A short upstream promoter region mediates transcriptional regulation of the mouse doublecortin gene in differentiating neurons

Cell type-specific activity of the 2kb-long Dcx promoter fragment

We first wanted to determine whether pdcx 2kb was sufficient to induce transcriptional activity and to drive it specifically in cells expressing the endogenous Dcx gene, namely neuronal precursor cells. Two mouse cell types were used: cerebellar neurons from post-natal day 3 mice (PND3) in which endogenous Dcx expression has already been reported [18] and mouse embryonic stem (ESR1) cells.

Cell specificity of pdcx 2kb transcriptional activity was observed in both heterogeneous cell cultures by transient expression of the eGFP protein upon transfection with pdcx 2kb-eGFP constructs (Figure 1b). In primary cerebellar cells, eGFP under the control of pdcx 2kb was detected only in cells expressing the endogenous Dcx gene and not in other cell types, such as GFAP-expressing astrocytes (Figure 1b: A-H). Control experiments using the widely active pCMV-eGFP construct revealed that eGFP fluorescence was detected indifferently in all cell types, neurons and glial cells alike. Both GFAP- and Dcx-positive cells were fluorescent (Figure 1b: I-P), showing that the transfection process was not cell-specific. Similarly, eGFP expression was only detected in ESR1 cells after neuronal differentiation and was limited to Dcx-positive or βIII-Tubulin-positive cells, confirming pdcx 2kb selectivity to Dcx-positive neuronal precursor cells in vitro (data not shown).

Ex vivo electroporation of embryonic mouse brains was also conducted to confirm cell selectivity of pdcx 2kb activity in a more physiological setting. pdcx 2kb-eGFP was transfected in brains of E14.5-E15.5 mice, a specific time point of corticogenesis [19]. Dcx expression was observed in migrating cortical neurons originating from progenitor cells located near the ventricular proliferative zone. Following pdcx 2kb-eGFP transfection, fluorescence was observed around the electroporation site in the proliferative zone of the cerebral cortex. Fluorescent cells displayed morphology similar to that of migrating neuronal precursors (Figure 1c). Moreover, all fluorescent cells also expressed Dcx. Taken together, these qualitative analyses show that the 2kb-long fragment upstream of the Dcx gene possesses regulatory sequences sufficient to mimic the activity of the endogenous Dcx gene in mouse embryos and to limit its activity to Dcx-positive neuronal precursors.

Transcriptional activity of the Dcx promoter in differentiating ES cells

To study more efficiently the transcriptional regulation of the mouse Dcx gene, we chose to use mouse embryonic stem (ESR1) cells: these cells are able to proliferate indefinitely in vitro while retaining pluripotency, and also to differentiate into a large variety of cell types in vitro. Hence, mouse ES cells can differentiate into neural precursors able to generate functional neurons [20], astrocytes and oligodendrocytes [21, 22]. Using our protocol (see Material and Methods), induction of neuronal differentiation leads to a progressive change in ESR1 cell morphology to a characteristic cobblestone structure and, by neuronal differentiation day 8 (DD8), ESR1 cells begin to form long cellular processes. From DD8/10 on, bipolar cells proliferate and form rosettes.

Dcx protein was detected at DD6 of ESR1 differentiation (Figure 2) and was still present in immature neuronal cells-derived from ESR1 cells at DD8 and DD18. Based on these observations and on the detection of Dcx mRNA at DD4, we decided to use ESR1 at DD6 for further analysis of the transcriptional activity of the mouse Dcx promoter.

To characterize relevant regions of pdcx, we dissected the 2kb-long fragment into shorter fragments of 1.2 kb, 1 kb (obtained by PCR) and 249 bp (obtained by enzymatic restriction), and inserted them upstream of the luciferase reporter gene.

Comparison of the respective activities of all constructs (Figure 3a) in ESR1 cells at DD8 revealed a higher activity in cells transfected with pdcx 2kb than in control cells (figure 3b). Transcriptional activities were also higher in cells transfected with shorter fragments (pdcx 1kb and pdcx 249bp) than with pdcx 2kb (Figure 3b). In contrast, the 1.2kb-long pdcx fragment did not induce a transcriptional activity different from control, suggesting the presence of a transcriptional repressor region in this 200 bp fragment. In addition, deletion of 79 bp at the 3'end of pdcx 1kb (pdcx 1kb-) prevented the pdcx 1kb-induced increase of activity. This last result suggested the presence of important regulatory domains at the 3'end of pdcx, especially in a 79bp-long fragment. Similar effects on the transcriptional activities of the different constructs were observed in PND3 cerebellar cells (Figure 3c), confirming the presence of similar regulatory control mechanisms in neuronal precursors from both origins (ESR1 cells and primary cerebellar cells). However, a discrepancy is observed between the relative activity of pdcx 1kb and pdcx 249bp between both cell cultures (Figure 3b and 3c), maybe reflecting the fact that the target regulatory elements necessary for the Dcx gene expression are different in two cell types. Indeed, it is quite possible that activators or inhibitor elements, located in pdcx 1kb and pdcx 249bp respectively, are most needed during the differentiation of CGN compared to neuroblasts from ESC. At that stage (DD6), ESR1 cells are plated on gelatin-coated plates while cerebellar cells are plated on poly-ornithine substrate (necessary for cell adhesion) (see also below).

