Embryonic Stem Cells as a Platform for Analyzing Neural Gene Transcription


  • Xiaodong Zhang,

    1. Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri, USA
    Search for more papers by this author
  • Scott A. Horrell,

    1. Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri, USA
    Search for more papers by this author
  • Deany Delaney,

    1. Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri, USA
    Search for more papers by this author
  • David I. Gottlieb Ph.D.

    Corresponding author
    1. Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri, USA
    • Department of Anatomy and Neurobiology, Box 8108, Washington University School of Medicine, 660 S. Euclid Avenue, St. Louis, Missouri 63110, USA. Telephone: 314-362-2758; Fax: 314-362-3446
    Search for more papers by this author


There is a need for improved methods to analyze transcriptional control of mammalian stem cell genes. We propose that embryonic stem cells (ESCs) will have broad utility as a model system, because they can be manipulated genetically and then differentiated into many cell types in vitro, avoiding the need to make mice. Results are presented demonstrating the utility of ESCs for analyzing cis-acting sequences using Olig2 as a model gene. Olig2 is a transcription factor that plays a key role in the development of a ventral compartment of the nervous system and the oligodendrocyte lineage. The functional role of an upstream region (USR) of the Olig2 gene was investigated in ESCs engineered at the undifferentiated stage and then differentiated into ventral neural cells with sonic hedgehog and retinoic acid. Deletion of the USR from the native gene via gene targeting eliminates expression in ventral neural cells differentiated in cell culture. The USR is also essential for regulated expression of an Olig2 transgene inserted at a defined foreign chromosomal site. A subregion of the USR has nonspecific promoter activity in transient transfection assays in cells that do not express Olig2. Taken together, the data demonstrate that the USR contains a promoter for the Olig2 gene and suggest that repression contributes to specific expression. The technology used in this study can be applied to a wide range of genes and cell types and will facilitate research on cis-acting DNA elements of mammalian genes.

Disclosure of potential conflicts of interest is found at the end of this article.


Author contributions: X.Z.: Conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; S.A.H.: collection and assembly of data, data analysis and interpretation; D.D.: collection and assembly of data; data analysis and interpretation; final approval of manuscript; D.I.G.: conception and design; manuscript writing; data analysis and interpretation; final approval of manuscript.

Spatially and temporally specific gene transcription is one of the most fundamental processes in the normal development of mammalian stem cells. Currently, gene transcription in stem cells is studied by an established set of methodologies that have significant limitations. Transient transfections of promoter-reporter constructs are useful but analyze gene regulatory elements in a nonchromosomal context. This limitation is important as it is now recognized that chromatin neighborhoods are highly diverse and are key determinants of where and when genes are expressed [1]. Another widely used approach is creation of transgenic mice in which transgenes are randomly inserted in chromosomal sites. Unlike transient transfections, this approach places transgenes in a chromosomal context and has been successful for delineation of regulatory cis-acting DNA sequences. However, the chromosomal site at which the transgene is inserted strongly influences the expression pattern of the transgene. This necessitates analysis of multiple lines of mice that vary in the pattern of expression and often involves a subjective choice of which pattern best approximates the true expression pattern of the gene. Also, the need to make transgenic mice strictly limits the number of experiments that can be done so that it is usually impractical to perform extensive structure-function analyses of cis-acting elements using a series of constructs. Because of the limitations of existing approaches, the field of discovering and understanding cis-regulatory elements for stem cell genes is in need of improved technology.

Embryonic stem cells (ESCs) can be the basis of a new option for studying stem cell gene transcription that overcomes the limitations described above. Undifferentiated ESCs are suitable for genetic engineering approaches such as gene targeting and recombinase-mediated cassette exchange (RMCE). By using these techniques, precisely planned alterations of native genes such as insertion of reporters, deletions of nearby or distant DNA sequences, and mutational substitutions can be made. Thus, it is feasible to design and produce changes in native ESC genes that are ideally suited to analyzing regulatory elements. There is also an alternative to making mice for analyzing the expression of the engineered genes because methods for differentiating ESC into many of the major lineages of the embryo are available [2]. These ESC-derived cell populations undergo the key steps in early lineage commitment and provide cells with the most important characteristics of embryo-derived lineages. Differentiated ESC-derived cell populations are an attractive alternative to transgenic mice for the readout of genetic changes engineered into the undifferentiated ESCs.

