Long noncoding RNAs in neuronal-glial fate specification and oligodendrocyte lineage maturation
© Mercer et al. 2010
Received: 27 June 2009
Accepted: 5 February 2010
Published: 5 February 2010
Skip to main content
© Mercer et al. 2010
Received: 27 June 2009
Accepted: 5 February 2010
Published: 5 February 2010
Long non-protein-coding RNAs (ncRNAs) are emerging as important regulators of cellular differentiation and are widely expressed in the brain.
Here we show that many long ncRNAs exhibit dynamic expression patterns during neuronal and oligodendrocyte (OL) lineage specification, neuronal-glial fate transitions, and progressive stages of OL lineage elaboration including myelination. Consideration of the genomic context of these dynamically regulated ncRNAs showed they were part of complex transcriptional loci that encompass key neural developmental protein-coding genes, with which they exhibit concordant expression profiles as indicated by both microarray and in situ hybridization analyses. These included ncRNAs associated with differentiation-specific nuclear subdomains such as Gomafu and Neat1, and ncRNAs associated with developmental enhancers and genes encoding important transcription factors and homeotic proteins. We also observed changes in ncRNA expression profiles in response to treatment with trichostatin A, a histone deacetylase inhibitor that prevents the progression of OL progenitors into post-mitotic OLs by altering lineage-specific gene expression programs.
This is the first report of long ncRNA expression in neuronal and glial cell differentiation and of the modulation of ncRNA expression by modification of chromatin architecture. These observations explicitly link ncRNA dynamics to neural stem cell fate decisions, specification and epigenetic reprogramming and may have important implications for understanding and treating neuropsychiatric diseases.
It has recently become evident that the majority of the mammalian genome is transcribed in a developmentally regulated manner producing tens of thousands of interleaved short and long ncRNAs [1–4]. Long ncRNAs, defined as greater than 200 nt in length , exhibit specific temporal and spatial expression patterns in mouse brain , embryonic stem cell differentiation  and T-cell differentiation . NcRNAs are also implicated in neural developmental events and in the pathogenesis of neuropsychiatric diseases . These long ncRNAs are being shown to have a wide range of functions, including roles in the regulation of chromosomal architecture and dynamics, transcription, post-transcriptional processing, RNA editing, RNA trafficking and organelle biogenesis [4, 10–14]. One major emergent function of long ncRNAs is their ability to modulate the epigenetic status of nearby protein-coding genes by recruiting chromatin activator or repressor complexes [4, 12, 15].
Here we present the first systematic examination of long ncRNA expression profiles during differentiation of embryonic forebrain-derived neural stem cells (NSCs). We focused on a subset of cortical GABAergic neurons (GABANs) and oligodendrocytes (OLs) that are initially specified from sonic hedgehog (Shh)-responsive, Nkx2.1-expressing bipotent neuronal-OL progenitor (N/OP) species within the ventromedial forebrain. We have previously reported that these N/OPs undergo long-distance tangential migration to cerebral cortex where they sequentially give rise to GABANs and OLs [16–21]. We employed custom designed microarrays to examine expression of noncoding and protein-coding transcripts during these neural developmental transitions, finding dynamic expression of ncRNAs associated with key neural development genes. We also observed differential expression of specific ncRNAs after treatment of oligodendrocyte progenitors (OLPs) with trichostatin A (TSA), a histone deacetylase (HDAC) inhibitor that prevents progression of OLPs into post-mitotic oligodendrocytes (PMOs) by suppressing oligodendrocyte-specific genes and inducing epigenetic reprogramming [22–26]. These developmental changes underscore the importance of epigenetic regulation in the elaboration of neural cell types and imply functional relationships between chromatin remodeling and selective deployment of ncRNAs.
We found that 9% (332 out of 3,659) of probes targeting ncRNAs were expressed above background, of which 169 (5%) were significantly differentially expressed (B-statistic > 1; Figure 1) at one or more developmental stages compared to the NSC control [Additional file 2). These data reveal that a smaller percentage of ncRNAs are differentially expressed relative to protein-coding mRNAs. While this difference may be due to relatively poor annotation of the long noncoding transcriptome, it is consistent with previous reports in other developmental models [1, 27] and may reflect the fact that ncRNAs are primarily regulatory in nature requiring only subtle changes in expression to carry out their functional roles . We also validated our microarray findings with QRT-PCR and found an 87% correlation between the gene expression profiles from the microarray data and QRT-PCR results [Additional file 3].
