanism of action appears to be mediated through a dsRNA/protein interaction, rather than through siRNA or miRNA. The discovery of small modulatory dsRNAs (smRNAs) extends the important contribution of noncoding RNAs as key regulators of cell behavior at both transcriptional and posttranscriptional levels.

Introduction:

The expression of cell type-specific differentiation genes is thought to depend on both positive and negative gene expression controls, which are implemented throughout the developmental history of cells. Numerous genetic studies provide evidence that cell type-transpecific gene expression activators and repressors are essential components of the process. In addition to the conventional transcription machinery, gene expression control by small noncoding RNAs, at the posttranscriptional level, appears to be essential (Eddy, 2001; Fire et al., 1998; Pasquinelli, 2002). Base complementarity allows very small noncoding RNAs to be sequence-specific, and since they act both in cis and trans, their potential functional roles at specific regulatory stages can be expanded. Noncoding RNA genes, which produce transcripts, can function directly as structural, catalytic, or regulatory RNAs, rather than as expressed mRNAs that encode proteins. Recently, several groups have carried out systematic noncoding RNA gene identification screens (Dostie et al., 2003; Lau et al., 2001; Lee and Ambros, 2001). Studies indicate that the prevalence of noncoding RNA genes has indeed been underestimated. Plants, flies, worms, mice, and humans all harbor significant numbers of small RNAs that are likely to play regulatory roles. Although most of the identified noncoding RNAs have unknown functions, their sequences are typically conserved among different species, and many have intriguing expression patterns in different tissues or stages of development. Therefore, noncoding RNAs may have a general role in modulating gene expression in many aspects of development, such as tissue-specific patterning and cell fate specification.

The regulatory mechanisms of gene expression, which determine cell fates giving rise to each lineage, remain largely unknown. Cell fate decisions might involve the regulatory activities of noncoding RNAs. To examine whether noncoding RNAs contribute to cell fate specification of adult neural stem cells, we isolated small, noncoding RNAs from adult hippocampal neural stem cells during lineage-specific differentiation. To identify possible target nucleic acids, public database searches were used for genomic sequences. Among the large number of small noncoding RNAs that appeared, one unique sequence emerged from the “neurogenesis” noncoding RNA pools. This sequence defined the NRSE/ RE1 (neuron restrictive silencer element), which is recognized by the NRSF/REST transcriptional regulator (Chong et al., 1995; Schoenherr and Anderson, 1995).

In the CNS, neuronal restricted silencing factor/RE-1 silencing transcription factor (NRSF/REST) plays a critical role as a key transcriptional repressor for neuron-specific genes in nonneuronal cells (Chen et al., 1998; Huang et al., 1999; Palm et al., 1998; Schoenherr et al., 1996). NRSF/REST is a kruppel family zinc finger protein and binds specifically to a 21- to 23-base pair (bp) conserved DNA response element (NRSE/RE1). NRSE/RE1 sequences are encoded within a broad range of genes involved in neuronal development and function, including ion channels, neurotransmitter receptors and their synthesizing enzymes, receptor-associated factors, neurotrophins, synaptic vesicle proteins, growth-associated and cytoskeletal and adhesion molecule factors involved in axonal guidance, transport machinery, transcription factors, and cofactors. The consensus NRSE/RE1 sequence is conserved between Xenopus, mouse, rat, chicken, sheep, and human. NRSF/REST mediates transcriptional repression through the association of the N-terminal repressor domain with the mSin3/histone deacetylase-1/2 (HDAC1/2) complex and through the association of C-terminal repressor domain with the CoREST complex (by recruitment of MeCP2 or HDACs) (Huang et al., 1999; Lunyak et al., 2002; Naruse et al., 1999).

In this work, we report that the identified, noncoding RNA-containing NRSE sequence forms double-stranded RNA (dsRNA) in lengths of about 20 bp, and that the NRSE dsRNA activates expression of NRSE/RE1-containing genes during an early stage of neurogenesis. The NRSE dsRNA modulates the NRSE/RE1 DNA-NRSF/REST protein machinery to switch neuronal gene expression from a repressed state in stem cells to an active state in early neurons. The NRSE dsRNA is necessary and sufficient to direct multipotent neural stem cells specifically down a neuronal lineage, suggesting it can function as an endogenous inducer of neuronal differentiation. The apparent gene activation effects of the NRSE dsRNA clearly distinguish it from the gene silencing effects of cellular miRNA/siRNAs and suggest a novel function for noncoding RNAs at a transcriptional level.

Results:

Identification of a Neuron-Specific, Small Noncoding RNA from Adult Hippocampal Neural Stem Cells

To investigate the role of small, noncoding RNAs in the differentiation of neural stem cells, 20- to 40-nucleotide (nt) RNAs were cloned from total RNA extracted from adult hippocampal neural stem cells (HCN-A94; Gage et al., 1995). We obtained more than 50 unknown noncoding RNAs and decided to focus on one RNA sequence that contained a match to the 21 nt NRSE/RE1 DNA sequence in the antisense orientation (asNRSE; Supplemental Figure S1 available at http://www.cell.com/cgi/content/full/116/6/779/DC1). 

 Screening of Noncoding RNAs

 To investigate the role of small noncoding RNAs in the differentiation of neural stem cells, 20- to 40-nucleotide
 (nt) RNAs were cloned and sequenced from total RNA extracted from adult hippocampal neural stem cells.
 Total RNA (about 1 mg) extracted from HCN-A94 cells that have been treated with RA + FSK for 4 days was
 loaded on a denaturing 15% PAGE with size marker, EZ-Load molecular ruler (BioRad). 20- to
 40-nucleotide (nt) RNAs were separated and recovered. The purified RNA was ligated with 5'
 phosphorylated nucleotide 3' adapter by T4 RNA ligase (NEB). The complementary primer for the adapter
 was hybridized and RNA sequences were converted into dsDNA by using Superscript II cDNA cloning kit
 following the manufacturer's instructions (Gibco-BRL).  Annealed 5' end DNA adaptor was ligated
 with these dsDNA by T4 DNA ligase and amplified by PCR with primers having 5' adapter and 3' adapter
 sequences. PCR products were cloned into pCR II plasmid by using the TOPO TA cloning kit (Invitrogen). We
 examined 6 different neural cell culture samples and obtained more than 50 unknown noncoding RNAs (data not shown). Cloned sequences were analyzed by Celera mouse gene database (Celera Discovery System). We
 decided to focus on one RNA sequence that contained a match to the 21-nt NRSE/RE1 DNA sequence in the
 antisense orientation (asNRSE; Supplemental Figure S1).

 Diversity of Regulatory Repression Patterns of Neuron-Specific Genes with the NRSE/RE1-Containing Neuron-Specific Genes

 We examined the function of NRSE dsRNA in gene specific activation.  Both HDACs and DNA methylation
 have been shown to be critical roles for gene silencing on chromatin.  HCN A94 cells were treated with either
 an HDAC inhibitor trichostatin A (TSA) or with de-methylation reagent 5'-aza-cytidine (5AzaC). Total
 RNA was extracted from the cell that had been treated with 5 nM TSA for 2 days and 3 mM 5AzaC for 4
 days.  RNA samples were also prepared from the HCN A94 cells infected with lentivirus to express either
 NRSE dsRNA or mtNRSE dsRNA (RNA was extracted 4 days after the infection).  By using these RNAs,
 RT PCR analysis was performed with specific genes (mGluR2, NaCh II, SCG10, Synapsin I and M4 AchR)
 and a control gene that lacks NRSE/RE1 sequences (GAPDH).

 Compared with the level of control mRNA in untreated HCN A94 progenitor cultures, endogenous expression
 levels of mRNAs of mGluR2, NaCh II, SCG10, and Synapsin I were increased in TSA-treated cells with no
 effects on the expression on GAPDH (Supplemental Figure S2).  The level of M4 AChR mRNA in NRSE
 dsRNA-expressing cells was slightly higher than that of the TSA-treated controls.  However, the treatment
 with 5AzaC showed significant activation of the M4 AChR gene. Similarly, treatment with 5AzaC dramatically
 increased the expression of mGluR2 and NaCh II mRNAs. These differences in repression responses among
 neuronal genes reinforce the idea that there may be diversity in the REST/NRSF regulatory machinery.