In this report, we illustrate the use of ESC to map cis-regulatory regions by analyzing the Olig2 gene. Olig2 is a basic helix-loop-helix transcription factor expressed in the developing nervous system, oligodendrocyte precursor cells and oligodendrocytes of the mature nervous system [3]. In the embryo, Olig2 plays an essential role in patterning a ventral compartment that gives rise to motoneurons and the oligodendrocyte lineage [3, [4], [5], [6], [7], [8], [9]–10]. In the neonate and adult, Olig2 is important in the development and maintenance of the oligodendrocyte lineage [3]. Expression of Olig2 is highly specific to subsets of cells at each stage of nervous system development, suggesting that the gene contains multiple regulatory elements [11]. Because Olig2 plays such a central role in differentiation, understanding how Olig2 itself is regulated is important for deciphering the larger transcriptional network that controls development of the ventral neural compartment [12]. At present, relatively little is known about cis-regulatory elements that regulate Olig2 transcription. A cap site for mouse Olig2 mRNA has been mapped by the Japan Transcription Start Site Project (http://dbtss.hgc.jp/index.html) in neonatal brain, and the human Olig2 gene has two transcriptional start sites in a tumor cell line [13]. A downstream enhancer element that drives expression in motoneurons has been mapped [14]. Bacterial artificial chromosome (BAC) transgenes show dramatic upregulation in ventral neural cells, demonstrating the presence of regulatory elements, but these were not further mapped [15].

Because there are no functional studies of the promoter for either mouse or human Olig2, locating and characterizing its promoter is an essential next step in understanding how the gene is regulated. In addition to cap site mapping, functional tests are essential for ascribing promoter function. A growing body of evidence suggests that each chromosomal location of the genome has unique properties [16, [17], [18]–19]. Thus, it is important to analyze regulatory elements in their native chromosomal context as well as with the established methods of stable transgenes and transient transfections. In this report, we define an upstream region (USR) containing a promoter of the Olig2 gene by use of the ability to engineer mouse ESC genes by homologous recombination and RMCE [20]. We also depend on an established culture system to differentiate ESCs into a ventral neural lineage of cells with key physiological and molecular properties of native neural cells such as ventral stem cells and motoneurons [15, 21, [22]–23]. First, the putative promoter region of the native Olig2 gene was deleted by gene targeting in ESCs. Next, the same region was evaluated for its ability to drive transcription in a defined foreign chromosomal site. Finally, the region was tested in transient transfection assays. The results of all three tests strongly suggest that the USR contains a promoter for the Olig2 gene.

This study underscores the potential of ESCs as a general system for analyzing transcription in mammals. Other genes expressed in the ventral neural lineage can be analyzed using the exact approach reported here. The approach can readily be extended to other neural subsets for which robust differentiation protocols are available, such as dopaminergic neurons, retinal cells, mature oligodendrocytes, and cerebellar cells [24, [25], [26], [27], [28]–29]. The same overall strategy is feasible in non-neural lineages that can be derived from ESCs, such as hematopoietic and cardiac lineages among others [2]. Finally, many of the methods will be applicable for analyzing control of gene expression in human ESCs.

Materials and Methods

ESC Culture

The RW4 ESC line and its engineered derivatives were used throughout. Undifferentiated ESCs were cultured by routine methods for feeder cell-free culture. Cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, 10% newborn calf serum, supplemental nucleosides, leukemia inhibitory factor (1000 units/ml), and beta-mercaptoethanol (0.11 mM) on gelatin-coated plastic substrates [30]. Ventral neural differentiation of ESCs was performed by an established method using retinoic acid (RA) and sonic hedgehog (Shh) induction [15, 21]. The Shh agonist HhAg1.4 from Curis, Inc. (Cambridge, MA, http://www.curis.com) was used throughout. Undifferentiated ESCs were scraped from flasks and cultured for 2 days as embryoid bodies (EBs) in DFK5 medium without inducing factors. EBs were then transferred to adhesive gelatin wells in DFK5, and RA (2 μM) and Shh agonist (30 nM) were added to induce ventral neural cells. Culture was continued for 3–5 days as indicated in the text.

DNA Preparation

DNA for detecting junctions from clones in 96-well plates was prepared by a standard method [31]. DNA from T25 flasks was prepared using a kit from Gentra Inc. (Minneapolis, http://www1.qiagen.com) following the manufacturer's instructions.

Targeting Vector and Gene Targeting

The targeting vector used to create ESC line CMVβ was constructed from the backbone plasmid p1338 (Entrez nucleotide accession number AF335419) by four successive steps. The neo cassette of p1338 was replaced with a PGKpuro cassette, and the right homology arm was inserted followed by cloning of the left homology arm. Both homology arms were obtained by polymerase chain reaction (PCR) from RW4 DNA using primers listed in supplemental online Table 1. Finally, the CMV-β-actin promoter was cloned between the LoxP sites. All new junctions in the vectors were verified by DNA sequencing.