Long ncRNAs are often organized as part of complex interleaved transcriptional networks that include protein-coding genes [1, 2]. A number of studies have shown that the association of long noncoding transcription with protein-coding genes has functional relevance, often with the ncRNA regulating expression of its protein-coding counterpart via epigenetic modifications and/or transcriptional co-activation/repression [14, 15, 29–31]. Therefore, we identified all ncRNAs expressed in our developmental models that are transcribed antisense to, bi-directionally, or from within the introns of overlapping or adjacent protein coding genes [Additional file 4]. In addition, we identified pairs of protein-coding RNAs and ncRNAs whose genomic loci are positionally conserved between mouse and human [Additional file 4] because this genomic conservation may also indicate a functional relationship between the two transcripts. In many cases, the associated protein-coding genes were critical neuro-developmental genes. As an illustrative example, the downregulation of a ncRNA (AK053922) is associated with the Gli3 locus in all progeny of N/OPs, whereas Gli3 was upregulated in N/OPs but downregulated in all progeny of N/OPs. This relationship may be of significant interest given that Gli3 has been shown to act as a bifunctional transcriptional switch that can either repress or activate Shh signaling  to help specify distinct neuronal cell types .
Many ncRNAs contain conserved secondary structures that are crucial to their function . These conformational features may be bound by proteins , processed into smaller regulatory ncRNAs  or even fulfill intrinsic catalytic functions . To provide support for the proposed functionality of the expressed ncRNAs, we looked for evidence of RNA secondary structures within differentially expressed ncRNA transcripts using the folding program RNAz . We found 74 expressed ncRNAs that contained high confidence predicted conserved secondary structures (RNAz P > 0.9) [Additional file 5] such as stem-loops [Additional file 6], known to be present in miRNA precursors, small nucleolar RNAs (snoRNAs) and other RNAs with previously characterized functions.
We then combined these genomic analyses to identify candidate ncRNAs for further functional analysis. These are discussed in more detail below in the context of neuronal and oligodendroglial lineage commitment, neuronal-glial lineage switching and progressive stages of oligodendrocyte lineage elaboration.
A subset of nestin-immunoreactive NSCs within the ventral forebrain give rise to bipotent N/OPs following exposure to a gradient of Shh signaling [17, 20] which are characterized by expression of two bHLH transcription factors (Olig2 and Mash1) and their potential to give rise to GABANs and OLs but not astrocytes [16, 20]. Our microarray analysis revealed that the initial fate restriction of regional forebrain NSCs to bipotent N/OPs was associated with relatively modest changes in gene expression profiles, possibly reflecting the relative similarities between these developmental cell types. Nonetheless, there were differences in factors that may be important for promoting NSC lineage restriction and abrogating self-renewal and maintenance programs, including 65 upregulated and 167 downregulated mRNAs [Additional file 1]. Not surprisingly, some genes previously implicated in neurogenesis and gliogenesis exhibited dynamic changes in mRNA expression including transcription factors (52% of upregulated genes; p < 0.01) and cell-surface receptor linked signal transduction pathways (21% of downregulated genes; p < 0.01). For example, we found upregulated mRNAs included genes such as Mash1, an important transcription factor that promotes lineage restriction of NSCs into N/OPs [38–40].
We also identified dynamically expressed ncRNAs that may fulfill functional roles in NSC fate restriction. Similar to mRNAs, relatively few changes in ncRNA expression were observed during the transition from NSCs to N/OPs, with 17 ncRNAs exhibiting differential expression (8 upregulated and 9 downregulated) [Additional file 2]. These included a previously described ncRNA, Gomafu, shown to be localized to a novel subnuclear domain in a distinct subset of differentiating neurons in the mouse nervous system , which was exclusively downregulated in N/OPs but upregulated in all subsequent stages of OL lineage specification and maturation [Additional file 7]. Furthermore, in situ hybridization (ISH) in the adult mouse brain showed Gomafu expression in specific neuronal populations and nuclear localization within Purkinje cells [Additional file 7]. Our results show that Gomafu is not only expressed during neurogenesis but also during OL lineage specification.