Supplemental Figure S1. The Sequence that Contained a Match to the 21-nt NRSE/RE1 DNA
 Sequence in the Antisense Orientation Is Shown. The nucleotides (shown in black) flanking the NRSE/REST sequence are not from the 3' adapter.  It is not clear whether these flanking sequences are derived from a genomic sequence or whether these sequences have been generated during the process of cloning/ligation.

Supplemental Figure S2. RT-PCR Analysis of NRSE/RE1-Containing Genes in HCN A94 Cells.  Cells were treated with either an HDAC inhibitor trichostatin A (TSA) or with demethylation reagent 5’ -aza-cytidine (5AzaC).

The NRSE/RE1 sequence is usually localized within promoter regions of neuron-specific genes and is recognized by the NRSF/ REST protein to restrict neuron-specific gene expression; however, the function of a DNA element within an RNA sequence is unclear.

Northern analyses revealed an asNRSE RNA corresponding to about 20 nt in length within the neuronal population (cells treated with 1 mM retinoic acid [RA] and 5 mM forskolin [FSK] for 4 days). Surprisingly, our control probe for sense-strand revealed an ~20 nt sense NRSE RNA(sNRSE) within the same neuronal population (Figure 1A, probe for sense) suggesting that these RNAs might exist as double-stranded forms within the cell. Low amounts of both sense and antisense RNAs could also be detected within the progenitor population, but the expression levels in the neuronal population were much higher relative to the progenitor population. The expression of low levels of both RNAs within the progenitor cultures could be due to the presence of some cells already committed to specific lineages. Nevertheless, we could not detect any NRSE RNAs within an astrocyte population.

To determine which neuronal stages express the dsRNA, a time-course Northern blot analysis was performed after neuronal induction. Cells at 2 and 4 days after induction of neural differentiation (RA+FSK) contained the highest amounts of the NRSE dsRNA; as maturation proceeded, the levels of the NRSE dsRNA apparently decreased (Figure 1B). These data show that the NRSE dsRNA appeared at an early stage of neurogenesis rather than at more mature stages.

Figure 1. Identification of NRSE dsRNA from Adult Hippocampal Neuronal Cells

    (A) Northern blot analysis of NRSE dsRNA. Both asNRSE and sNRSE RNA corresponding to approximately 20 nt in length exist in the neuronal population.

    (B) Time-course Northern blot analysis after RA/FSK induction. Cells 2 and 4 days after induction of neural
    differentiation contain the highest amounts of the NRSE dsRNA, suggesting that NRSE dsRNA appears at an early stage of neurogenesis.

Neuronal Lineage Induction by the NRSE dsRNA

To determine the function of the NRSE dsRNA, we expressed them in HCN-A94 cells. We made lentiviral vectors with U6 promoter-driven sNRSE, asNRSE and NRSE dsRNA expression cassettes. After infecting progenitor cells with virus, cells were maintained without FGF-2 for 4 days. In control infections (lentivirus with an empty U6 cassette), we did not observe any obvious effects on cell morphology (Figure 2A). Expression of single sNRSE or asNRSE RNA alone also had no obvious effects (Figure 2A). However, when we introduced both sNRSE and asNRSE RNA together, significant morphological changes were observed (Figure 2A). These cells extended processes indicative of differentiation and some cells made large but flat clusters with long processes.

Figure 2. NRSE dsRNA Induces Neuronal Differentiation of Progenitor Cells

    (A) Neural progenitor cells were infected with either a control virus or viruses expressing sense, antisense or
    dsNRSE RNAs.

    (B) Immunocytochemical analysis of cells with NRSE dsRNA. Expression of NRSE dsRNA in progenitor
    cultures resulted in increased numbers of neuron-specific marker-positive cells.

    (C) Quantitative analysis of NRSE dsRNA activity as an inducer of neuronal differentiation. The reporter assay was performed by using lineage-specific gene promoter-driven luciferase constructs. Scale bars are equal to 10 mm.

We next performed immunocytochemistry with markers of various differentiated neural lineages. Introduction of NRSE dsRNA in progenitor cultures resulted in increased immunocytochemical staining of neuron-specific markers, including bIII-tubulin (TUJ1), NF200, and calbindin (Figure 2B). Cells containing the NRSE dsRNA were completely negative for the astrocyte marker GFAP and oligodendrocyte marker RIP (Figure 2B).

Quantitative Analysis of NRSE dsRNA Activity as an Endogenous Inducer of Neuronal Differentiation

We next assessed the effects of noncoding NRSE dsRNA, both in progenitor cultures and during lineage-specific differentiation with reporter constructs. Stage-specific promoter-based reporter assays allowed us to quantify the activity of NRSE dsRNAs comparatively.

We used a Sox2 promoter-driven luciferase construct as an undifferentiated neural progenitor-specific reporter construct. Luciferase values from cells 4 days after mock virus (control) infection were set as 100% (Figure 2C; the immunostaining is also shown in the right image). No obvious difference was observed for sNRSE RNA and asNRSE RNA; however, there was a significant decrease in luciferase activity in cells infected with NRSE dsRNA.

A bIII-tubulin (TUJ1) promoter-driven luciferase construct was used for the neuron-specific reporter construct. RNA-expressing virus-infected progenitor cells were cultured in 1 mM RA and 5 mM FSK for 4 days. The luciferase activity increased more than 4 times when compared with the activity in the progenitor culture (data not shown). Many cells stained positive for TUJ1 (control with mock virus infection; Figure 2C); therefore, the luciferase value of the cell at this time point was taken as 100%. Expression of either sNRSE RNA or asNRSE RNA alone had no obvious effects on neuronal differentiation. In contrast, the NRSE dsRNA specifically increased the TUJ1 promoter-luciferase activity more than 2 times relative to control.

The GFAP and MBP promoters were prepared as lineage-specific luciferase assays for astrocyte and oligodendrocyte differentiation, respectively. To induce astrocyte differentiation, cells were treated with a combi- nation of 50 ng/ml BMP-2, 50 ng/ml LIF and 1% FCS. Four days later, the GFAP-luciferase activity increased more than 5-fold when compared with the activity in the progenitor culture (data not shown). To induce oligodendrocyte differentiation, FGF2 withdrawal of progenitor cultures results in some spontaneous differentiation. Two days later, the MBP-luciferase activity increased more than 3-fold when compared with the activity in progenitor cultures. We did not detect any obvious differences in the cases where sNRSE RNA and asNRSE RNA were expressed during astrocyte or oligodendrocyte differentiation. However, significant decreases in luciferase activity were detected when the NRSE dsRNA was introduced under each differentiation condition. Furthermore, NRSE dsRNA-expressing virus-infected cells that remained under each differentiation condition appeared to be neurons (TUJ1 positive, data not shown).

Increased Expression of Neuron-Specific Genes Containing NRSE/RE1 by the NRSE dsRNA

To investigate the mechanism of action of the NRSE dsRNA, we first considered whether the NRSE dsRNA might function as a miRNA/siRNA. It has been discovered that miRNA/siRNAs exist as 21–25 nt dsRNAs and target cellular mRNAs in a complementary fashion, leading to a process of posttranscriptional gene silencing (Hutvagner et al., 2001; Pasquinelli, 2002).

Figure 2 demonstrates that neuronal lineage induction is one of the major effects of the NRSE dsRNA. If the NRSE dsRNA mediated the silencing of the NRSF/REST gene itself by a miRNA/siRNA-like function and if NRSF/ REST functions as a repressor of neuronal gene expression, the repression of neuron-specific genes may be eliminated, resulting in neuronal lineage induction. However, there is no apparent NRSE sequence within the NRSF/REST mRNA, making it an unlikely target of the NRSE dsRNA at the posttranscriptional level. In fact, when total RNAs were extracted from HCN-A94 cells infected with NRSE dsRNA-expressing virus for 4 days, reverse transcription (RT) PCR analysis revealed that the introduction of the NRSE dsRNA did not appear to change the expression of the NRSF/REST itself (Figure 3A, top left).