For electroporation, 5 × 106 G-Olig2 ESCs were suspended in 1 ml of electroporation buffer with the appropriate plasmid in a 0.4-cm gap cuvette (Bio-Rad, Hercules, CA, http://www.bio-rad.com), and cells were electroporated at 230 V and 960 μF. After electroporation, cells were cultured overnight on 100-mm gelatinized tissue culture dishes in standard medium, puromycin (4 μg/ml) was added to provide selection, and clones were picked after 7 days of selection. Picked clones were expanded in 96-well plates with irradiated STO feeder cells. Recombinant clones were detected by junction PCR, and positive clones were expanded and frozen.

Cre Excision of β-Actin Promoter to Create USRΔ

CMVβ ESCs (5 × 106) were trypsinized and suspended in 90 μl of Amaxa nucleofector buffer (Amaxa Biosystems, Cologne, Germany, http://www.amaxa.com) for ESCs. Six micrograms of the Cre expression plasmid p1411 (gift of Tim Ley, Washington University) in 10 μl of Amaxa nucleofector solution was added, and cells were electroporated with Program A-13 of the Amaxa nucleofector. Transfected cells were plated in 35-mm gelatinized wells and cultured for 2 days. These cells were trypsinized and diluted into conditioned media at clonal density and cultured for 7 days to allow clones to grow. Clones were scanned for green fluorescent protein (GFP) expression, and a total of 12 GFP-negative clones were selected. These were picked and expanded to create permanent stocks, one of which was designated USRΔ. PCR of the new junction followed by DNA sequencing was used to confirm the deletion of the USR.

Olig2 Transgenes in Foreign Chromosomal Site

The B5 RMCE acceptor ESC line was created by randomly integrating a RMCE acceptor cassette into RW4 ESCs and deriving clonal lines. The acceptor cassette in B5 was mapped to Chr12:32,013,167 by a linker PCR-based technique [32]. Olig2 transgenes for these experiments were constructed by first modifying the Olig2 gene in BAC RP23-256L6 by recombineering as described [15]. This resulted in one transgene with the intact USR and another with that region deleted. Both had luciferase inserted in-frame in the open reading frame (ORF) to provide a reporter group. Next the transgenes were subcloned into a shuttle vector with Lox sites and the puro cassette by gap repair in Escherichia coli (supplemental online Fig. 2). ESCs were transfected with the shuttle vector, and the Cre expression plasmid p1411 and recombinant puro-resistant clones were selected. The structure of transgenes in TG1 and TG2 were verified by junction PCR, and Southern mapping was used to verify that only one copy per cell was integrated. Luciferase assays were carried out with a Promega kit (Promega, Madison, WI, http://www.promega.com) following the manufacturer's directions.

Transient Transfection Experiments

Vectors for transient transfection experiments were constructed by amplifying the indicated regions of the Olig2 gene by PCR using primers tailed with appropriate restriction sites and cloning the fragments into the pGL3 luciferase expression vector (Promega). The PCR primers are listed supplemental online Table 1. All constructs were verified by sequencing. Promoter-reporter DNA was transfected into ESCs via Lipofectamine, and cells were cultured for an additional 24 hours to allow expression of the transgene. Luciferase assays to measure transgene expression were performed by a standard method with a kit from Promega.


Induced cultures (see above) were incubated with 4′,6-diamidino-2-phenylindole dihydrochloride (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com) for 1 hour at 37°C before fixation. The wells were rinsed with cytoskeletal buffer (CB) (1.95 mg/ml 2-N-morpholino ethanesulfonic acid, 8.76 mg/ml NaCl, 5 mM EGTA, 5 mM MgCl2, and 0.9 mg/ml glucose; pH 6.1) twice, fixed in 3% paraformaldehyde in CB for 10 min, treated with 0.2% Triton X-100 (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) for 10 minutes, and then incubated for 1 hour in 3% bovine serum album in Tris-buffered saline (TBS) at pH 7.4. Primary antibodies (supplemental online Table 2) were applied overnight (4°C). After washing in TBS for 15 minutes three times, the cells were incubated with secondary antibodies: Texas red-conjugated (Molecular Probes) goat anti-rabbit IgG, Texas red-conjugated goat anti-mouse IgG, and Cy2-conjugated (Jackson Immunoresearch Laboratories, West Grove, PA, http://www.jacksonimmuno.com) goat anti-mouse IgG (all diluted 1:100). Experiments were performed in triplicate; first and second antibody omission controls, as well as nonimmune serum controls, were performed with each experiment to ensure the specificity of staining as described before [33, 34].