Bipotent N/OPs can differentiate into either GABANs or OLPs in response to different neural cytokines . We therefore examined changes in gene expression during the elaboration of GABANs to identify mRNAs and ncRNAs that may be involved in this type of neuronal lineage specification. We observed upregulation of 483 mRNAs, which were enriched for genes involved in nervous system development (17.3%; p < 0.03), neuronal function (9.88%, p < 0.05) and neurogenesis (10.67%, p < 0.05). We also identified 330 downregulated mRNAs [Additional file 1]. Interestingly, these downregulated mRNAs were enriched for predicted targets of let-7b (13.2% p < 0.01) and mir-135 (11.3%; p < 0.01), possibly being subject to repression by miRNAs. Indeed, both let-7b and mir-135 have previously been implicated in promoting neurogenesis [48, 49]. In addition, we identified 56 ncRNAs that are upregulated during GABAN neurogenesis, including a number of examples such as Gtl2, Rian, Evf2, and Copg2as that were previously shown to exhibit neuron-enriched expression profiles in the adult mouse brain . In contrast, we identified only 8 ncRNAs that were downregulated during GABAergic neurogenesis [Additional file 2].
We distinguished ncRNAs that may be involved specifically in GABAN lineage commitment by selecting ncRNAs upregulated during GABAN differentiation but downregulated during OL differentiation [Additional file 9]. We identified 4 ncRNAs, including an intergenic ncRNA transcript AK044422 that overlaps a highly conserved brain-specific miRNA, miR-124a. MiR-124a accounts for nearly half of all brain miRNAs  (Figure 2) and helps to regulate neuronal specification and differentiation [51–53]. One mechanism by which miR-124 was shown to promote neuronal differentiation is by repression of Ptbp1, a negative regulator of alternative splicing, resulting in activation of neuron-specific alternative splicing events . Indeed, we found that Ptbp1 and AK044422 exhibited complementary expression profiles. However, the AK044422 transcript may itself function intrinsically as a full-length ncRNA, rather than simply acting as a precursor. The AK044422 transcript is subject to post-transcriptional modifications including splicing and polyadenylation, contains a number of high confidence predicted stem-loop structures (Fig 2.), and is highly conserved with homologous transcripts present in humans (AK091593, AL832535) and rats (BF402302). Indeed, the AK044422 transcript was recently identified in a genome wide screen for functional lncRNAs, termed lincRNAs, that exhibit a distinct chromatin signature indicative of transcriptional regulation and high levels of expression . Furthermore, ISH in adult mouse brain revealed that the full-length AK044422 transcript is expressed in the mature central nervous system (Figure 2), suggesting an involvement in neurogenesis independent of mir-124a.
Because a large number of developmental gene-related enhancers have been shown to drive transcription in the developing forebrain during early neurogenesis , we examined ncRNAs expressed in GABANs that are associated with highly conserved enhancers. We identified 3 ncRNAs that were associated with highly conserved elements driving transcription in the forebrain [Additional file 10]. Two of these ncRNAs were associated with ultraconserved enhancers at the Dlx genomic loci, one of which, Evf-2, was previously shown to regulate binding of the Dlx2 transcription factor to the ultraconserved enhancer to activate the expression of the encompassed Dlx6 gene during neurogenesis .
We also identified a similar ncRNA antisense to the Dlx1 gene. Both Dlx1 and the antisense ncRNA, Dlx1AS, are upregulated during GABAN differentiation but downregulated during OL differentiation [Additional file 10]. These observations are consistent with published data showing that Dlx1 and Dlx2 promote GABAN lineage specification by suppressing the important oligogenic bHLH transcription factor, Olig2  and are necessary for tangential migration of ventral forebrain-derived progenitor cells [17, 56–58]. Dlx1AS is transcribed from an ultraconserved enhancer that drives expression in the midbrain, forebrain and olfactory system during development and in regions associated with neurogenesis in the adult mouse brain . The third ncRNA, AK005755, which is also transcribed from an ultraconserved enhancer driving expression in the developing forebrain, has not been previously described and is not associated with any annotated developmental genes. Nonetheless, expression of this spliced transcript during GABAergic neurogenesis, coupled with its extremely conserved enhancer, suggests it plays a role in the developing forebrain.