NRSE sequences are preferentially localized within promoter regions of neuron-specific genes. We performed sequence database search, and found more than 60 NRSE/RE1 sequences in the mouse genome. We next examined the direct effects of NRSE dsRNA on the expression of genes that have the NRSE/RE1 element in their promoters (SCG10, Synapsin I , NaCh II, M4 mAChR, and mGluR2). RT-PCR analysis revealed that the NRSE dsRNA increased expression levels of NRSE/RE1-containing genes (Figure 3A). In progenitor stages, expression levels of SCG10, Synapsin I, NaCh II, M4 mAChR, and mGluR2 were very low. Upon introduction of NRSE dsRNA, significant transcriptional activation was observed (Figure 3A). These gene-activating events appeared to be NRSE/RE1-gene specific; no obvious increases in the expression of GAPDH and b-actin genes were detected.

To determine how widespread the NRSE dsRNA-dependent gene activation was, we monitored the activity with reporter assay using the mGluR2 promoter. The mGluR2 promoter containing an NRSE/RE1 DNA element was fused to EGFP and this construct was transfected into various cell types infected with or without the NRSE dsRNA lentivirus. In HCN-A94 progenitor cultures, most of the cells were negative for GFP expression. In contrast, in progenitor cultures with the NRSE dsRNA, the number of GFP-positive cells increased (Figure 3B). Neurosphere cultures were prepared from whole brain of 10-day-old ICR strain mice, and primary neural stem cells were derived from ventricular zone, hippocampus, and whole brain of the 129/SvJ strain of adult mice. In all cases, the introduction of NRSE dsRNA mediated a substantial increase in mGluR2 promoter activity (Figure 3B).

Figure 3. Effect of NRSE dsRNA on the Expression of Neuron-Specific Genes Containing the NRSE/RE1
    DNA Element

    (A) RT-PCR analysis showed that NRSE dsRNA increased transcription levels of NRSE/RE1-containing genes without affecting the expression of NRSF/REST.

    (B) Reporter assay for NRSE dsRNA activity using the mGluR2 promoter-driven EGFP construct. Effects of
    NRSE dsRNA on gene activation were assessed in neurosphere cultures and primary neural stem cells. Relative fluorescence intensity was plotted on a log scale (right). Scale bar is equal to 10 mm.

    (C) Schematic diagram of the constructs to examine the requirement of sequence specificity of NRSE. The
    NRSE/RE1 element was fused upstream of TATA box and linked to the luciferase gene. Mutated NRSE/RE1
    DNA element on the reporter construct (mtNRSE-TATA) and mutated NRSE dsRNA expression construct at
    a critical recognition site (mtNRSE dsRNA) were also prepared.

    (D) Sequence requirement of both NRSE dsRNA and the NRSE/RE1 DNA element for gene activation.
    Luciferase assay showed that NRSE dsRNA-dependent gene activation requires a specific sequence homology between the NRSE/RE1 DNA element and NRSE dsRNA.

Critical Sequence Requirement of Both the NRSE dsRNA and NRSE/RE1 DNA Element for Gene Activation

To investigate the sequence requirement and specificity, we prepared a set of simple reporter constructs. The NRSE/RE1 element was fused upstream from the 260 bp CMV minimal promoter carrying a TATA box and linked to the luciferase gene (NRSE-TATA, Figure 3C). A mutated NRSE/RE1 element was prepared similarly (mtNRSE-TATA). We also made expression cassettes for the NRSE dsRNA and a mutated NRSE dsRNA (mtNRSE dsRNA).

When we introduced TATA-luciferase constructs lacking the NRSE/RE1 element, no differences in the luciferase activities were detected between cells infected with no RNA (control)-, NRSE dsRNA- and mtNRSE dsRNA- expressing virus constructs (Figure 3D, gray bars). In contrast, when we introduced NRSE-TATA-luciferase constructs, the NRSE dsRNA increased the expression levels of NRSE-TATA-luciferase gene more than 2.5 times (Figure 3D, orange bars). We next tested the effects of a mutated NRSE dsRNA on NRSE-TATA-luciferase activity. The mutations changed sequence specificity while preserving dsRNA structure. Interestingly, when the NRSE dsRNA was mutated (mtNRSE dsRNA), no additional increase relative to control was observed (Figure 3D). On the other hand, the introduction of a mutated NRSE/RE1 DNA element (mtNRSE-TATA) in combination with an intact NRSE dsRNA was not enough to induce further gene activation (Figure 3D, green bars). These results show that NRSE dsRNA-dependent gene activation requires a critical sequence homology between the NRSE/RE1 DNA element and the NRSE dsRNA.

RNA-Directed Chromatin Changes of NRSE/RE1-Containing Genes in Adult Hippocampal Neural Cells

NRSF/REST proteins interact with histone deacetylase (HDAC1) and methyl-CpG binding protein (MeCP2) to form a repressive chromatin state in nonneuronal cells (Huang et al., 1999; Lunyak et al., 2002). To investigate the nature of the transcriptional activation of NRSE/RE1-containing genes by the NRSE dsRNA, we performed chromatin immunoprecipitation (ChIP) assays. The promoter regions of mGluR2 and SCG10 genes were assessed as representative NRSE/RE1-containing genes, since these genes have been characterized in mechanistic studies of the NRSF/REST repressor complex (Myers et al., 1998; Naruse et al., 1999).

We prepared ChIP samples from HCN-A94 cells during progenitor (with FGF2) and differentiated stages (neurons, astrocytes, and oligodendrocytes). During stem/progenitor stages, as well as nonneuronal stages (oligodendrocytes and astrocytes), both mGluR2 and SCG10 genes were associated with NRSF/REST and HDAC1 (Figure 4A, second and third rows). Notably, NRSF/REST was always found to be associated with endogenous mGluR2 and SCG10 promoters, in the region of the NRSE/RE1 (Figure 4A, second row). In the case of mGluR2 promoter, in addition to HDAC1, methyl-DNA binding proteins of MeCP2 and MBD1 were found associated with the NRSE/RE1 region (Figure 4A, fourth and fifth rows). As for the SCG10 promoter, we observed decreased association of MeCP2 and MBD1, even though this gene is apparently repressed in the nonneuronal state, suggesting diversity of the repression machinery depending on specific gene/promoters containing NRSE/RE1 elements.

Figure 4. NRSE dsRNA-Directed Chromatin Changes of NRSE/RE1-Containing Genes

    (A) ChIP assay for chromatin regulating factors.

    (B) Reporter assay for NRSE dsRNA function to convert NRSF/REST from a repressor to an activator using
    the GluR2 promoter-driven luciferase construct. Intact and mutated mGluR2-luciferase constructs were
    prepared and the level of luciferase activity driven by each promoter in the presence or absence of NRSE
    dsRNA expression was compared.

In contrast, clear evidence of derepressed chromatin states was seen for both mGluR2 and SCG10 genes in neurons, where decreased association of HDAC1 with the NRSE/RE1 element was found (Figure 4A, third row, third lane). The CREB binding protein (CBP)/p300 family of transcriptional coactivators possessing histone acetyltransferase activity has been shown to interact with various transcription factors to activate genes (Bannister and Kouzarides, 1996). We detected an increase in the association of CBP, acetylated histone H4, and acetylated histone H3 with both promoters when cells were in a state in which these genes are actively expressed (Figure 4A, sixth to eighth rows, third lanes). SWI/SNF chromatin-remodeling factors, BRG1 and BAF170, were also found to associate with the NRSE/RE1 element, as part of a possible machinery to remodel the chromatin state for active expression of neuron-specific genes in neuronal cells (Figure 4A, eighth and ninth rows, third lanes).

Importantly, the ChIP assay revealed that, upon the introduction of NRSE dsRNA into progenitor cells, there was decreased association of the repressor proteins MeCP2, MBD1, and HDAC1 with the NRSE/RE1, resulting in the activation of neuronal genes. The fact that NRSF/REST still occupied the NRSE/RE1 locus suggests that NRSF/REST may be involved in an alternative chromatin structure with acetylated histones to activate transcription. For this transition step to occur, chromatin-remodeling factors like BAF170 and BRG1, which had previously been shown to bind NRSF/REST (Battaglioli et al., 2002), may be required for remodeling the chromatin through their ATPase activity.