Cells were viewed with a Nikon TE2000S fluorescence microscope (Nikon Instruments Inc., Melville, NY, http://www.nikoninstruments.com). Images were acquired on the MetaVue image analysis software (Molecular Devices, Sunnyvale, CA, http://www.moleculardevices.com/home.html).

Fluorescence-Activated Cell Sorting

EBs were dissociated in trypsin-EDTA, resuspended in HEPES-buffered DMEM, and then analyzed on a Epics XL-MCL fluorescence-activated cell sorting (FACS) machine (Beckman-Coulter, Fullerton, CA, http://www.beckmancoulter.com).

Primer Extension Analysis

Primer extension assays were performed to map the 5′ end of the Olig2 transcript. Total RNA was extracted from ESCs and ESC-derived neural cells using a Pureyield RNA Midiprep system (Promega) following the manufacturer's instructions. The primer used in these studies has the sequence 5′gatgatctaagctctcgaatgatccttctt and is approximately 120 nucleotides (nt) downstream of the cap site. The primer was 5′-labeled with 32P with γ-labeled ATP and T4 polynucleotide kinase (New England BioLabs, Ipswich, MA, http://www.neb.com), and 32 μg of RNA was incubated with 16 fmol of labeled primer for 3 minutes at 80°C followed by 15 minutes at 65°C to anneal. Additional reverse transcriptase (RT) components (RETROscript kit, Ambion, Austin, TX, http://www.ambion.com) were added, and the reaction was incubated at 44°C for 60 minutes followed by a 10-minute incubation at 92°C to inactivate the Moloney murine leukemia virus RT. RT products were loaded and run on a Novex 10% TBE-urea gel (Invitrogen, Carlsbad, CA, http://probes.invitrogen.com) at 180 V for 2 hours along with end-labeled markers. Gels were imaged on a Storm PhosphorImager model 860 (Molecular Dynamics, Sunnyvale, CA, http://www.gehealthcare.com).



Improved technology for analyzing gene transcription in developing cell lineages is needed for the reasons presented in the Introduction. A central goal of this work is to demonstrate that ESCs are an excellent system to investigate cis-acting elements. This is done by answering a basic question about transcriptional regulation of Olig2, a gene that plays a key role in neural stem cell biology. Although the results focus on Olig2, the approach can be readily generalized to other neural genes and also to other stem cell lineages.

ESC Lines with Engineered Olig2 Genes

Olig2 is an autosomal gene located on chromosome 16 with two exons separated by a small (∼800 base pairs [bp]) intron (Fig. 1A). The start of exon 1 has been determined by cap site mapping by the Japan Transcription Start Site Project. We performed a primer extension analysis that gave a band consistent with the cap site assigned by the Japan Transcription Start Site Project (supplemental online Fig. 3). These start sites are highly similar to the assigned start site in the human [13 and Japan Transcription Start Site Project]. There is a region of ∼250 bp of high conservation between mammalian species upstream of the assigned start site (UCSC Mouse Genome, http://genome.ucsc.edu) further marking this as a transcriptional start site. Mice with one allele of Olig2 knocked out develop normally; therefore, one copy of Olig2 may be engineered for expression studies without altering cell fate [5, 8]. In this study, we use multiple techniques to assess the function of an upstream region of the Olig2 gene. A map giving the relationship of the segments studied to the gene is shown in Figure 1A. Figure 1B is a table of the 30-nt endpoint sequences of these segments. These data will permit investigators to independently verify and extend the present studies.

Figure Figure 1..

DNA regions relevant to this study. (A): The native Olig2 gene is illustrated (top line) together with the location of manipulated regions (lower lines); red rectangles are exons. Regions include the USR removed by gene targeting to create cell line USRΔ, the transgenes (TG1 and TG2) inserted in a foreign chromosomal site, and the segments (3.8, 2.0, and 1.1 kb) used in transient transfection assays. The position of reporter genes is not illustrated. (B): Sequence of the end points for DNA regions. End point sequences for all segments in A are listed. These sequences allow independent replication of experiments presented. Abbreviations: kb, kilobases; USR, upstream region.