Application of PDGF-AA to N/OPs for 4 days in vitro was sufficient to promote the elaboration of OLPs (Figure 1B). We confirmed by immunofluorescence microscopy that N/OPs express the corresponding receptor (PDGFRα) expression prior to and independent of the application of PDGF-AA [Additional file 11]. The developmental transition of N/OPs to OLs was associated with the upregulation of 665 mRNAs including a number of OL-selective genes, such as Olig1 and Olig2, and the downregulation of 650 mRNAs [Additional file 1]. We also found that 42 ncRNAs were downregulated, whereas 58 ncRNAs were upregulated, six of which exhibited highly correlated expression profiles to Olig1, having a distinct peak during OLP differentiation (Pearsons R2 > 0.9) [Additional file 12; Additional file 13].
To identify ncRNAs with specific roles in OL lineage commitment, we selected ncRNAs that were upregulated in OLPs but downregulated in GABANs. This analysis identified only one ncRNA, a novel 1.2 kb transcript (AK079380) organized bi-directionally to the Sox8 gene (Figure 2). Sox8 is a member of the SRY-box transcription factor family, many of which play key roles in OL lineage elaboration, including Sox9 and Sox11 for OL specification and Sox8 and Sox10 for progressive OL maturation [59, 60]. The ncRNA we identified, which we termed Sox8OT (Sox8 Opposite Transcript), exhibited concordant expression (Pearsons R2 > 0.9) to the Sox8 mRNA. Furthermore, a positionally conserved homologous Sox8OT transcript is also present in humans (BC098409) and rats (BM382844), suggesting the bi-directional relationship between Sox8 and Sox8OT is relevant. The bidirectional organization of the Sox8 gene and the Sox8OT ncRNA suggests that these transcripts are subject to common modes of regulation, possibly by sharing a promoter sequence. Indeed, recently published genome-wide maps of chromatin modifications  indicate a shared domain subject to dynamic modifications during the transition from embryonic stem to neuronal progenitor cells [Additional file 14]. Together, these data suggest Sox8OT functions in OL lineage commitment, possibly by regulating the expression of the associated Sox8 gene. As an alternative strategy to identify ncRNAs with specific roles in OL lineage commitment, we identified ncRNAs that were differentially expressed only in OLPs, and found that 16 were downregulated while 14 were upregulated including Dleu2, a previously described ncRNA overlapping but transcribed in the opposite direction of the Trim13 (RFP2/LEU5) gene whose expression is upregulated during early OL maturation and terminal differentiation .
The withdrawal of OL cytokines (PDGF-AA) from OLP culture conditions results in cell cycle exit (Figure 1A), though the actions of environmental factors such as triiodothyronine, cyclin dependent kinase inhibitors (CDKIs: p27KIP1), inhibitory HLH transcription factors (ID4) and other transcription factors (YY1) that participate in the combinatorial molecular programs regulating this process [23, 63–66]. In PMOs, we identified 768 upregulated mRNAs including genes involved in lipid metabolic processes (11.4%, p < 0.02), such as Mbp, a structural component of the myelin sheath . We also observed downregulation of 562 mRNAs including genes involved in cell division (32.7%, p < 0.01), mitosis (23.4%, p < 0.01), DNA replication (13.8%, p < 0.01), and chromosome organization and biogenesis (8.65%, p < 0.01). Similarly, OLP cell cycle exit was accompanied by the upregulation of 38 ncRNAs and the downregulation of 26 ncRNAs. Many of the ncRNAs are associated with genomic loci encompassing protein-coding genes with previously described functions in development and maturation. For example, we found a highly conserved ncRNA, AK019302, hosted within the intron of Bai3, which was downregulated in PMOs. Bai3, which exhibits a similar expression profile, is a developmentally expressed angiostatic gene that is downregulated in the early phases of brain ischemia and in high-grade gliomas  and has been associated with mental retardation  and schizophrenia .
OL terminal differentiation is associated with the outgrowth of cytoplasmic processes and the accelerated production of myelin , and this process is regulated by neural cytokines, various classes of transcription factors, CDKIs (p21CIP1), and epigenetic factors [72, 73]. In MYOs, we found 733 upregulated mRNAs that were enriched for genes involved in lipid metabolic processes (8.7%; p < 0.01), biopolymer metabolic processes (21.2%, p < 0.01), and actin cytoskeleton organization (8.3%, p < 0.01), including a dramatic increase in the expression of myelin genes such as Plp1, Mbp and Mal. In addition, we found 51 ncRNAs that were upregulated during OL terminal differentiation and myelination, including 8 that were broadly upregulated during progressive stages of OL lineage elaboration, exhibiting a highly correlated (Pearsons R2 > 0.9) [Additional file 9; Additional file 12] expression profile with Mog, an important glycoprotein found on the surface of MYOs . In contrast, we found that 35 ncRNAs were downregulated, including 27 ncRNAs that exhibited a highly correlated expression profile (Pearsons R2 > 0.9) [Additional file 12; Additional file 13] to Melk, a factor important for promoting NSC proliferation [1, 75] that was downregulated during OL lineage elaboration.
