Chromatin architecture in stem cells determines the pattern of gene expression and thereby cell identity and fate. The chromatin-modifying agents trichostatin A (TSA) and 5-Aza-2′-deoxycytidine (AzaC) affect histone acetylation and DNA methylation, respectively, and thereby influence chromatin structure and gene expression. In our previous work, we demonstrated that TSA/AzaC treatment of neurosphere cells induces hematopoietic activity in vivo that is long-term, multilineage, and transplantable. Here, we have analyzed the TSA/AzaC-induced changes in gene expression by global gene expression profiling. TSA/AzaC caused both up- and downregulation of genes, without increasing the total number of expressed genes. Chromosome analysis showed no hot spot of TSA/AzaC impact on a particular chromosome or chromosomal region. Hierarchical cluster analysis revealed common gene expression patterns among neurosphere cells treated with TSA/AzaC, embryonic stem (ES) cells, and hematopoietic stem cells. Furthermore, our analysis identified several stem cell genes and pluripotency-associated genes that are induced by TSA/AzaC in neurosphere cells, including Cd34, Cd133, Oct4, Nanog, Klf4, Bex1, and the Dppa family members Dppa2, 3, 4, and 5. Sox2 and c-Myc are constitutively expressed in neurosphere cells. We propose a model in which TSA/AzaC, by removal of epigenetic inhibition, induces the reactivation of several stem cell and pluripotency-associated genes, and their coordinate expression enlarges the differentiation potential of somatic precursor cells.
Disclosure of potential conflicts of interest is found at the end of this article.
Stem cells are functionally characterized by their high self-renewal activity and their multilineage differentiation potential [1, 2]. In multicellular organisms, specific stem cell types with distinct developmental potentials occur during development. Transient pluripotent cells, which can differentiate into derivatives of all three germ layers, are generated during blastocyst development. Adult stem cells, developing at later stages, are more restricted in their potential, since they can differentiate into progenitors and mature effector cell types of only one stem cell system. Adult stem cells have been identified in a variety of tissues in the adult organism and are important for lifelong tissue homeostasis and repair . Adult stem cells show specific functional properties and express specific gene patterns that are distinct from those of pluripotent embryonic stem (ES) cells and terminally differentiated cells. Pluripotency is a transient feature that becomes increasingly restricted during development . ES cell self-renewal and differentiation are regulated by specialized regulatory circuitry, involving transcriptional regulators, including OCT4, SOX2, NANOG, and Polycomb group proteins, and various signaling pathways (e.g., STAT3, BMP, and WNT signaling) [2, , , –6]. Recent studies have demonstrated that fetal and adult cells can be reprogrammed to pluripotency by ectopic expression of four genes (Oct4, Sox2, c-Myc, and Klf4), referred to as induced pluripotent stem (iPS) cells [7, , –10]. iPS cells acquire an epigenetic state similar to that of ES cells and can form viable chimeras and contribute to the germ line.
Over the past few years, several studies have reported on an enlarged developmental potential of adult stem cells that extends to tissues other than the tissue of origin [11, –13]. For example, hematopoietic stem cells (HSC) were reported to generate muscle, liver, and neural cells. However, many of these initial observations were not reproduced and are now being met with skepticism. Cells were found to adopt the phenotype of other cells, for example, due to spontaneous cell fusion [13, , –16] and/or epigenetic alterations . Furthermore, epigenetic dysregulation is also observed during stem cell aging .
In a previous study, we observed that transient epigenetic modification of neurosphere cells by the chromatin-modifying agents trichostatin A (TSA) and 5-Aza-2′-deoxycytidine (AzaC) make them acquire hematopoietic activity in vivo, albeit with low frequency . The TSA/AzaC-induced hematopoietic activity was long-term, multilineage, and transplantable. Similarly, in vitro-cultured human CD34+ hematopoietic stem/progenitor cells retain the ability to repopulate immunodeficient mice if pretreated with AzaC and histone deacetylase inhibitors [20, 21]. TSA and AzaC, by inhibiting histone deacetylation and DNA methylation, respectively, generate an altered, transcriptionally active chromatin structure that is expected to affect global gene expression and thus to influence cell fate decisions. Frequently, treatment with both chromatin-modifying agents is required for reactivation of epigenetically silenced genes .