We also examined the function of NRSE dsRNA in gene-specific activation. HCN A94 cells were treated
with either an HDAC inhibitor trichostatin A (TSA) or with demethylation reagent 5'-aza-cytidine (5AzaC). Endogenous expression levels of mRNAs of mGluR2, NaCh II, SCG10, and Synapsin I were increased in TSA-treated cells, compared with untreated cells (Supplemental Figure S2 available on Cell website). The treatment with 5AzaC showed significant activation of the M4 AChR, mGluR2, and NaCh II mRNAs. Some of these differences in repression responses among neuronal genes reinforce the idea that there may be diversity in the REST/NRSF regulatory machinery.

NRSF/REST Is Converted from a Transcriptional Repressor to an Activator in the Presence of NRSE dsRNA

To determine whether the transactivation of genes by NRSE dsRNA was caused by derepression or by a functional switch of NRSF/REST from repressor to activator, we made GluR2 promoter-driven luciferase constructs [wild-type (GluR2-luciferase, Figure 4B) and mutated NRSE substituted with random nucleotides (mtGluR2-luciferase, Figure 4B)], and compared the level of luciferase activity with and without expression of NRSE dsRNA in adult neural stem cells (Figure 4B). NRSF/REST cannot bind to mutated NRSE sequences (Kraner et al., 1992). In the case of the mutated NRSE construct (mtGluR2-luciferase), the relative luciferase activity is seen at baseline levels, presumably due to a release of NRSF-mediated repression (derepression). We observed at least a 2-fold increase in the wild-type GluR2- luciferase construct upon introduction of the NRSE dsRNA, but not in the mtGluR2-Luciferase construct, indicative of an activation effect. This activation was never observed with the introduction of a mutant NRSE dsRNA or a control vector; in fact there was an active repression of GluR2-luciferase, consistent with NRSF/ REST actions as a repressor. Taken together, these results suggest that: (1) NRSF/REST functions as a repressor in the absence of NRSE dsRNA, (2) NRSF/REST converts to an activator in the presence of NRSE dsRNA, and (3) the activator function of NRSF/REST is dependent on having both a wild-type NRSE/RE1 DNA sequence and a wild-type NRSE dsRNA.

The Loss of Nuclear Localizing NRSE dsRNA Blocks Neuronal Differentiation in Adult Hippocampal Stem Cells

To determine if NRSE dsRNA is necessary for neuronal differentiation, we designed a ribozyme (Rz; Figure 5A) that can specifically cleave one of the strands of the dsRNA sequence, thus inactivating the expression of the NRSE dsRNA. For ribozymes, additional proteins are not needed for catalysis; they only require Mg ²⁺ ions, which are abundant in cells (Eckstein and Lilley, 1996; Warashina et al., 2000). Since it is important to select the appropriate promoter to express the ribozyme in the compartment of the cell where the target RNA is located (Koseki et al., 1998), we first analyzed the localization of NRSE dsRNA by Northern blotting. We found both antisense and sense NRSE RNAs dominantly expressed in the nuclear fraction (Figure 5A), reinforcing the finding that NRSE dsRNAs are not acting as miRNAs, which target cytoplasmic mRNAs to inhibit their translation. Treatment of progenitor cells with the ribozyme completely abolished expression of the NRSE dsRNA. An inactive ribozyme (I-Rz) with one nucleic acid substitution in the catalytic domain was prepared as a negative control and did not affect NRSE dsRNA expression.

Figure 5. Loss of NRSE dsRNA Blocks Neuronal Differentiation

    (A) Nuclear localization of the NRSE dsRNA was confirmed by Northern blot analysis. The ribozyme (Rz)
    sequence was designed to cleave the asNRSE RNA, and the effect was assessed by Northern blot using
    fractionated RNAs extracted from 4-day RA+FSK-treated HCN cells. Nuclear fraction, (N); cytoplasmic
    fraction, (C).

    (B) Cells with Rz cleaving antisense NRSE RNA showed strong antidifferentiation effects. Scale bar is equal to 10 mm.

    (C) The effect of Rz on the function of NRSE dsRNA in each lineage. Cell type-specific promoter-based
    reporter assay was performed in HCN A94 cells. The results shown are the averages of results from three sets of experiments.

We introduced both nuclear specific U6-driven functional Rz and I-Rz into HCN A94 cells by lentiviral infection. No obvious effects were detected at the progenitor stage (since NRSE dsRNAs are not expressed at this stage) compared with the cells in which the NRSE dsRNA had been introduced by lentivirus (Figure 5B). When the culture was switched into the neuronal differentiation condition, in the case of I-Rz, normal neuronal differentiation was observed (Figure 5B, bottom right). However, when the Rz targeting NRSE dsRNA was introduced, cells displayed strong antidifferentiation effects even with RA+FSK stimulation and resembled the morphology of cells in progenitor stages (Figure 5B, top right).

To determine the effect of Rz in each differentiation pathway, a cell type-specific promoter-based reporter assay was performed similar to that shown in Figure 2C. Luciferase values from cells 4 days after control mock virus-infection were taken as 100%. Under progenitor culture conditions, Sox2 promoter-driven luciferase values resulted in no difference in the cases of Rz and I-Rz treatment, probably due to the lack of endogenous dsRNA (Figure 5C).

Under the neuronal condition, the level of the NRSE dsRNA increased, as well as TUJ1 promoter-driven luciferase activity. Introduction of the Rz in this condition significantly reduced the TUJ1-luciferase activity, whereas the I-Rz had no effect (Figure 5C). Under astrocyte or oligodendrocyte differentiation conditions, no obvious differences were detected in the levels of luciferase driven from the GFAP or MBP promoter, respectively, upon either Rz or I-Rz introduction (Figure 5C). Mutant NRSE dsRNA (mtNRSE dsRNA) had no effect on various luciferase assays.

NRSE dsRNA in the Nuclei of Cells Differentiating into Neurons

We next carried out in situ hybridization against NRSE dsRNA and immunostaining for NRSF/REST protein simultaneously. As illustrated in Figure 6A, DAPI (blue) and NRSE RNA (green) colocalized in the nucleus of HCN-A94 cells in neurons (RA+FSK for 4 days, upper images). However, colocalization did not occur when DNA was condensed during cell division (white arrow, Figure 6A, upper image). During DNA condensation, NRSE RNAs remained in the nuclear domain but appeared to be outside of the condensed chromosomal region (white arrows, Figure 6A, upper image). NRSE dsRNA localization in mitotic cells seems to reflect the localization of histone acetylase/proteins, which also appear beyond the condensed chromosomal region. The nature of their actions on transcriptional regulation is in accord with the finding that transcription is repressed during mitosis (Kruhlak et al., 2001). Molecules smaller than 50~70 kDa can translocate back and forth through the nuclear pore through a process of natural diffusion (Stehno-Bittel et al., 1995). Since NRSE dsRNAs are ~20 bp in length (less than 20 kDa), they would likely diffuse throughout the cell. However, NRSE dsRNAs were located specifically in the nucleus, suggesting as yet unknown molecule(s) restricting the localization of the dsRNA to the nucleus.

Figure 6. Localization of NRSE dsRNAs in the Nucleus of Neuronal Cells and Their Interaction with
    NRSF/REST Protein

    (A) Nuclear localization of NRSE dsRNAs and NRSF/REST. Scale bar is equal to 10 mm.

    (B) Binding of NRSE dsRNAs to endogenous NRSF/REST proteins. Proteins that had bound to the
    biotin-labeled oligonucleotides were “pulled down” with streptavidin beads and were analyzed by Western blot.

    (C) EMSA of NRSF/REST protein against NRSE dsRNA and dsDNA. While the concentration of each
    nucleotide was fixed as 20 mM, protein amount was increased 2-fold by each lane depending on arrow
    direction.

NRSF/REST proteins are mainly localized in the nucleus, regardless of cell division (magenta, Figure 6A). Even though NRSF/REST is expressed in all of the cells, the cells expressing higher amounts of NRSE RNAs were also TUJ1-(yellow) positive (red arrows). Conversely, cells expressing NRSE RNAs at the lowest levels were TUJ1-negative (blue arrows).