A variety of ncRNAs were broadly upregulated in OLPs, PMOs and MYOs including AK141895 and AK133808, associated with the Slc44a1 and Rab11b genomic loci, respectively. In these cases, both the ncRNAs and the protein-coding transcripts were strongly upregulated during OL lineage elaboration suggesting that they may function coordinately to promote appropriate OL maturation and expression of myelin proteins. We similarly identified a ncRNA antisense to Ecsit, an important node between BMP and Toll-Like Receptor (TLR) pathways that integrates morphogenetic signaling with pathogen recognition and innate immunity during development [76, 77]. The ncRNA, AK050588, is upregulated in GABANs, OLPs and MYOs, in contrast to Ecsit, which is downregulated in these species. The observed inverse expression pattern between AK050588 and Ecsit is similar to a correlation we recently observed in embryonic stem cell differentiation  and suggests coordination between the expression profiles of this pair of transcripts.
We employed QRT-PCR to examine the expression of ncRNAs identified within this study as strong candidates for functional roles in OL or neuronal differentiation. Broadly, we found parallel changes in cellular phenotype and alterations in ncRNA expression profiles induced by TSA. In addition to examples described above, we included other ncRNAs associated with neural developmental genes that exhibited representative developmental expression profiles. Almost all selected ncRNAs (14 of 15) exhibited significant (p < 0.05) changes in expression at 24 or 48 h (Figure 3). The majority of these changes were sustained or developed during later stages (i.e., at 48 h), suggesting the expression of these ncRNAs is subject to regulation by mechanisms of histone deacetylation.
While we observed both increased and decreased expression of long ncRNAs, the majority was downregulated, often dramatically, supporting the putative role of these ncRNAs in contributing to the progressive differentiation of OLs (Figure 3). Downregulated ncRNAs included those associated with protein-coding genes involved in progressive OL differentiation. In addition to those previously described including Sox8OT, Gomafu, AK014263 (hosted in an intron of Csmd1) and AK019302 (hosted in an intron of Bai3), we also observed dramatic downregulation of ncRNA AK020939 which is hosted within an intron of Lhfpl3, a member of the tetraspanin proteins which promote proliferation, migration and differentiation of OLs [82, 83].
Recent evidence has shown that histone deacetylation is responsible for restricting neuronal gene expression, with the inhibition of HDACs inducing neuronal differentiation . Therefore, it is of interest that a number of candidate ncRNAs exhibit an increase in expression in response to HDAC inhibition, possibly suggesting a contribution to neuronal rather than OL differentiation. For example, the previously described ncRNA Gtl2 was upregulated after TSA treatment. Indeed, within his study we observed upregulation of Gtl2 during GABAN differentiation. Gtl2 is expressed in restricted domains of the developing forebrain including ventral telencephalon, neocortex, hippocampus and diencephalon, suggesting roles in regional neuronal development and differentiation . Finally, we found that Neat1 transcription exhibited complex transcriptional dynamics after TSA treatment, being strongly upregulated after TSA treatment under stochastic (probabilistic) oligodendroglial specification conditions, whereas it was downregulated after TSA treatment under the influence of instructive (inductive) signals for OL lineage elaboration (see Methods; Figure 3 and Additional file 15).
In this study, we utilized a custom-designed microarray platform to probe the transcription of both protein-coding and noncoding transcripts during ventral forebrain-derived NSC lineage restriction, neuronal and oligodendrocyte (OL) lineage specification, neuronal-glial fate transitions, and progressive stages of OL lineage elaboration including myelination. Expression profiles for mRNAs in our study were similar to previously reported findings [62, 86] showing modulation of key signaling pathways, transcription factors, and epigenetic regulators that help to define the molecular codes governing neural lineage elaboration. Additionally, we also observed expression of components of the myelin sheath.