To obtain insight into the altered gene expression program induced by TSA and AzaC, we performed global gene expression analysis of neurosphere cells treated with TSA, AzaC, or TSA plus AzaC (TSA/AzaC). Our analysis revealed that TSA/AzaC induced transient expression of several stem cell and pluripotency-associated genes, possibly by removal of epigenetic inhibition. We suggest that the reactivation of such genes has an impact on the developmental competence of TSA/AzaC-treated neurosphere cells.
Materials and Methods
Cells and Cell Culture
Neurospheres were established from E14.5 forebrain of wild-type (wt) C57BL/6 or Bcl-2 transgenic mice (backcrossed on C57BL/6) [19, 23]. Neurospheres were cultured in Neural Basal Medium with B27 supplement (2% vol/vol), 100 U/ml penicillin/streptomycin, 2 mM l-glutamine (all from Gibco-BRL, Gaithersburg, MD, http://www.gibcobrl.com), 20 ng/ml basic fibroblast growth factor, and 20 ng/ml epithelial growth factor (both from Cell Concepts, Umkirch, Germany, http://www.cellconcepts.de) as described . Single-cell suspensions (1 × 105 cells per milliliter) were prepared by enzymatic dissociation with Accumax (PAA Laboratories, Linz, Austria, http://www.paa.at), and cells were treated with 150 nM TSA and/or 500 nM AzaC (both from Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) for 2 days or were left untreated.
To induce neural differentiation, single-cell suspensions of neurosphere cultures were seeded on polyornithine-coated coverslips at 1 × 106 cells per slide and cultured for up to 13 days under differentiation conditions in Neural Basal Medium containing B27 supplement (2% vol/vol), 2 mM l-glutamine, 100 U/ml penicillin/streptomycin (all from Gibco-BRL), and NeuroCult Differentiation Supplement (10% vol/vol; StemCell Technologies, Vancouver, BC, Canada, http://www.stemcell.com; ).
Mouse R1 ES cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 15% (vol/vol) heat-inactivated fetal calf serum (FCS), 2 mM l-glutamine, nonessential amino acids (all from Gibco-BRL), 50 μM β-mercaptoethanol (Serva, Heidelberg, Germany, http://www.serva.de), and 10 ng/ml recombinant human leukemia inhibitory factor on mouse embryonic fibroblasts . Differentiation of R1 ES cells into neural precursor cells was done as described [24, 25]. Briefly, embryoid bodies (EBs) were generated in hanging drops for 2 days. EBs were then cultured in bacterial dishes for an additional 2 days in Iscove's modified Dulbecco's medium (Gibco-BRL) with 20% (vol/vol) FCS plus supplements and then plated onto tissue culture dishes. Nestin+ precursor cells were obtained at days 8 and 11 in DMEM/Ham's F-12 medium with 5 μg/ml insulin, 30 nM sodium selenite (both from Sigma-Aldrich), 50 μg/ml transferrin, and 5 μg/ml fibronectin (both from Gibco-BRL). HSC were isolated from bone marrow of 6–12-week-old mice after depletion of lineage-positive cells using immunomagnetic beads followed by fluorescence-activated cell sorting for c-kit and Sca-1 double-positive cells as described by Hieronymus et al. .
Cells were fixed with methanol for 1 minute at −20°C (neuron and astroglia double stainings) and with 4% paraformaldehyde in phosphate-buffered saline (pH 7.4) for 20 minutes at room temperature (astroglia and oligodendrocyte double stainings). Fixed cells were then stained with specific antibodies and analyzed by immunofluorescence microscopy. The following antibodies were used: neurons, anti-tubulin β-III (clone TUJ-1; R&D Systems Inc., Minneapolis, http://www.rndsystems.com), 1:200; secondary antibody mouse IgG, Cy3-conjugated (Chemicon, Chandlers Ford, U.K., http://www.chemicon.com), 1:1,000; astroglia, anti-glial fibrillary acidic protein (Dako, Glostrup, Denmark, http://www.dako.com), 1:500; secondary antibody rabbit IgG, Cy2-conjugated (Chemicon), 1:1,000; oligodendrocytes, anti-O4 (R&D Systems), 1:200; secondary antibody mouse IgG, Cy3-conjugated (Chemicon), 1:1,000. Nuclei were counterstained with 4,6-diamidino-2-phenylindole.