Interaction between NRSF/REST and the NRSE dsRNA

To examine the potential interaction between NRSF/REST protein and NRSE dsRNA, we incubated the cell extract with biotin-labeled NRSE dsRNAs. We also prepared biotin-labeled NRSE dsDNAs as a positive control. To assess the specificity in the interaction, negative controls with the partial sequence of the multicloning site (MCS) in pBlueScript II SK+ were prepared. Bound proteins were “pulled down” and analyzed by Western blot (Figure 6B). The immunoblot revealed that both the NRSE dsDNAs and dsRNAs bound NRSF/REST, demonstrating an interaction between NRSF/REST and NRSE dsRNA.

To compare the affinity between NRSE dsDNA and NRSE dsRNA to NRSF/REST protein, we carried out electrophoretic mobility shift assay (EMSA). NRSF/REST with cMyc-tag was expressed in 293T cells and immunoprecipitated with anti-cMyc antibody. After purification, the protein was incubated with either NRSE dsDNAs or dsRNAs. We tested a range of NRSF/REST protein concentrations; the highest one tested produced a shift in dsDNA migration, whereas a 16-fold lower concentration of NRSF/REST protein was enough to produce a shift in dsRNA (Figure 6C). Surprisingly, these data revealed that the affinity of NRSF/REST to the NRSE dsRNAs was much higher than the affinity to NRSE dsDNAs. Binding of NRSF/REST to the sequence of the MCS control dsRNAs or dsDNAs was not observed. Furthermore, no apparent band-shift was observed in samples of bovine serum albumin (BSA) incubated with NRSE dsRNAs or dsDNAs. This highly specific binding between NRSE dsRNA and NRSF/REST protein may contribute to a functional switch of the NRSF/REST machinery from transcriptional repressor to activator.

Expression of NRSF/REST Protein and NRSE dsRNA in the Adult Hippocampus

We next examined the expression of NRSF/REST mRNA and NRSE dsRNA in the adult mouse hippocampus by in situ hybridization. Although NRSF/REST mRNA was expressed in nonneuronal glial cells (data not shown), the mRNA was highly expressed in hippocampal neurons (Figure 7), suggesting that NRSF/REST is playing not only a role as a transcriptional repressor in nonneuronal cells but also a role in neurons in vivo. To verify the specificity of the in situ experiments, we did additional experiments with two negative control probes: a probe with the same nucleotide content containing a scrambled sequence and a probe with the same sequence in the reverse direction. There was no detectable signal with either negative control probe (data not shown). An enlarged view of NRSF/REST and NRSE dsRNA expression is shown in Figure 7B. Similar expression patterns of NRSF/REST have been previously documented (Kallunki et al., 1998; Palm et al., 1998; Timmusk et al., 1999). Interestingly, the expression of NRSE dsRNA was highly restricted in the subgranular region of dentate gyrus, in a region where adult neurogenesis is continuously occurring (van Praag et al., 2002; Kempermann, 2003) (Figure 7), supporting the in vitro data that NRSE dsRNAs function at an early stage in neuronal differentiation (Figures 1 and 2).

Figure 7. In Situ Hybridization Analysis for NRSF/REST mRNA and NRSE dsRNA in Adult Mouse
    Hippocampus

    (A and B) NRSE dsRNA expression (red) was highly restricted within the subgranular layer of the dentate gyrus, whereas NRSF/REST mRNA was expressed (green) in a widespread neuronal area in adult hippocampus.

    (B) Higher magnification view.

These results suggest the existence of interactions between proteins (NRSF/REST complex) and dsRNAs (the NRSE dsRNA) in addition to dsDNAs (NRSE/RE1 element). After the participation of dsRNAs in cells at early stages of neurogenesis, the NRSF/REST complex alters binding partners from repressors to activators to initiate transcription of neuron-specific genes. Derepression events might include global changes in cells, but at least the key players—proteins, dsDNAs, and dsRNAs—could recognize each other within the nucleus in order to direct neurogenesis.

Discussion:

NRSE dsRNA Stimulates Neuronal Differentiation through an Interaction with NRSF/REST Complex

In this study, a dsRNA with a restrictive silencer element was identified as a functional transcriptional activator.Genes important for neuronal properties contain the NRSE sequence, which is recognized by the protein NRSF/REST (Palm et al., 1998; Schoenherr et al., 1996). The maintenance of neuronal gene repression in nonneuronal cells depends on the ability of NRSF/REST to bind the NRSE sequence (Chen et al., 1998; Huang et al., 1999; Lunyak et al., 2002). To repress gene expression, NRSF/REST recruits negative transcriptional regulators such as HDACs and methyl-DNA binding proteins (Lunyak et al., 2002; Naruse et al., 1999). The question of how multipotent adult neural stem cells switch from actively repressing neuron-specific genes in the “stem cell state” to actively expressing neuron-specific genes in the “differentiated state” appears to be explained, at least in part, by the NRSE dsRNA. The cell that will become a neuron activates transcription of genes marked by the NRSE. These cells supply noncoding RNA that forms dsRNA with an NRSE sequence. The dsRNA interacts with NRSF/REST machinery, resulting in the NRSF/REST switching cofactors from repressors to activators. This intrinsic ability of the NRSF/REST machinery implies that NRSF/REST can function as a flexible mediator of NRSE regulatory elements.

Multipotent neural stem cells require a highly selective gene regulation system to achieve uniquely different fates. During the uncommitted stem cell state, the genes required for neural differentiation are repressed (Figure 8), whereas commitment to the neuronal lineage requires repression of stem cell-, astrocyte-, and oligodendrocyte-specific genes and activation of neuron-specific genes (Figure 8). Adult hippocampal neural stem cells would be considered one of the cell types that need selective gene regulation for endogenous fate determination. Noncoding dsRNAs encoding NRSE sequences play a unique role in NRSE element dependent gene regulation, without alteration of the expression of a key protein player, NRSF/REST, at the transcriptional level. The expression of NRSE dsRNA modulates the function of NRSF/REST between activator and repressor of neurogenesis.

Figure 8. Schematic Representation of Activation Events by NRSE dsRNA

    NRSE dsRNA can trigger gene expression of neuron-specific genes through interaction with NRSF/REST
    transcriptional machinery. This interaction results in the NRSF/REST complex no longer binding to HDACs,
    MeCP2, and MBD1.

We have shown that NRSE dsRNAs can act as inducers of neuronal differentiation. Interestingly, introduction of NRSE dsRNA alone is sufficient to activate NRSE/RE1-containing neuron-specific genes and induce neuronal differentiation. Furthermore, introduction of a ribozyme targeted against the NRSE dsRNA has antineuronal differentiation effects, suggesting that NRSE dsRNAs are also necessary to induce neuronal differentiation. It should be noted that the NRSF/REST mRNA is highly expressed in adult hippocampal neurons as well as in nonneuronal cells, and the expression of NRSE dsRNAs was highly restricted in the subgranular layer of dentate gyrus, one of the neurogenic regions in the adult mammalian brain (van Praag et al., 2002; Kempermann, 2003). These findings imply that NRSE dsRNAs also participate in neuronal differentiation in vivo.

Possible Mechanism of NRSE dsRNA-Mediated Neuronal Differentiation

NRSE dsRNA binds NRSF/REST as well as the NRSE dsDNA (Figure 6B). A simple model of the mechanism of dsRNA-dependent activation is that the dsRNA captures NRSF/REST as a decoy and releases the genome from the repression. However, the ChIP analysis (Figure 4A) indicates that NRSF/REST protein remains stably associated with the NRSE/RE1 machinery in both the “stem cell state” and the “differentiated states.” These results also show that the nature of the NRSE-containing chromatin changes from a repressed state (association with HDACs, MBD1, and MeCP2) to an activated state (association with acetylated histones) and does not involve a change in the association of the NRSF/REST protein itself. How can we explain a change in NRSF/REST function, even though NRSF/REST proteins appear to remain physically associated at NRSE/RE1 sites within different cell stages? We postulate several models (Figure 8). Model A, based on pull-down experiments (Figures 6B and 6C) and mutation analyses (Figure 3D), proposes that there is a physical interaction between the NRSE dsRNA and NRSF/REST protein, suggesting a critical sequence dependency between dsRNAs and proteins. Basically, NRSE/RE1-containing neuronal genes are actively repressed by the NRSF/REST machinery (through the association of HDACs and methyl-DNA binding proteins). At the onset of neuronal differentiation, the dsRNA interacts directly with NRSE dsDNA-NRSF/REST machinery within the genome and triggers an organizational change in transcriptional activation (Figure 8, Model A). Another possibility is that the NRSF/REST protein acts as a homodimer. In this case, the NRSE dsRNA could bind one monomer of NRSF/REST while the other monomer remains physically associated with dsDNA/chromatin. In Model B, through physical interactions with the NRSF/REST complex, NRSE dsRNAs alter NRSF/REST function, possibly by inducing a conformation change in NRSF/REST and/or associated proteins (Figure 8, Model B). In both cases, after an interaction with NRSEdsRNAs, the NRSF/REST complex can no longer associate with repressor proteins, such as HDACs, MeCP2, and MBD1.