Our first goal was to determine whether the region upstream of exon 1 is essential for gene expression by deleting it from the native gene in an ESC line. To perform this analysis, two new ESC lines with engineered Olig2 genes were constructed by modifying ESC line G-Olig2, a GFP knock-in cell line (Fig. 2B). G-Olig2 was constructed previously by gene targeting and has a GFP cDNA followed by a neo cassette replacing the ORF of Olig2 [35]. The expression pattern of GFP in G-Olig2 ESCs differentiated in cell culture closely resembles that of the native Olig2 gene, showing that GFP serves as a surrogate for Olig2 gene expression [15, 35, [36]–37]. G-Olig2 is thus suitable for deletion analysis of cis-acting elements. ESC line CMVβ was constructed by replacing the upstream 2 kilobases (kb) of the Olig2 gene with the CMVβ-actin promoter, which is optimized for strong expression. A targeting construct with a puro cassette followed by the CMVβ-actin promoter was inserted in the modified Olig2 allele of G-Olig2 so that the 2 kb upstream of Olig2 is deleted and replaced with the incoming cassette (Fig. 2C). This allows easy detection of recombinant clones, because random integrants are GFP-negative (the Olig2 promoter is off in ESCs) but targeted clones are GFP-positive. Two clones of ∼500 were intensely GFP-positive, consistent with correct gene targeting, which was verified as described below. ESC line USRΔ was engineered by excising the puro cassette CMVβ-actin promoter from CMVβ cells via flanking LoxP sites. CMVβ cells were transfected with a Cre-expressing plasmid and screened for GFP-negative clones, three of which had the planned excision of the CMVβ-actin promoter; one of these was named USRΔ.

Figure Figure 2..

Deletion of USR of native Olig2 gene. (A): Native gene. Exons (red) and the ORF are indicated as well as the 2 kb (USR) in blue. (B): G-Olig2. GFP knock-in created by HR provides a reporter cell line. Note that the GFP neo cassette replaces the ORF. Xs mark sites of HR. (C): CMVβ. The USR of G-Olig2 is replaced via HR by a puro cassette + CMVβ-actin promoter unit. (D): USRΔ. The puro cassette + CMVβ-actin promoter is removed by Cre recombination of the LoxP sites (green) to create a gene with the USR deleted. Abbreviations: GFP, green fluorescence protein; HR, homologous recombination; kb, kilobases; ORF, open reading frame; USR, upstream region.

Structure of Engineered Olig2 Genes

CMVβ and USRΔ ESC lines were analyzed by PCR to determine the structure of engineered Olig2 genes using the primers shown in Figure 3A. The inserted CMVβ promoter was mapped by PCR amplification of the left and right novel junctions created by the insertion. As seen in Figure 3B, the predicted 925-bp left junction product is present in CMVβ but absent in USRΔ, G-Olig2, and RW4 control ESCs. For the right junction the predicted 2,513-bp band is present in CMVβ cells and is absent in USRΔ, G-Olig2, and RW4 cells. For ESC line USRΔ, PCR demonstrated that the puromycin promoter cassette had been excised (Fig. 3B) The predicted 671-bp product of primers F3 and R3 is present in USRΔ but absent in CMVβ, G-Olig2, and RW4 cells because the long intervening sequences prevent efficient PCR. To further verify the excision, the PCR product from USRΔ was sequenced. The results show exactly the predicted sequence (Fig. 3C), in which sequences that are 2 kb apart in the intact gene are brought together and flank the LoxP site in USRΔ. Also, the inserted GFP cDNA was sequenced and found to be intact. Thus, lack of GFP expression (see below) is due to lack of transcription, not to mutation in GFP.

Figure Figure 3..

Mapping the structure of engineered genes. (A): PCR analysis of junctions in CMVβ and USRΔ cell lines. Structure of the genes is shown along with the primers used in polymerase chain reactions (PCRs). (B): PCR results. Top panel, left and right junctions of the CMVβ-actin promoter insertion. Left junction with primers F1–R1: lane 1, CMV β; lane 2, USRΔ; lane 3, G-Olig2; lane 4, RW4. Right junction with primers F2–R2: lane 5, CMVβ; lane 6, USRΔ; lane 7, G-Olig2; lane 8, RW4. Bottom panel, PCR across the deleted USR 2-kb region with primers F3 and R3. Lane 9, USRΔ; lane 10, CMVβ; lane 11, G-Olig2; lane 12, RW4. (C): Sequences at the junction in USRΔ are as predicted. Sequences that flank the USR and are 2 kb apart in native genes are now closely apposed and flank the LoxP site (green). Abbreviations: GFP, green fluorescence protein; kb, kilobases; USR, upstream region.

Additional evidence for these structures was obtained by determining the puromycin sensitivity of the three lines (data not shown). G-Olig2 lacks a puro cassette and is puromycin-sensitive. CMVβ has an inserted puro cassette and is functionally puromycin-resistant. USRΔ has the cassette excised and reverts to puromycin sensitivity.