RNA Extraction and Microarray Analysis
Total RNA was extracted using RNeasy Midi kit (Qiagen, Hilden, Germany, http://www1.qiagen.com) followed by DNase digestion. RNA was subjected to DNA microarray analysis as described before [26, 27]. Briefly, first- and second-strand cDNA synthesis was done from 7 μg of total RNA with the SuperScript Choice System (Gibco-BRL) and T7-(dT)24 primer (Biotez, Berlin, http://www.biotez.de). In vitro transcription of double-stranded cDNA with biotinylated UTP and CTP (PerkinElmer Life and Analytical Sciences, Boston, http://www.perkinelmer.com) was performed with the MEGAscript kit (Ambion, Foster City, CA, http://www.ambion.com) according to the manufacturer's instructions. Per sample, 10 μg of cRNA was hybridized to Affymetrix mouse MG-U74Av2 GeneChip arrays (Santa Clara, CA, http://www.affymetrix.com) at 45°C for 16 hours. GeneChip arrays were stained, washed, and scanned according to the manufacturer's protocol.
Neurospheres from wt C57BL/6 mice were treated with 150 nM TSA plus 500 nM AzaC for 6, 12, 24, and 48 hours, and RNA was isolated and subjected to reverse transcription (RT)-polymerase chain reaction (PCR) analysis. Briefly, cDNA was synthesized from 1 μg of total RNA using SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) and random hexamer oligonucleotides (Fermentas Life Sciences, St. Leon-Rot, Germany, http://www.fermentas.com). cDNA was then used for PCR amplification with Taq DNA polymerase (Fermentas Life Sciences). PCR fragments were analyzed in 2% agarose gels, and images were recorded with Gel Doc system (Bio-Rad, Hercules, CA, http://www.bio-rad.com). The amount of PCR product was assessed semiquantitatively relative to β-actin. Average values of three independent TSA/AzaC experiments were subjected to cluster analysis, and results were displayed in heatmap format .
PCR primer pairs were designed at the junctions between exons of the genes of interest following recommendations for quantitative PCR (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). Primer sequences and PCR conditions are given in supplemental online Table 1.
Image analysis was done using Affymetrix GeneChip Operating Software (GCOS). Principal component analysis (PCA) analysis  and filtering were performed using GeneSpring software (Agilent, Santa Clara, CA, http://www.agilent.com) with data normalization per chip to 50th percentile and per gene to the median. For hotspot detection the MicroArray Chromosome Analysis Tool (MACAT) package of BioConductor (http://www.bioconductor.org) [28, 30] was used.
PCA was done on a list of 6,941 genes, from three independent experiments (experiments 1–3, described below), that were selected for a significant expression level (absolute raw expression value, >60; p < .06) in at least 3 of the 12 samples analyzed. Hierarchical clustering was performed using Pearson correlation coefficient and the McQuitty linkage method . Gene expression values were normalized using GCRMA . Genes were considered expressed with a normalized expression value >10 in at least three data sets and were differentially regulated by >2 fold (change t test, p < .06) upon treatment. Gene lists were generated by filtering for fold change and intersection in Venn diagrams. All data sets were submitted to the Gene Expression Omnibus database  (http://www.ncbi.nlm.nih.gov/geo; accession numbers GSE587 and GSE2375).
AzaC and TSA Alter Gene Expression in Neurosphere Cells
In culture, neural stem cells from mouse embryonic forebrain form floating colonies  referred to as neurospheres (supplemental online Fig. 1A). Such neurospheres contain cells that can generate (a) new colonies and (b) cells that, following plating under differentiation conditions, can generate neuronal and glial cell types (supplemental online Fig. 1B; ). TSA/AzaC treatment of neurosphere cells increased the level of histone H4 acetylation and decreased methyl-CpG-binding protein (MeCP2) activity . This is expected to lead to an altered chromatin architecture and gene expression pattern [35, 36]. Transplantation of TSA/AzaC-treated neurosphere cells into irradiated recipient mice resulted in a fraction of animals that showed neurosphere-derived long-term, multilineage hematopoietic engraftment .