NRSF/REST is a kruppel family zinc finger protein that contains a DNA binding domain with an eight zinc-finger cluster and one zinc finger at carboxyl terminal (Palm et al., 1998; Shimojo et al., 2001; Tapia-Ramirez et al., 1997). A zinc finger domain includes 2 conserved cysteine and 2 conserved histidine residues in a C-2-C-12- H-3-H (C₂H₂) type motif. Zinc finger domains have been found in numerous nucleic acid binding proteins and interact with nucleotides in the major groove of the nucleic acid. They have the ability to bind to both RNA and DNA. C₂H₂ motif zinc finger proteins represent one of the most common nucleic acid binding motifs found in nature (Hoovers et al., 1992). Proteins containing these motifs are generally DNA binding transcription factors that recognize specific sequences in the context of the B-form helix. However, some zinc finger proteins are also able to bind single-stranded RNA, double-stranded RNA, and RNA-DNA hybrids. Recently, a zinc finger protein was identified that possessed higher binding affinity to A-form dsRNA and RNA-DNA hybrids than to B-form dsDNA helix (Finerty and Bass, 1999). Interestingly, the NRSF/REST protein contains similar features to these RNA binding motifs. Although as mentioned above, Figures 6B and 6C showed possible interactions between NRSF/REST and the NRSE dsRNA, it remains to be determined whether NRSF/REST alone is capable of binding to the dsRNA or if this association involves a larger protein complex capable of binding NRSE dsRNA. Some proteins may specifically and dominantly recognize the short dsRNA form itself in the nucleus, and this complex interacts with NRSF/REST within the genome at NRSE/RE1 loci, as a common machinery for the dsRNA-dependent transcriptional regulation (Figure 8, Model B).

Noncoding Small dsRNA Regulates Gene Expression at a Transcriptional Level

In animals, the dsRNA-specific endonuclease, Dicer, produces miRNAs and siRNAs for gene silencing (Bernstein et al., 2001; Hutvagner et al., 2001). miRNA/siRNAs target mRNAs through their sequence homology, leading to gene silencing via the Dicer complex within the cytoplasm. Posttranscriptional gene silencing by non-coding RNA critically contributes to regulation of developmental timing, spatial patterning of cell fates, and cellular physiology (Eddy, 2001; Fire et al., 1998; Pasquinelli, 2002).

Pre-miRNAs approximately 70 nt in length are made within the nucleus, and a protein complex(es) recognizes them, exports them to the cytoplasm, and passes them to the next players for various gene silencing events (Lee et al., 2002). The identified 21–25 bp NRSE dsRNA is smaller; if there are no molecules to keep them within the nucleus, they should naturally diffuse through the nuclear pore and out into the cytoplasm. Since NRSE dsRNAs are clearly localized only in the nucleus (Figures 5A and 6A), there must be molecule(s) involved in sequestering their localization. One major candidate is the NRSF/REST protein, since it can recognize the NRSE dsRNA through sequence specificity by the zinc-finger motifs. Other candidates are the above-mentioned specific proteins that recognize the short dsRNA form itself in the nucleus and perhaps act as global regulators, like Dicer for a miRNAs/siRNA regulatory mechanism (Figure 8, Model B).

The dsRNAs so far identified may regulate mRNA expression at a posttranscriptional step in the cytoplasm. The currently reported NRSE dsRNAs appear to function exclusively at the transcriptional level, suggesting a novel aspect of noncoding dsRNA function. Even though their functions are different, their nucleotide lengths are almost the same (21–25 bp), enabling them to diffuse in cells without limitation. Therefore, dsRNAs might be sequestered in specific cellular compartments through interaction with their cognate protein partner(s) to mediate effects in a spatial-temporal and sequence-dependent manner of target mRNA, DNA, and proteins. Many questions remain, including the exact mechanism that produces small modulatory dsRNAs within the nucleus, and whether a noncoding NRSE gene(s) exists. We believe that the NRSE dsRNA defines a class of functional noncoding RNAs that have primary roles in regulating gene expression at the transcriptional level, and we propose that this class be named small modulatory RNAs (smRNAs).

Experimental Procedures:

Cell Culture

HCN A94 cells were cultured as described (Gage et al., 1995). For neuronal differentiation, cells were cultured in N2 medium (Invitrogen) containing RA (1 mM, Sigma) and forskolin (5 mM, Sigma). For astrocyte differentiation, cells were cultured with 50 ng/ml BMP-2 (R&D systems), 50 ng/ml LIF (Chemicon), and 1% FCS (HyClone) for 4–10 days. For oligodendrocyte differentiation, cells were cultured in N2 medium after FGF-2 withdrawal for 2–4 days. Cell imaging was performed using microscope (Nikon TE300) with a SPOT camera.

Construction of Plasmids

The sNRSE RNA- and asNRSE-expressing lentiviral vectors were constructed by using CSC PW, a lentiviral vector. The CMV promoter was digested out. U6 promoter drives each NRSE RNA sequence with the terminator at the 3' end first amplified by PCR. Each U6 cassette was subcloned into CSC PW. Ribozyme-expressing vectors were constructed similarly. The production of lentivirus has been described elsewhere (Pfeifer et al., 2001), and infections were almost 100% (viral titers were >1.5 x 10 ⁴ Tu/ng defined by the non-P24 assay).

Murine Sox2 promoter on Sox2 pBS SK (gift from Dr. Rizzino) was inserted into pNeoLuci at the site of MCS (Clontech). Murine TUJ1-, GFAP-, MBP-and rat GluR2-promoters were cloned by PCR from genomic DNA and each promoter was inserted into pNeoLuci. The mtGluR2-luciferase construct with mutated NRSE/RE1 substituted with random nucleotides (from TTCAGCACCACGGACAGCGCC to GCATCCGCACCGCTAGCGCAG), was also prepared. The TATA, NRSE-TATA and mtNRSE-TATA luciferase reporter plasmids were constructed in pGL2-basic plasmid (Promega).

Northern Blotting Analysis

Total RNA was extracted with TRIzol reagent (Gibco-BRL). To prepare cytoplasmic fraction, cells were incubated in digitonin lysis buffer (50 mM HEPES/KOH, [pH 7.5], 50 mM potassium acetate, 8 mM MgCl₂, 2 mM EGTA, and 50 mg/mL digitonin) on ice for 10 min. The lysate was centrifuged at 1,000 x g and the supernatant was collected as the cytoplasmic fraction. The pellets resuspended in NP-40 buffer (20 mM Tris-HCl, [pH 7.5], 50 mM KCl, 10 mM NaCl, 1 mM EDTA, and 1% NP-40) were used as the nuclear fraction.
Purified RNA was loaded on a 3.5% NuSieve-Seakem agarose gel (FMC Inc.) and transferred to a Hybond-N nylon membrane (Amersham Biosciences). The membrane was probed with synthetic oligonucleotides that were complementary to the sequences of each sNRSE or asNRSE that had been labeled with ³²P by T4 polynucleotide kinase (NEB). Prehybridization and hybridization were carried out using EazyHyb solution (Clontech) following manufacturer’s instructions.

In Situ Hybridization and Immunofluorescence Studies

Cells were fixed in fix/permeabilization buffer (50 mM HEPES/KOH, [pH 7.5], 50 mM potassium acetate, 8 mM MgCl2, 2 mM EGTA, 2% paraformaldehyde, 0.1% NP-40, and 0.02% SDS) for 15 min. The FITC-/rhodamine-labeled oligodeoxynucleotide probes matching complementary to asNRSE and NRSF/REST mRNA were denatured for 10 min at 70^oC and chilled. Hybridization buffer, containing 20% dextran sulfate and 2% BSA in 4 x SSC, with probes were placed on the cells for 16 hr. Cells were rinsed in 2 x SSC/50% formamide and in 2 x SSC for 20 min each.