Although we examined protein-coding genes, we focused primarily on characterizing the expression profiles of long ncRNAs, which are being increasingly appreciated in the regulation of developmental gene expression programs [3, 4, 7, 28]. Previous studies of neural differentiation have reported expression profiles for only a small number of long ncRNAs, such as Nkx2.2AS and the imprinted H19 ncRNA [62, 86, 87]. In contrast, we examined a large spectrum of long ncRNAs and found dramatic changes in expression that are unique to specific stages of cellular differentiation. Many of these ncRNAs are transcribed from complex genomic loci that encompass protein-coding genes with defined roles in neuronal and glial lineage elaboration, and these pairs of ncRNAs and associated protein-coding genes often exhibited coordinated expression during developmental transitions. Therefore, we suggest that these ncRNAs may influence the expression of the associated protein-coding genes similar to previously characterized examples such as Nkx2.2AS , HOTAIR , p15AS , p21AS  and Evf2 . For example, many ncRNAs were associated with transcription factors that were themselves differentially regulated during OL differentiation [62, 86]. Long ncRNAs, such as those identified in this study, may therefore contribute to the combinatorial transcriptional codes involved in lineage specification and terminal differentiation including myelination.
The coordinated expression of long ncRNAs with critical neural developmental protein coding genes suggests that both are subject to shared regulatory mechanisms and that long ncRNAs are integrated into complex environmentally-mediated neural and glial developmental gene expression programs. We, therefore, considered whether the well-documented ability of epigenetic mechanisms to promote decisive changes in neuronal and glial gene expression programs also includes the capacity to regulate long ncRNA expression. Indeed, treatment with the HDAC inhibitor, TSA, effectively arrests the developmental progression of OLPs into PMOs  and also affects the expression of long ncRNAs, suggesting that these are not only regulated by epigenetic modifications but also these transcripts are incorporated into a broader lineage specific developmental gene expression program. Furthermore, HDAC inhibition also has differential effects on certain ncRNAs in an environmentally responsive manner. For example, treatment of OLPs with TSA in the presence (inductive) or absence (stochastic) of CNTF results in opposing profiles of Neat1 expression, implying that long ncRNAs might be deployed in a context-specific manner to dynamically modulate seminal fate decisions within the OL lineage. Although the use of HDAC inhibitors has been essential for interrogating the epigenetic mechanisms that govern OL lineage maturation, the cellular effects of TSA administration are complex and our understanding of the mechanisms mediating ncRNA expression are still evolving. Nonetheless, long ncRNAs may represent important targets for regenerative therapies in the CNS because they exhibit both exquisite site-selective as well as genome wide effects [5, 13] associated with dynamic OL developmental transition states characterized by considerable epigenetic flux [22–24, 26, 78, 79, 81].
This study is the first systematic examination of the expression of long ncRNAs in forebrain-mediated NSC maturational processes, specifically OL differentiation and development. It is also the first demonstration that long ncRNAs, many of which are involved in regulating epigenetic processes, are also themselves subject to regulation by epigenetic processes. It adds to a growing consensus that not only are large portions of the mammalian genome transcribed  but also expressed in a regulated manner . The specific expression of ncRNAs has been recently observed in other developmental programs and comparisons between these studies reveal both common and unique features [7, 88]. Because the long noncoding transcriptome is still relatively poorly annotated and only partially interrogated by this study, it seems likely that many more long ncRNAs will be shown to be expressed during development and their importance for neuronal and glial biology will be more fully understood and appreciated. These studies will have particular relevance in the future as the role of noncoding transcripts in neurological disease states becomes increasingly recognized .
Early embryonic (E14) multipotent and more lineage-restricted progenitor species derived from the ventral forebrain were plated at clonal density (1000 cells/mL) and propagated in serum free media (SFM) containing specified factors for various time intervals, and subsequently examined by immunofluorescence microscopy to define neural lineage profiles, expression of selected transcription factors (Figure 1) and to detect cellular proliferation (BrdU, 10 μM, added 3 h prior to fixation) as previously described [20, 90–92]. Multipotent progenitor clones were generated by application of basic fibroblast growth factor (bFGF, 10 ng/mL) and were cultured 7 days in vitro (DIV) and were passaged twice to enrich the self-renewing NCSs. Subsequently, NSC clones were dissociated by trypsinization and were exposed to bFGF and the N-terminal active form of Shh (N-Shh, 50 ng/ml) for 2 DIV to generate N/OP clones . GABANs were generated from N/OPs by the application of BMP2 (10 ng/mL) for 2 DIV. OLPs were generated from N/OPs by the application of PDGF AA (10 ng/mL) for 4 DIV [Additional file 11] [20, 71]. PMOs and MYOs were generated from on 6 DIV late-stage OLPs by withdrawal of PDGF and concurrent addition of CNTF (20 ng/mL) for an additional 2 DIV and 4 DIV respectively [65, 71]. We followed institutional IACUC guidelines for experiments in which primary mouse tissue specimens were used (Einstein Institute for Animal Studies protocol # 20080709).