In summary, mapping data support the structures of CMVβ and USRΔ genes illustrated in Figure 2.

The 2-kb Upstream Region Is Essential for Olig2 Gene Expression

The effect of deleting the USR on gene expression was determined by comparing GFP expression of the intact gene (G-Olig2 cells) and the gene with the deleted USR (USRΔ cells). Expression was observed throughout the developmental series from undifferentiated ESCs to induced ventral neural cells. A standard protocol for differentiating ESCs into ventral neurons [15, 21] was used throughout (Fig. 4A). In brief, ESCs are cultured for 2 days as EBs in serum-free medium without inducers and then plated as undissociated EBs on to an adhesive surface in the presence of the inducers RA and Shh. Controls with no inducers (−/−) and either RA alone (RA/−) or Shh alone (Shh/−) are included in most experiments. In this and all other experiments Shh refers to the agonist HhAg1.4 from Curis, Inc.

Figure Figure 4..

GFP expression in the three genotypes. (A): Differentiation procedure for ESC. The flow chart for differentiation in cell culture is outlined. ESCs are first cultured as EBs without inducing factors and then replated in the presence of inducing factors. (B): Expression in embryoid bodies cultured for 1 day. (B, A1 and A2): G-Olig2 phase contrast and fluorescence. (B, B1 and B2): CMVβ phase-contrast and fluorescence; (B, C1 and C2): USRΔ phase-contrast and fluorescence expression is specific to cells with the CMVβ promoter. (C): Expression in 2−/3++ cells (cultured 2 days as EBs and 3 days with RA and Shh). Column 1, phase contrast; column 2, GFP. Row A, G-Olig2; row B, CMVβ; row C, USRΔ. GFP expression is strong in G-Olig2 and CMVβ cultures but absent in USRΔ cultures. (D): GFP expression in 4-day plated, induced cells; 20 adhered aggregates for each genotype and induction condition were viewed and scored for presence of GFP+ cells. Note the absence of expression in USRΔ cultures. (E): FACS analysis of cells 5 days after plating. The genotype and induction conditions are displayed. The results show the percentage of cells that are positive for each condition. USRΔ cells are uniformly negative even under inducing conditions. Abbreviations: EB, embryoid body; ES, embryonic stem; FACS, fluorescence-activated cell sorter; GFP, green fluorescent protein; RA, retinoic acid; Shh, sonic hedgehog.

The pattern of expression with the intact USR (G-Olig2 cells) is as reported previously [15]. Undifferentiated ESCs (data not shown) and 1D EB stage (Fig. 4B, panels A1 and A2) are GFP. At 4 days after plating, expression depends on the presence of inducers (Fig. 4D, 4E). Cultures without inducers and cultures with Shh alone are GFP. Cultures with RA alone have very rare scattered GFP+ cells in some aggregates (11 of 20) (Fig. 4D). Cultures with both RA and Shh have abundant intensely GFP+ cells (Fig. 4C, panels A1 and A2) and 20 of 20 EBs were GFP+ (Fig. 4D). FACS analysis supports these conclusions (Fig. 4E).

The expression pattern when the USR is replaced with a strong general promoter (CMVβ cells) is predicted to be general, and the data support this prediction. CMVβ cells are intensely and uniformly GFP+ at the ESC stage (data not shown) and also at the 1D EB stage (Fig. 4B, panels B1 and B2). At the 3-day postplating stage, all cultures have abundant GFP+ cells independent of added inducers (Fig. 4C, panels B1 and B2, 4D). FACS analysis supports the microscopic data (Fig. 4E). In summary, expression of the Olig2 gene driven by a strong general promoter is general and does not depend on inducers.

Deletion of the USR (USRΔ cells) abolishes expression at all stages. ESCs (data not shown) and 1D EBs are negative (Fig. 4B, panels C1 and C2). After plating USRΔ cultures with or without inducers have no GFP+ cells (Fig. 4C, panels C1 and C2, 4D). FACS analysis corroborates this conclusion (Fig. 4E). In summary, deleting the USRΔ of the Olig2 gene abolishes expression in all the cell types produced in our culture systems, and we conclude that this region is essential for expression.

Neural Differentiation After Three Cycles of Chromosomal Engineering

The lineage leading to ESC line USRΔ has undergone three successive rounds of chromosomal engineering: (a) replacement of the Olig2 ORF with GFP; (b) replacement of the upstream 2 kb with the CMV promoter; and (c) excision of that cassette mediated by CRE. Nevertheless, the cells still undergo neural differentiation (Fig. 5). There is abundant expression of Olig1, Olig2, Nkx2.2, nestin, and β-tubulin III. The adherent cells also form extensive neurites (supplemental online Fig. 1). Thus, multiple rounds of chromosomal engineering can be carried out without loss of key developmental capabilities.