To determine the TSA/AzaC-induced changes in gene expression, neurosphere cultures were established and treated with AzaC, TSA, or TSA/AzaC for 2 days or were left untreated. Neurosphere cells from Bcl-2 transgenic mice and wt C57BL/6 mice were used , and RNA was prepared and subjected to microarray analysis. Of the ∼12,000 genes analyzed, 42%–54% were found to be expressed in both untreated and treated cells (Table 1). Neurospheres from Bcl-2 transgenic mice were used to protect cells from drug-induced apoptosis [23, 37], and the percentage of expressed genes was lower in Bcl-2 neurosphere cells (42.8%–47.9%) compared with wt neurosphere cells (51.4%–54%).
Table Table 1.. Global gene expression of TSA/AzaC-treated neurosphere cells was assessed by DNA microarray analysis
Surprisingly, neither treatment with AzaC or TSA alone nor simultaneous treatment with both compounds increased the total number of expressed genes. Drug treatment did, however, affect gene expression levels, involving both increases and decreases in gene expression (Table 1). In TSA- or TSA/AzaC-treated cells, 20.9% ± 2.1% of the genes analyzed showed a change in expression by more than twofold: 9.6% ± 0.9% of the genes were upregulated, whereas 11.3% ± 1.5% were downregulated. AzaC alone only marginally influenced gene expression, probably because the microarray analysis was performed after 48 hours of treatment: 3.5% ± 0.7% of the genes were upregulated, and 4.5% ± 0.9% were downregulated. Thus, TSA was more effective than AzaC in inducing changes in gene expression, and there was no further increase in the number of genes affected by simultaneous treatment with TSA and AzaC.
Genome-Wide Analysis of TSA/AzaC-Induced Changes in Gene Expression
To determine whether specific chromosomal regions are particularly susceptible to TSA/AzaC treatment, we studied the chromosomal distribution of differentially expressed genes by MACAT BioConductor software [28, 30]. There was no clear preference of a given chromosome for changes in gene expression induced by TSA/AzaC treatment for any of the 21 mouse chromosomes studied (Fig. 1; data not shown). In addition, no hot spots of TSA/AzaC impact on distinct chromosomal regions were observed.
Next, gene array data were analyzed by PCA, which clusters data sets according to their degree of correlation . PCA demonstrated that TSA and TSA/AzaC samples cluster together, indicating a similarity of the samples at the gene expression level (Fig. 2). AzaC-treated samples cluster with untreated samples (PC1). We also noticed that samples from Bcl-2 neurospheres cluster together and are separated from wt neurospheres (PC2).
To further extend the PCA observations, a list of genes was generated comprising genes that were more than twofold differently regulated upon TSA/AzaC treatment (described in Materials and Methods). This gene list was then subjected to hierarchical clustering  (Fig. 3). Given the observation that TSA/AzaC-treated neurosphere cells acquired long-term hematopoietic potential , gene array data of purified bone marrow-derived Lin−c-kit+Sca-1+ HSC were included in the analysis. In addition, since TSA/AzaC treatment induced gene expression of early embryonic and pluripotency-associated genes (described below), undifferentiated ES cells were also analyzed. ES cells differentiated in vitro into nestin+ cells  served as an additional control to allow the identification of genes expressed in ES cells but not in their differentiated progeny.
Hierarchical clustering enabled us to clearly distinguish TSA/AzaC-induced and -repressed genes, which were consistently regulated upon treatment in both wt and Bcl-2 transgenic neurosphere cells (Fig. 3). Interestingly, some of the TSA/AzaC-induced genes were found to be also expressed in HSC and/or ES cells, such as Bex1/Rex3, an early embryonic gene already expressed at the two-cell stage . ES cells form a distinct cluster but share a number of TSA/AzaC-induced and -repressed genes, such as Zfp277/NIRF4, which is found in early embryonic stem cells .