The application of TSA (10 ng/mL, Sigma) to OLs was done at the time of withdrawal of PDGF in the presence (inductive oligodendrogliogenesis) or absence (stochastic oligodendrogliogenesis) of CNTF that is necessary to promote cell cycle exit and OL differentiation. In addition to study of the four developmental stages encompassing oligodendrogliogenesis (N/OPs → OLPs → PMOs → MYOs), we also analyzed the expression patterns of selected ncRNAs at the time of PDGF withdrawal without CNTF or TSA addition (T0), and at 24 h and 48 h after TSA treatment in the presence and absence of CNTF [72, 93]. We also compared these values to data obtained from OLs not treated with TSA at 24 h and at 48 h .
Antibodies against the neuroepithelial marker, nestin (mIgG1, 1:200, Pharmingen) were used to label both NSCs and N/OPs. N/OPs were distinguished from NSCs by the expression of the bHLH factors, Olig2 (goat IgG, 1:500, R&D) and Mash1 (mouse, mIgG1, 1:250, Pharmingen). Progressive stages of OL lineage elaboration were documented by the expression of the OL progenitor marker, O4 (mIgM, 1:600 Sigma), the PMO marker, GC/O1 (mIgM, 1:300 Chemicon) and the MYO marker, MBP (mIgG2b, 1:500 Sternberger Monoclonals Inc.). GABANs were identified by immunoreactivity against GABA (rabbit IgG, 1:1000 Sigma). Expression of PDGFRα on N/OPs was assessed using an antibody against PDGFRα (goat IgG, 1:300, R&D Systems). Nuclear and cell cycle-associated reagents (BrdU, mIgG1, 1:400, Novocastra; Hoechst 33342, 2 μg/mL) were utilized to assess cell proliferation.
To generate the various neural stem, progenitor and more lineage-restricted neuronal and OL lineage species required for immunocytochemical analysis, the following growth factor preparations were utilized: recombinant bFGF (Collaborative Biomedical Products), N-Shh, PDGF-AA and CNTF (R&D Systems) and BMP2 (gift from Genetics Institute).
RNA from mouse tissues and cell cultures was purified using Trizol (Invitrogen) or RNeasy Mini Kit (Qiagen) and treated with DNase I (Invitrogen), according to the protocols provided by the manufacturers. The quality of purified total RNA samples was assessed with an RNA 6000 Nano assay kit using the Agilent 2100 Bioanalyzer (Agilent Technologies) according to the manufacturer's instructions. For microarray experiments, RNA was amplified and labeled using the Amino Allyl Message Amp II kit (Ambion) following the instructions provided by the manufacturer. Amplified aRNA from each time point, as well as a reference pooled sample comprising a mixture of RNA from all time points, was labeled with either Cy3 or Cy5 monoreactive dyes (Amersham Biosciences) according to the MessageAmp II protocol (Ambion). The quality and quantity of amplified RNA samples were assessed using the Agilent 2100 Bioanalyzer as described above.
The microarrays contained 22,038 65-mer oligonucleotide probes from the Mouse OligoLibrary (Compugen, San Jose, CA, USA) and 2,118 70-mer oligonucleotide probes that were designed to target ncRNAs, including known mouse pre-miRNAs from miRBase , longer mouse ncRNAs from RNAdb , and "high confidence" ncRNAs identified from the FANTOM3 project . The custom 70-mer probes were printed alongside Mouse OligoLibrary probes on Power Matrix slides (Full Moon BioSystems, Sunnyvale, CA, USA) at the SRC Microarray Facility (University of Queensland, Brisbane, Australia). The quality of the print run was verified by hybridizing random 10-mer oligonucleotides to the 1st and last slides of the run. The array design is available from the ArrayExpress Data Warehouse (EMBL-EBI; ArrayExpress Accession Number A-MEXP-1070).