Immunofluorescence studies were performed basically as described (Gage et al., 1995): rabbit anti-b tubulin-III (TUJ1; 1/7500, Covance), guinea pig anti-GFAP (1:500; Advanced Immunochemical, Inc.), rabbit anti-NF200 (Advanced Immunochemical, Inc.), mouse anti-RIP (1/250, Immuno), rabbit anti-calbindin (Advanced Immunochemical, Inc.) and DAPI (Sigma). All secondary antibodies were from Jackson ImmunoResearch. Images were analyzed using Bio-Rad Radiance confocal imaging system (Hercules, CA).

Luciferase Assay

Luciferase activity was measured with Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s protocol. The luminescent signal was quantitated with a luminometer (Lumant LB 9501). As an internal control, a plasmid containing Renilla luciferase gene was cotransfected.

Chromatin Immunoprecipitation (ChIP), RT-PCR

ChIP assay was done essentially as described (Takizawa et al., 2001) by using ChIP assay kit (Upstate). We used the monoclonal 12C11 antibody (gift from Dr. Anderson) and NRSF-P18 antibody (Santa Cruz), as the antibody for NRSF/REST. RT-PCR was performed by using total RNA extracted from HCN A94 cells. 1 mg RNA was used for first-strand cDNA synthesis with SuperScript II (GibcoBRL). PCR primer sequences are available upon request.

Pull-Down Assay with Biotin-Labeled Oligonucleotide

Biotin-labeled RNA was synthesized with an AmpliScribe T7 transcription kit (Epicentre Technologies). Streptavidin-agarose beads (Gibco BRL) were washed with binding buffer (20 mM Tris-HCl, [pH 7.5], 60 mM KCl, 2.5 mM EDTA, and 0.1% Triton X-100) and suspended. While the beads were kept on ice, cell extract was
mixed with 70 mg of biotinylated RNA. After incubation on ice for 10 min, the total volume was adjusted to 1 ml with binding buffer.

Then the sample was transferred to the tube with agarose beads, and the tube was rotated slowly overnight at 4^o C. The beads were washed 5 times with wash buffer (20 mM Tris-HCl, [pH 7.5], 350 mM KCl, and 0.01% NP-40) and resuspended in binding buffer. Proteins were eluted by boiling the beads and were separated by SDS-PAGE.

EMSA

The expression plasmid of cMyc-tagged NRSF/REST (gift from Dr. Anderson) was transfected in 293T cells, and the lysate was incubated with cMyc-antibody in RB buffer (20 mM Tris-HCl, [pH 7.5], 50 mM MgCl₂, 10 mM NaCl, and 1 mM EDTA) overnight at 4^oC. After 1 hr of incubation with Fast Flow protein agarose beads (Upstate) at room temperature, associated proteins were precipitated. The beads were washed 3 times with wash buffer and resuspended in RB buffer. Proteins were eluted by pH 2.0 elution buffer (Upstate) and were neutralized immediately with Tris-Cl pH 8.5 buffer. The protein solution was purified and concentrated with Millipore Centricon (Amicon). Resultant protein solutions were sequentially diluted and each solution was incubated with preannealed and prestained 20 mM oligonucleotides for 30 min at room temperature. Oligonucleotide prestaining was done with SYBR green I (Molecular Probes) for DNA and SYBR green II (Molecular Probes) for RNA. Samples were loaded on 2% Nusieve agarose gel and the image was developed by Eagle Eye II (Stratagene).

Acknowledgments:

We thank David Anderson for the gift of the NRSF/REST antibodies and myc-tagged NRSF/REST constructs. We thank Andrew Fire and Tony Hunter for critical reviewing of our manuscript. We are grateful for the technical assistance of Lynne Moore and Bobbi Miller and to M.L. Gage for editorial comments. K.N. was supported by a JSPS Postdoctoral Fellowship for Research Abroad. J.H. was supported by the Hewitt Foundation for Medical Research. T.K. and K.T. were supported by various grants from AIST and Department of Chemistry and Biotechnology, School of Engineering, The University of Tokyo, Tokyo 113-8656, Japan. F.H.G. was supported by the Lookout Fund, The Christopher Reeves Paralysis Foundation, Michael J. Fox Foundation, and the National Institutes of Health: National Institute on Aging, and National Institute of Neurological Disease and Stroke.

References

Battaglioli, E., Andres, M.E., Rose, D.W., Chenoweth, J.G., Rosenfeld, M.G., Anderson, M.E., and Mandel, G. (2002). REST repression of neuronal genes requires components of the hSWI.SNF complex. J. Biol. Chem. 277, 41038–41045.

Bernstein, E., Caudy, A.A., Hammond, S.M., and Hannon, G.J. (2001). Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409, 363–366.

Bannister, A.J., and Kouzarides, T. (1996). The CBP co-activator is a histone acetyltransferase. Nature 384, 641–643.

Chen, Z.F., Paquette, A.J., and Anderson, D.J. (1998). NRSF/REST is required in vivo for repression of multiple neuronal target genes during embryogenesis. Nat. Genet. 20, 136–142.

Chong, J.A., Tapia-Ramirez, J., Kim, S., Toledo-Aral, J.J., Zheng, Y., Boutros, M.C., Altshuller, Y.M., Frohman, M.A., Kraner, S.D., and Mandel, G. (1995). REST: a mammalian silencer protein that restricts sodium channel gene expression to neurons. Cell 80, 949–957.

Dostie, J., Mourelatos, Z., Yang, M., Sharma, A., and Dreyfuss, G. (2003). Numerous microRNPs in neuronal cells containing novel microRNAs. RNA 9, 180–186.

Eckstein, F., and Lilley, D.M.J. (1996). Catalytic RNA. In Nucleic Acids and Molecular Biology, F. Eckstein and D.M.J. Lilley, eds. (Berlin: Springer-Verlag).

Eddy, S.R. (2001). Non-coding RNA genes and the modern RNA world. Nat. Rev. Genet. 2, 919–929.

Finerty, P.J., Jr., and Bass, B.L. (1999). Subsets of the zinc finger motifs in dsRBP-ZFa can bind double-stranded RNA. Biochemistry 38, 4001–4007.

Fire, A., Xu, S., Montgomery, M.K., Kostas, S.A., Driver, S.E., and Mello, C.C. (1998). Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391, 806–811.

Gage, F.H., Coates, P.W., Palmer, T.D., Kuhn, H.G., Fisher, L.J., Suhonen, J.O., Peterson, D.A., Suhr, S.T., and Ray, J. (1995). Survival and differentiation of adult neuronal progenitor cells transplanted to the adult brain. Proc. Natl. Acad. Sci. USA 92, 11879–11883.

Hoovers, J.M., Mannens, M., John, R., Bliek, J., van Heyningen, V., Porteous, D.J., Leschot, N.J., Westerveld, A., and Little, P.F. (1992). High-resolution localization of 69 potential human zinc finger protein genes: a number are clustered. Genomics 12, 254–263.

Huang, Y., Myers, S.J., and Dingledine, R. (1999). Transcriptional Centri-repression by REST: recruitment of Sin3A and histone deacetylase to neuronal genes. Nat. Neurosci. 2, 867–872.

Hutvagner, G., McLachlan, J., Pasquinelli, A.E., Balint, E., Tuschl, T., and Zamore, P.D. (2001). A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7 small temporal RNA. Science 293, 834–838.

Kallunki, P., Edelman, G.M., and Jones, F.S. (1998). The neural restrictive silencer element can act as both a repressor and enhancer of L1 cell adhesion molecule gene expression during postnatal development. Proc. Natl. Acad. Sci. USA 95, 3233–3238.

Kempermann, G., Gast, D., Kronenberg, G., Yamaguchi, M., and Gage, G.H. (2003). Early determination and long-term persistence of adult-generated new neurons in the hippocampus of mice. Development 130, 391–399.

Koseki, S., Tanabe, T., Tani, K., Asano, S., Shioda, T., Nagai, Y., Shimada, T., Ohkawa, J., and Taira, K. (1998). Factors governing the activity in vivo of ribozymes transcribed by RNA polymerase III. J. Virol. 73, 1868–1877.