The same relative activities of the promoter fragments were maintained throughout the entire differentiation program in ESR1 cells (Figure 4). Indeed, when ESR1 cells were transfected every second day throughout the entire differentiation period and the luciferase activity measured 48 hours later, increasing levels of activities were observed along the process with a peak at the transfer onto poly-ornithine/lamin substrate for constructs pdcx 2kb, pdcx 1kb and pdcx 249bp. The time-course studies confirmed the inhibition or the significant reduction of the transcriptional activity of pdcx at all stages by deletion of 79b at the 3' end of the pdcx sequences in constructs pdcx 1kb and pdcx 249bp (pdcx 1kb^- and pdcx 249bp^-, respectively). Finally, we also examined the cellular specificity of pdcx 1kb and pdcx 249bp in differentiated ESR1 and PND3 cerebellar cells and confirmed that, similar to the pdcx 2kb construct (Figure 1b), both pdcx 1kb and pdcx 249bp fragments restricted the expression of the eGFP reporter gene in Dcx-expressing cells (data not shown).

Transcription factors regulating the mouse Dcx promoter

Comparative analyses were performed on the sequences of mouse, rat, human and chicken 2kb-long Dcx promoter fragments using the VISTA software (Figure 5). A high sequence similarity was found between mouse and rat pdcx 2kb sequences. Comparison between mouse and human sequences revealed that sequence conservation was not detected throughout the 2kb-long sequence, but was limited to a ± 500 bp portion located at the pdcx 2kb 3'-end. Similarly, sequence conservation between the mouse pdcx 2kb and the chicken pdcx 2kb was only observed in a 183bp-long fragment located at the 3'-end. In addition, the strongest sequence conservation between mouse, rat, human and chicken pdcx 2kb sequences was detected within this 183bp-long fragment (Figure 5). Further analysis, using the MatInspector software, revealed that many putative binding sites for members of several transcription factor families known to participate in neuronal differentiation/migration such as NEUROD, NEUROG, BRN, PAX, HOX, SOX and DLX are present in the 249bp-, 1kb-, 2kb-long Dcx promoter fragments (Additional file 1: Table S1). In particular, this analysis revealed four conserved consensus binding sites for known transcription factors in the 79b-long fragment located at the 3'end of the Dcx promoter (Figure 6): two perfect consensus sites for NF-Y/CAAT, one perfect site for LEF/TCF which are effectors of the canonical Wnt pathway and one imperfect site for HNF6/OC2 (75% conserved relative to consensus), members of the ONECUT family [23–25].

Expression of putative transcription factors during neuronal differentiation of mESC

To investigate which of these transcription factors, present in the 79 bp fragment and potentially involved in Dcx gene regulation, are expressed during neuronal differentiation of ESR1 cells, the levels of mRNA coding for Dcx and for each factor (Lef, Tcf1, Tcf3 and Tcf4, Hnf6 and Oc2, Nf-ya, Nf-yb and Nf-yc) were measured at different time-points during neuronal differentiation (Figure 7).

The abundance of each transcript at each stage was calculated relative to the Gapdh housekeeping gene mRNA, based on published results [26–28]. The observed pattern of Dcx transcript expression is consistent with that of Dcx protein synthesis during neuronal differentiation of ESR1 cells and with the observed transcriptional activity of the Dcx promoter (see above). No Dcx transcript was detected in undifferentiated ESR1 cells. Levels of Dcx transcripts progressively increased to reach a maximum at DD18 followed by a reduction at DD28. Each transcription factor transcript was detected in ESR1 cells. The temporal profile of Lef, Hnf6 and Tcf4 transcription factors presented patterns of expression similar to that of Dcx, suggesting that expression of these transcription factors is also dependent on neuronal differentiating time. Their levels were relatively low during the early phase of ESR1 cell differentiation (DD0-8), progressively increased during the late phase of the differentiation program to peak at DD18 when Dcx transcript levels were maximum. In contrast, Nf-ya, Nf-yb, Nf-yc and Oc2 transcripts displayed relatively constant steady state levels, with little variation during the entire differentiation program. The relative amounts of Tcf1 and Tcf3, seem to depend on the differentiation process, with Tcf3 decreasing during differentiation [29] and Tcf1 being apparently sensitive to the culture matrix substrate.