TSA/AzaC-Induced HSC and Pluripotency Genes in Neurosphere Cells
Further data mining of TSA/AzaC-induced genes identified the HSC marker Cd34 and several early embryonic and pluripotency-associated genes (Oct4, Nanog, Klf4, and members of the Dppa gene family). Therefore, to assess the kinetics of changes in gene expression, neurosphere cells were treated with TSA/AzaC for 6, 12, 24, and 48 hours, and RNA was subjected to RT-PCR analysis. The HSC marker gene Cd34, several ES cell genes (Oct4 and Nanog) and pluripotency-associated genes (such as Ddx4/Vasa, a known early marker of germinal commitment [2, 40, 41]) were analyzed. Bex1/Rex3  and the developmental pluripotency-associated (Dppa) genes Dppa2, 3, 4, and 5  were included in this analysis (Fig. 4). Quantified RT-PCR data were subjected to hierarchical clustering and displayed in heatmap format to show patterns of coregulated genes (Fig. 5).
The HSC marker Cd34 was progressively induced starting at 24 hours, and there was also an induction of the stem cell gene Cd133. Expression of the ES cell transcription factors Oct4 and Klf4 was transiently upregulated after 24 hours, and their expression ceased at 48 hours. Bex1/Rex3 and Ddx4 were induced after 12 hours. The pluripotency-associated genes Nanog and the Dppa family members Dppa2, 3, 4, and 5 were upregulated after 24–48 hours of TSA/AzaC treatment. At the same time, we observed a decrease in expression of the neuronal lineage markers nestin and β5-tubulin (Figs. 4, 5). Importantly Dppa3, 4, and 5 showed a biphasic profile, with an initial downregulation followed by an upregulation after 24 hours. Sox2 and c-Myc, two genes implicated in pluripotency induction [7, , –10], were already abundantly expressed in neurosphere cells, and their expression remained unaffected by TSA/AzaC treatment (Fig. 6). In addition, their expression levels were similar to those in ES cells.
To directly assess the potential pluripotent ES cell-like phenotype, TSA/AzaC-treated neurosphere cells were injected under the kidney capsule of recipient mice, and teratoma formation was studied . Untreated neurosphere cells and ES cells were used as controls. There was no teratoma formation from TSA/AzaC neurosphere cells even after an extended period of time (8 weeks; supplemental online Fig. 2). As expected, there were no teratomas formed for untreated controls, whereas injection of ES cells yielded massive teratomas after 2–3 weeks postinjection. Mice that developed from blastocysts injected with treated neurosphere cells (16 animals) or that were transplanted intravenously with TSA/AzaC neurosphere cells (20 animals) also did not develop tumors (data not shown).
Thus, epigenetic modification of neurosphere cells by TSA/AzaC induced the expression of several genes associated with stem cell phenotype and/or pluripotency, possibly by removal of epigenetic silencing. Reactivation of such genes might be involved in enlarging the developmental competence of these cells. Furthermore, the TSA/AzaC-induced expression of pluripotency-associated genes was transient, and TSA/AzaC-treated cells did not develop teratomas.
TSA/AzaC induced hematopoietic potential in neurosphere cells , and we hypothesized that treatment causes a reprogramming of cells toward (a) a hematopoietic fate or (b) a pluripotent program, from which point cells get committed into the hematopoietic lineage. To investigate this hypothesis, TSA/AzaC-treated neurosphere cells (and, as controls, hematopoietic stem cells and ES cells) were subjected to gene expression profiling by DNA microarray analysis. We show that TSA/AzaC treatment induced in neurosphere cells the expression of several genes associated with stem cell phenotype and pluripotency that might be responsible for or contribute to enlarge the developmental competence of these cells. TSA/AzaC-induced genes included (a) the HSC marker gene Cd34, (b) pluripotency-associated genes, such as Oct4, Nanog, Klf4, and Dppa2, 3, 4, and 5, and (c) the early embryonic gene Bex1/Rex3. Frequently, such genes are epigenetically silenced during development and/or cell differentiation, and TSA/AzaC might reactivate them by removing this epigenetic inhibition.