Labeled RNAs from each of the five developmental stages was hybridized with the NSC sample to individual microarrays. Three biological replicates were performed for each time point. Blocking, hybridization and washing was performed according to the manufacturer's instructions (Full Moon BioSystems, Sunnyvale, CA, USA). Slides were scanned at 5 μm resolution using a DNA microarray scanner (Agilent Technologies). Feature extraction was performed using ImaGene software (BioDiscovery, El Segundo, CA, USA), with manual grid adjustment and auto-spot finding and segmentation. Data was exported from ImaGene as text files, then uploaded and analyzed using the Linear Models for Microarray Data (LIMMA) software package via the R Project for Statistical Computing http://www.r-project.org. Data was background-corrected, normalized both within and between arrays , and differential expression analysis was performed by fitting a linear model of the data to the experimental design matrix and then calculating Bayesian statistics (B statistics; posterior log odds) adjusted for multiple testing using Benjamini-Hochberg analysis . Raw and processed microarray data is available at the ArrayExpress Data Warehouse (EMBL-EBI; ArrayExpress Accession Number E-TABM-874).
The primers were designed and synthesized (Invitrogen) to amplify selected ncRNA sequences [Additional file 16].
Quantitative real-time PCR (QRT-PCR) was performed according to manufacturer's protocols (Qiagen). QRT-PCR reactions were performed in a 96-well plate, each reaction consisted of a 25 μL mixture including 12.5 μL of 2× QuantiFast SYBR Green PCR Master Mix (Qiagen), 1 μM of forward and reverse sequences of the corresponding primers and 40 ng of cDNA. Mouse 18S ribosomal RNA was used as an internal control to ensure accuracy of the results. QRT-PCR amplifications were performed using the ABI PRISM 7000 Sequence Detection System (Applied Biosystems). Amplification conditions were as follows: 5 min at 95°C to activate the HotStarTaq Plus DNA Polymerase followed by 40 cycles of 10 s at 95°C and 30 s 60°C. QRT-PCR conditions were optimized by adjusting the primer concentrations and the quantity of cDNA. A dissociation curve was performed to ensure the validity of each specific PCR product. QRT-PCR results were analyzed using Relative Expression Software Tool  software http://rest.gene-quantification.info/.
Although the Mouse OligoLibrary probe set was predominantly designed to recognize known or putative protein-coding transcripts, several thousand probes targeted miscellaneous cDNAs and ESTs whose coding status was not well characterized at the time this commercial probe set was first produced. To update the annotation of these probes and to clarify whether they targeted protein-coding or noncoding regions, a computational pipeline was designed to re-annotate the entire probe set. Sequences for all probes were mapped to the February 2006 (NCBI Build 36) assembly of the mouse genome using BLAT  (parameters: minScore = 50, minIdentity = 99, stepSize = 5, tileSize = 11, ooc = 11.ooc). Probes that could not be reliably mapped were excluded from the study. Targeted transcripts were then defined as protein-coding and noncoding as described previously . The genomic context of ncRNAs (relative to protein-coding genes) was also determined as described previously .
The secondary structural composition of expressed ncRNAs was determined by intersecting their chromosomal positions with those of the RNAz structural predictions made across the entire mouse genome (using confidence threshold levels of P > 0.5 and P > 0.9), which were performed as previously described [6, 34]. High confidence predicted RNA secondary structures were rendered with CONTRAfold .
The Gene Ontology project http://www.geneontology.org/ provides a controlled vocabulary to functionally annotate genes. By studying GO terms associated with a large set of genes, one obtains useful information about the types of genes represented. Lists of differentially expressed protein-coding genes were uploaded into FastiGO+ http://babelomics.bioinfo.cipf.es. Statistically over-represented GO terms in the biological process and molecular function and microRNA target enrichments were obtained by applying a p-value cutoff as indicated, correcting for multiple testing with Benjamini False Discovery Rate.
TRM was supported by an Australian Postgraduate Award. MED was funded by a Foundation for Research, Science, and Technology, New Zealand Fellowship. JSM is supported by an Australian Research Council Federation Fellowship (FF0561986), as well as by the University of Queensland and the Queensland State Government. MFM is supported by grants from the National Institutes of Health (NS38902, MH66290, HD01799), as well as by the F.M. Kirby, the Rosanne H. Silbermann, the Alpern Family, the Mildred and Bernard H. Kayden and the Roslyn and Leslie Goldstein Foundations.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.