Kraner, S.D., Chong, J.A., Tsay, H.-J., and Mandel, G. (1992). Silencing the type II sodium channel gene: a model for neural-specific gene regulation. Neuron 9, 37–44.

Kruhlak, M.J., Hendzel, M.J., Fischle, W., Bertos, N.R., Hameed, S., Yang, X.J., Verdin, E., and Bazett-Jones, D.P. (2001). Regulation of global acetylation in mitosis through loss of histone acetyltransferases and deacetylases from chromatin. J. Biol. Chem. 276, 38307– 38319.

Lau, N.C., Lim, L.P., Weinstein, E.G., and Bartel, D.P. (2001). An abundant class of tiny RNAs with probable regulatory roles in Caenorhabditis elegans. Science 294, 858–862.

Lee, R.C., and Ambros, V. (2001). An extensive class of small RNAs in Caenorhabditis elegans. Science 294, 862–864.

Lee, Y., Jeon, K., Lee, J.T., Kim, S., and Kim, V.N. (2002). MicroRNA maturation: stepwise processing and subcellular localization. EMBO J. 21, 4663–4670.

Lunyak, V.V., Burgess, R., Prefontaine, G.G., Nelson, C., Sze, S.H., Chenoweth, J., Schwartz, P., Pevzner, P.A., Glass, C., Mandel, G., and Rosenfeld, M.G. (2002). Corepressor-dependent silencing of chromosomal regions encoding neuronal genes. Science 298, 1747–1752.

Myers, S.J., Peters, J., Huang, Y., Comer, M.B., Barthel, F., and Dingledine, R. (1998). Transcriptional regulation of the GluR2 gene: neural-specific expression, multiple promoters, and regulatory elements. J. Neurosci. 18, 6723–6739.

Naruse, Y., Aoki, T., Kojima, T., and Mori, N. (1999). Neural restrictive silencer factor recruits mSin3 and histone deacetylase complex to repress neuron-specific target genes. Proc. Natl. Acad. Sci. USA 96, 13691–13696.

Palm, K., Belluardo, N., Metsis, M., and Timmusk, T. (1998). Neuronal expression of zinc finger transcription factor REST/NRSF/XBR gene. J. Neurosci. 18, 1280–1296.

Pasquinelli, A.E. (2002). MicroRNAs: deviants no longer. Trends Genet. 18, 171–173.

Pfeifer, A., Brandon, E.P., Kootstra, N., Gage, F.H., and Verma, I.M. (2001). Delivery of the Cre recombinase by a self-deleting lentiviral vector: efficient gene targeting in vivo. Proc. Natl. Acad. Sci. USA 98, 11450–11455.

Schoenherr, C.J., and Anderson, D.J. (1995). The neuron-restrictive silencer factor (NRSF): a coordinate repressor of multiple neuron-specific genes. Science 267, 1360–1363.

Schoenherr, C.J., Paquette, A.J., and Anderson, D.J. (1996). Identification of potential target genes for the neuron-restrictive silencer factor. Proc. Natl. Acad. Sci. USA 93, 9881–9886.

Shimojo, M., Lee, J.H., and Hersh, L.B. (2001). Role of zinc finger domains of the transcription factor neuron-restrictive silencer factor/repressor element-1 silencing transcription factor in DNA binding and nuclear localization. J. Biol. Chem. 276, 13121–13126.

Stehno-Bittel, L., Perez-Terzic, C., and Clapham, D.E. (1995). Diffusion across the nuclear envelope inhibited by depletion of the nuclear Ca ²⁺ store. Science 270, 1835–1838.

Takizawa, T., Nakashima, K., Namihira, M., Ochiai, W., Uemura, A., Yanagisawa, M., Fujita, N., Nakao, M., and Taga, T. (2001). DNA methylation is a critical cell-intrinsic determinant of astrocyte differentiation in the fetal brain. Dev. Cell 1, 749–758.

Tapia-Ramirez, J., Eggen, B.J., Peral-Rubio, M.J., Toledo-Aral, J.J., and Mandel, G. (1997). A single zinc finger motif in the silencing factor REST represses the neural-specific type II sodium channel promoter. Proc. Natl. Acad. Sci. USA 94, 1177–1182.

Timmusk, T., Palm, K., Lendahl, U., and Metsis, M. (1999). Brain-derived neurotrophic factor expression in vivo under the control of neuron-specific-restrictive silencer element. J. Biol. Chem. 274, 1078–1084.

van Praag, H., Schinder, A.F., Christie, B.R., Toni, N., Palmer, T.D., and Gage, F.H. (2002). Functional neurogenesis in the adult hippocampus. Nature 415, 1030–1034.

Warashina, M., Takagi, Y., Stec, W.J., and Taira, K. (2000). Differences among mechanisms of ribozyme-catalyzed reactions. Curr. Opin. Biotechnol. 11, 354–362.

This article by Tomoko Kuwabara, Jenny Hsieh, Kinichi Nakashima, Kazunari Taira, and Fred H. Gage presents, for the first time, solid evidence for the direct positive effect by nuclear small RNA molecular species on the activation of DNA transcription of specific genes within intact mammalian cells [1-10]. Beyond that important fact, it is also the first demonstration of the RNA-induced conversion of adult neural stem cells to differentiated mature brain neurons. Finally, it raises important questions concerning the molecular induction of embryonic stem cells, as revealed in earlier studies of induction by RNA within sea urchin embryos [11-12].
( Links to RNA and Biological Causality. )

John Frenster, April 7, 2004.      Return to Top.

Additional References:

1. Hovsepian JA, and Frenster JH, "Bioassays of Isolated Nuclear RNA Species as Activators of DNA Transcription".

2. Gottesfeld JM, and Barbas CF III, "RNA as a Transcriptional Activator".

3. Hovsepian JA, and Frenster JH, "RNA-Induced Melting of DNA during Selective Gene Transcription".

4. Coughlin CM, Vance BA, Grupp SA, and Vonderheide RH, "RNA-transfected CD40-activated B cells induce functional T cell responses against viral and tumor antigen targets: implications for pediatric immunotherapy".

5. Iwakiri D, Eizuru Y, Tokunaga M, and Takada K, "Autocrine Growth of Epstein-Barr Virus-Positive Gastric Carcinoma Cells Mediated by an Epstein-Barr Virus-Encoded Small RNA".

6. Ling J, Pi W, Yu X, Bengra C, Long Q, Jin H, Seyfang A, and Tuan D, "The ERV-9 LTR Enhancer is Not Blocked by the HS5 Insulator and Synthesizes Through the HS5 Site Non-Coding, Long RNAs that Regulate LTR Enhancer Function", Nucleic Acids Research, vol. 31, no. 15, pp. 4582-4596 (August 1, 2003).

7. Lanz RB, Chua SS, Barron N, Söder BM, DeMayo F, and O'Malley BW, "Steroid Receptor RNA Activator Stimulates Proliferation as Well as Apoptosis In Vivo".

8. Frenster JH, and Hovsepian JA, "Overshoot in Late Telophase for RNA Re-Programming of Mitotic Chromatin".

9. Buskirk AR, Kehayova PD, Landrigan A, and Liu DR, "In Vivo Evolution of an RNA-Based Transcriptional Activator".

10a. Saha S, Ansari AZ, Jarell KA, and Ptashne M, "RNA Sequences that Work as Transcriptional Activating Regions".

10b. De Carvalho S, "Effect of RNA from Normal Human Marrow on Leukaemic Marrow In-Vivo".

10c. Frenster JH, and Hovsepian JA, "Activator RNA Exchange during Interphase Chromatin Reprogramming".

11. Czihak G, "Evidence for inductive properties of the micromere RNA in sea urchin embryos".

12. Kronenberg LH, and Humphreys T, "Double-Stranded Ribonucleic Acid in Sea Urchin Embryos".

Frenster JH,  "Ultrastructural Probes of Active DNA Sites, and the RNA Activators of DNA".

Top of Page - Euchromatin Network - Current Research - Forums - Other Sites - Future Events -

For Further Information or Feedback:
e-mail:   frenster@euchromatin.net
Phone:   +1 650 367 6483
Fax:   +1 650 364 1773