TSA/AzaC treatment did not increase the percentage of expressed genes in neurosphere cells and did not show any preference for a particular genomic region or chromosome, but it did significantly alter the repertoire of expressed genes. As cell identity is determined by the expression of a defined set of genes, changes in gene expression are expected to lead to changes in cell phenotype and function. TSA/AzaC induced the HSC gene Cd34 and the stem cell gene Cd133, which is also expressed in HSC, suggesting that neurosphere cells can adopt features of HSC. Thereby, such cells might acquire a level of competence that, following transplantation and homing in the bone marrow, generates sustained and transplantable hematopoietic activity .
Recently, Takahashi and Yamanaka , Okita et al. , and two additional studies [9, 10] identified four reprogramming genes (Oct4, Sox2, c-Myc, and Klf4) that by retroviral transfer induced an ES cell-like state in fibroblasts referred to as iPS cells. Interestingly, TSA/AzaC reactivated expression of Oct4 and Klf4 in neurosphere cells, whereas Sox2 and c-Myc were constitutively expressed. Thus, it is tempting to speculate that this combination of factors induced an ES cell-like state in neurosphere cells, similar to what is observed in fibroblasts [7, , –10]. Therefore, TSA/AzaC-induced reactivation of pluripotency-associated genes, such as Oct4, Nanog, and Klf4, together with constitutively expressed Sox2 and c-Myc, might generate a state of competence that, following adoptive transfer into animals and integration in the HSC niche, causes commitment to HSC. TSA/AzaC-treated neurosphere cells did not generate teratomas, indicating that they did not acquire full ES cell activities. This might be because the reactivation of pluripotency-associated genes is transient and, under the experimental conditions used here, does not result in a stable ES cell state.
In addition, the reprogramming system described previously [7, , –10] used retroviral transduction, and as a result, further fortuitous gene activation by retroviral integration might occur and contribute to successful reprogramming. TSA/AzaC-treated cells expressed other pluripotency-associated genes in addition to Oct4, Sox2, c-Myc, and Klf4, and these might very well be involved in or even be required for reprogramming of neurosphere cells. These genes included several members of the Dppa family, such as Dppa2, Dppa4, and Dppa5, the germ line genes Ddx4/vasa/Mvh, and Dppa3/Stella. Dppa genes exhibit an expression pattern similar to that of Oct4  and might be specifically involved in induction and/or maintenance of pluripotency in stem cells. We hypothesize that these genes might be part of the cellular machinery required for the completion of the reprogramming events observed.
In addition, we noted that concomitantly with induction of pluripotency-associated genes, neuronal lineage genes, such as nestin and β5-tubulin, were downregulated. This might be because expression of pluripotency-associated genes and lineage marker genes are mutually exclusive. On the other hand, c-Myc and Sox2 were abundantly expressed in both neurosphere cells and ES cells, and there was no further upregulation by TSA/AzaC treatment, yet they might be important players in the reprogramming event, apparently in concert with other factors.
The reprogramming system by TSA/AzaC relies on the activation of endogenous genes by using defined chemical compounds and transient destabilization of the epigenotype of somatic precursor cells. TSA/AzaC treatment of human CD34+ cells from bone marrow was found to increase the frequency of SCID mouse repopulating cells . This demonstrates the potential application of such strategies for enlarging the developmental capacity of human somatic stem cells and their use in medical therapy.
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
We thank V. Hornich and C. Becker for expert technical assistance, A. Offergeld for expert secretarial assistance, and N. Kirchhof and C. Schmittwolf for help during early phases of this study. We thank Tim Beissbart and Martin Vingron for support in bioinformatics during the initial phase of the project. This work was supported by grants from Deutsche Forschungsgemeinschaft (DFG/SPP1109, DFG GRK1048), Bundesministerium für Bildung und Forschung, and Interdisziplinäres Zentrum für Klinische Forschung BIOMAT. D.R. and R.E.-W. are co-first authors. C.H. is currently affiliated with the Institute for Physiology, Charite, Campus Benjamin Franklin, Berlin, Germany.