Since mRNA for each transcription factor was present in ESR1 cells and since no transcription factor could be discarded on the sole basis of the temporal expression profile of its transcript, selective mutagenesis experiments were undertaken.

To determine the relative importance of each DNA binding site present in the 79bp-long pdcx-luciferase fragment, we individually altered each putative binding site for LEF/TCF, HNF6/OC2 or NF-Y/CAAT in the Dcx promoter using site-directed mutagenesis to generate pdcx 249bp/Lef*, pdcx 249bp/Hnf6* and pdcx 249bp/Nf-y* (Figure 8). Then we compared the luciferase activities induced by each construct to that of pdcx 249bp by transient expression experiments in ESR1 cells at three stages of neuronal differentiation. At the beginning of the differentiation program (DD2), mutation of the NF-Y or HNF6 binding sites did not affect pdcx 249bp activity (Figure 8a). In contrast, mutation of the LEF/TCF site completely inhibited pdcx 249bp activity. Such inhibition was not detected in ESR1 cells at DD8 (Figure 8b). At that stage, the relative activity of every mutated pdcx 249b was below that of the wild type pdcx 249bp but higher than that of the 79bp-truncated pdcx 249bp. Finally, at DD20, when many ESR1 cells expressed the Dcx gene and displayed a neuronal phenotype, mutation of any of the four binding sites reduced the activity of pdcx 249bp to that of the 79bp-truncated pdcx 249bp (Figure 8c). Similar results were found using pdcx 1kb (data not shown) and in PND3 cerebellar cells (Figure 8d). Simultaneous mutation of all three binding sites was also performed (pdcx 249bp/Lef*/Hnf6*/Nf-y*). Transfection experiments revealed that the transcriptional activation of pdcx 249bp mutated for LEF, HNF6 and NF-Y binding sites is similar to the constructs with any one mutated site (Additional file 2: Figure S1) at all stages, suggesting that an additional, not yet identified factor is responsible for the observed residual activity at DD8. Altogether, these results demonstrate first of all the importance of LEF/TCF, HNF6/OC2 or NF-Y/CAAT DNA binding sites for Dcx promoter activity at various stages of neuronal differentiation and that the corresponding factors are part of the regulatory mechanisms controlling Dcx promoter activity. However, these transcription factors are not sufficient to induce full transcriptional activation of the Dcx gene. In addition, the results obtained with the LEF/TCF binding site reveal that this binding site could be more specifically involved in activating pdcx at early stages of neuronal differentiation of ESR1 cells.

Animals

All experimental procedures on animals were carried out according to the European Communities Council Directive (86/609/EEC) for care and use of Laboratory animals. The experimental protocols were reviewed and approved by the institutional Animal Care Committee.

Materials

ES cell culture medium ingredients (DMEM/F12, Neurobasal medium, B27 supplement, Bovine Serum Albumin and L-glutamine) were obtained from Invitrogen-Gibco (Merelbeke, Belgium); BMP4 (Bone Morphogenic Protein-4) was obtained from R&D Systems (Abingdon, United Kingdom); apo-transferrin, progesterone, insulin, putrescine, sodium selenite, laminin, poly-ornithine and all chemicals were obtained from Sigma-Aldrich (St. Louis, MO, USA) unless specified otherwise and ESGRO^® (LIF, Leukemia Inhibitory Factor) was from Chemicon International (Temicula, CA, USA) unless specified otherwise.

ESC Culture - Monoculture Differentiation and Transfection

ESR1 cells were maintained without feeder cells in serum-free culture medium. ESC were plated onto 0.1% gelatin-coated plates in N2B27 medium (as described by Ying et al., 2003 [28]) supplemented with ESGRO^® (LIF) 10³ Units and BMP4 10 ng/ml and incubated at 37°C in 5% CO₂. The medium was renewed every day. Cells were plated every 3 days using a dissociation solution (trypsin 0.05%, chicken serum 1%, EDTA 0.53 mM). To start monolayer differentiation, corresponding to DD0, undifferentiated cells were dissociated and plated onto 0.1% gelatin-coated plates at 10⁴cells/cm² in N2B27 medium supplemented with ESGRO^® (LIF) 10³ Units. The medium (non-supplemented N2B27) was changed every other day during 8 days, until DD8. Then, at DD8, cells were dissociated and plated on poly-ornithine (100 μg/ml) and laminin (0.5 μg/ml)-coated plates. The medium was changed 2 h and 24 h after plating and every 2 days until 20 days of differentiation on porn/lam substrate, corresponding to DD28. ESR1 cells can differentiate into neural precursors, which could generate functional neurons [53], astrocytes, and oligodendrocytes [21, 22].