Figure Figure 5..

Upstream region (USRΔ) cells treated with retinoic acid and sonic hedgehog for 4 days express appropriate markers. Immunolocalization demonstrates that Olig1 (A3–A4), Olig2 (B3–B4), and Nkx2.2 (C3–C4) (red) are expressed in nuclei; neural markers nestin (D3–D4) and β-tubulin III (E3–E4) (red) were expressed in the neurites and cell bodies. DAPI (blue) was a stain for DNA (A2–E2). A1–E1, differential interference contrast. Abbreviation: DAPI, 4,6-diamidino-2-phenylindole.

Promoter Activity of the Upstream Region in a Foreign Chromosomal Location

The deletion experiment shows that the USR of the Olig2 gene contains a promoter or other essential cis-acting sequence. One characteristic of promoters is that they are sometimes active in foreign chromosomal locations. We therefore investigated whether the USR was essential for expression in an identified foreign chromosomal site. ESC line B5 was engineered by random integration of an RMCE cassette followed by linker-mediated mapping [32] of the insertion site. The insertion site is located in chromosome 12 (Chr12:32,013,167). To test the activity of the USR, Olig2 transgenes with and without the USR were targeted via Cre recombinase-mediated cassette exchange to the acceptor site (Fig. 6A). Both transgenes have a luciferase reporter replacing the Olig2 ORF (Fig. 6B). ESC line TG1 has the intact transgene and line TG2 has the same endpoints but carries a deletion of the USR. In both TG1 and TG2 ESC lines the transgene inserted as a single crossover at the upstream Lox 511 site; TG1 and TG2 are identical except for the presence or absence of the USR. TG1 and TG2 ESCs were differentiated in the presence and absence of the factors indicated in the legend to Figure 4C, and the level of luciferase expression was assayed. In TG1 with the intact USR, RA alone or RA + Shh strongly induces expression whereas Shh alone is without effect. In sharp contrast, the deleted transgene in TG2 is expressed at a low basal level and is not induced by any of the factors. We conclude that the USR is necessary for inducible expression in a foreign chromosomal location. It is surprising that expression of the transgene is induced by RA alone whereas the native gene requires RA + Shh. This could be due to a repressive element in the native gene that is blocked by Shh and was not included in the transgene. Alternatively, an element from the chromosome near the insertion site may overcome the need for Shh. In either case the results show that the USR is essential for expression at a foreign location. It is also notable that both transgenes are expressed at low levels in noninduced EB cells. This could be due to lack of a distant repressor that was not included in the transgene or, alternatively, to read-through from a promoter flanking the integration site. In either case the low-level expression points to the importance of using the native locus as well as foreign sites for analyzing promoters.

Figure Figure 6..

USR is required for expression of transgene in a foreign site. (A): Structure of the two transgenes in the inserted in recombinase-mediated cassette exchange site at Chr12:32,013,167. Colored triangles are Lox sites used for Cre-mediated exchange. The bracket in TG1 indicates the USR present in TG1 but deleted in TG2. Note that TG1 and TG2 differ only by the presence or absence of the USR. (B): Expression of luciferase in cells cultured as embryoid bodies for 2 days without factors and then for 4 days with the factors indicated on top of the bars. Red bars are for the construct with the USR (TG1) and green bars for the deleted (TG2) construct; circles show duplicate cultures. The transgene with the USR is induced by RA and RA + Shh, but the transgene without this region is not. Abbreviations: RA, retinoid acid; Shh, sonic hedgehog; USR, upstream region.

Promoter Activity of the Upstream Region in Transient Transfection Assays

Transient transfection experiments were performed to determine whether the USR has promoter activity. Upstream segments were cloned 5′ to the luciferase gene in vector pGL3 to produce a series of promoter-reporter constructs (Fig. 7A). As a further measure of specificity, each segment was cloned in both orientations. Transient transfection of undifferentiated ESCs shows that a 3.8-kb upstream region is active in the + orientation but has much less activity in the − orientation (Fig. 7B). The upstream 2 kb alone is not active in the + orientation and has slight activity in the − orientation. Finally, the 1.1 kb immediately upstream of exon 1 is both active and orientation-sensitive; notably it is as active as the entire 3.8-upstream segment. The series of vectors was also transfected into STO cells and A293 cells with very similar results (data not shown). We conclude that the upstream 1.1-kb segment behaves as a promoter in ESCs, STO cells, and 293 cells, all of which are cell lines that do not express the Olig2 gene. Therefore, the promoter is nonspecific, and we hypothesize that specific expression is due to the interaction of a nonspecific promoter and one or more cis-acting repressor elements.

Figure Figure 7..

Promoter activity of the upstream region. (A): Segments of the upstream region cloned into luciferase expression vector. Each segment is cloned in the (+) and (−) orientation. (B): Transient expression of upstream segments given as luciferase activity normalized to the GL3 control vector. Height of bar is mean of three assays with the SD displayed as lines above the bar. Abbreviations: kb, kilobases; LUC, luciferase.


A central goal of this study was to assess the suitability of ESCs as a technology for analyzing mechanisms of gene transcription in stem cell lineages. Gene targeting in ESCs allows precise changes to be made in most genes. Until recently most gene targeting experiments focused on coding regions, but targeting of noncoding regulatory regions is also feasible. Once a regulatory region of an ESC gene is targeted, the effect on gene expression can be determined in many cases by studying differentiated cells produced in cell culture. Thus, for many (but not all) experiments, the expensive and rate-limiting step of making mice can be avoided. The results of our experiments support this use of ESCs for characterizing cis-acting elements of stem cell genes.

To illustrate the approach, we aimed to determine whether a USR of the mouse Olig2 gene is essential for expression. The region was chosen because it is immediately upstream of a mapped cap site and has a 250-bp region of high conservation. The data show that the region is, in fact, critical for expression, most likely because it contains a promoter. The simplest and most stringent test was to delete the USR from the native Olig2 gene in ESCs. This deletion results in a complete loss of gene expression in differentiated neural cells that normally express the Olig2 gene. The most likely interpretation is that the region has a promoter that is essential for expression in many types of neural cells. In a second experiment, transgenes with and without the USR were inserted by RMCE in the same defined foreign chromosomal site. The USR was essential for regulated expression at this foreign site, further underscoring its functional importance. Notably, the pattern of regulation of the transgene differed significantly from that of the native gene, again emphasizing the importance of carrying out studies on the native locus. Finally, sequences from the USR were tested for promoter activity by transient transfection assays. The 1.1-kb segment immediately upstream of the cap site had high and orientation-specific promoter activity. However, it is also active in cells that do not express Olig2, suggesting that the promoter is nonspecific. This predicts that the gene is repressed in ESCs and non-neural lineages and is expressed when the repression is blocked during development of the ventral compartment of the nervous system. The Olig2 gene has an enhancer-like sequence [14] that probably interacts with the USR region described here.

The ESC system generates many of the cell types characteristic of the ventral neural pathway including dividing stem cells, motoneurons, and oligodendrocytes [21, 22, 36, [37]–38]. However, it is conceivable that some Olig2-expressing cell types do not differentiate in the system, and regulatory elements used exclusively by such cells will be missed. Thus, the present experiments map one essential region but do not exclude the possibility of a second promoter that is active in cells present in the mouse but absent in the ESC culture system.

These studies on the Olig2 gene illustrate the strengths of the ESC system for gene transcription studies. One is the ability to delete and mutate regulatory regions in their native chromosomal context, which is the most stringent test of function. Recently, a 150-kb upstream region of the mouse globin locus was replaced by the corresponding human sequence using a recombinase based (RMCE) method [39]. This paradigm can be used for testing large series of mutated regulatory regions for a single gene. In another study, an upstream element of the protocadherin gene cluster was deleted and shown to be functional [40]. The second strength of ESCs is the ability to obtain differentiated cells directly in cell culture and thus obviate the expensive and time-consuming step of making mice. This allows systematic series of mutations to be analyzed affordably, a capability that is essential because of the complex array of regulatory regions of most genes. Finally, the cell culture system generates millions of cells and is thus well suited to biochemical approaches that are constrained by the small numbers of cells in embryos. These include chromatin immunoprecipitation-based analyses of transcription factors and protein-protein association studies as well as detailed studies of the kinetics associated with regulation. With the wide array of differentiated types that can be produced in cell culture from ESCs [2], this system is likely to be important in many fields of mammalian regulatory biology.

Disclosure of Potential Conflicts of Interest

The authors indicate no potential conflicts of interest.


This work was supported by National Institutes of Health Grant R01NS045809 (D.I.G.). We thank Tim Ley for plasmids, Christina Chen and Barak Cohen for bioinformatics analysis, and Lee Rubin and Curis Inc. for Shh agonist.