Transient transfection experiments were performed using FuGene-6 (Roche Applied Science, Roche, Mannheim, Germany) as described by the manufacturer. eGFP expression was checked 24 h, 48 h and 72 h after transfection or luciferase/β-galactosidase activity ratios were determined 48 hours after transfection. The luciferase activity was measured according to the Luciferase Assay System (from Promega, Leiden, The Netherlands) and is expressed in RLU (Relative Light Units). The β-galactosidase activity was determined by reacting 40 μl of lysate with 40 μl of β-galactosidase substrate containing ONPG (o-nitrophenyl-D-galactopyranoside) and the absorbance was measured at 420 nm.

Cerebellar Granule Neuron (CGN) Culture and Transfection

Primary cultures of CGNs were obtained from post-natal day 3 (PND3) 129/sv mice. Isolated cerebella were stripped off meninges in DPBS (Dulbecco's Phosphate-Buffered Saline) (Invitrogen-Gibco) supplemented with glucose (0.45%), minced, and treated with dissociation solution (0.25% trypsin, 0.01% DNase) for 25 min at 37°C. Cell dissociation was stopped by adding 0.5 ml of FBS (Fetal Bovine Serum) to the dissociation solution. Granule cells were then mechanically dissociated by two successive triturations and sedimentation steps in complete medium (DMEM, 25 mM KCl, 96 mM NaCl, 5 μM insulin, 2 mM glutamine, 1.5% glucose 30%, 1% sodium pyruvate, 5% heat-inactivated FBS, 10% horse serum), and resuspended in 3 ml of complete medium. Neurons were plated onto poly-ornithine (100 μg/ml)-coated 24-well culture plates (Nunc, Wiesbaden, Germany) at a density of ~1 × 10⁴ cells/drop and incubated at 37°C in 5% CO₂. The following day, transient transfection experiments were performed using FuGene-6 (Roche Applied Science) as described by the manufacturer. eGFP expression was checked 24 h, 48 h and 72 h after transfection and the cells were fixed for immunocytochemistry analysis or lysed for luciferase/β-galactosidase activity ratio determination 48 hours after transfection.

Immunocytochemistry (ICC) and Immunohistochemistry (IHC)

Western Blotting (WB)

After 2 washes with PBS, cells on 10 cm² plates were lysed by adding 300 μl lysis buffer (0.15 M NaCl, 0.05 M TrisBase, 1% TritonX-100, 1% Sodium deoxycholate, 0.1% SDS) supplemented with protease inhibitor cocktail (Roche) and PMSF (phenylmethylsulfonyl fluoride). After centrifugation at 4°C, the supernatant was removed and the protein content was determined using Bradford reagent (Bio-Rad Laboratories, Hercules, CA, USA). Samples were boiled in sample buffer (2×: 4% SDS, 10% β-mercaptoethanol, 135 mM Tris-HCl pH6.8, 20% glycerol, 1% bromophenol blue). Proteins (20 μg/sample) were separated by 10% SDS-PAGE, and transferred onto PVDF (Amersham Hybond-P). Blots were blocked in TBST-BSA 1% and incubated overnight with the primary antibody at 4°C. The following day, membranes were washed and then incubated 1 h with the secondary antibody at room temperature before washes. Detection was performed using ECL Plus method (Amersham).

Plasmid Construction

The Mus musculus Doublecortin cDNA sequence is available under GenBank Accession no. NM010025 and genomic DNA sequence available under GenBank Accession no. BX530055, (Sanger Institute, Cambridge United Kingdom).

Bacterial Colony PCR-based screening was performed on the Library clone RP23-377E2 Bacterial Artificial Chromosome (BAC) (Children's Hospital Oakland Research Institute, California), using the Elongase Enzyme Mix (Invitrogen-Gibco) and three different 5'-primers (pdcx forward F1, F1.2, F2) and one 3'-primer (pdcx reverse):

pdcx F 1: 5'-TTTGTCTCTCTCAGCCTCGG-3'

pdcx F 1.2: 5'-TTCTTAGGTGCTGCTTTCCC-3'

pdcx F 2: 5'-ACTGACCTCTGTTCAGTTCC-3'

pdcx R: 5'-GTTTTCTGCTGGTTGGGTG-3'

Each PCR product (1 kbp, 1.2 kbp, 2 kbp) was cloned into pGEM-T easy (Promega) and subjected to DNA sequencing with T7 and SP6 oligonucleotide primers (GIGA, University of Liège, Belgium). The derived sequence was confirmed by comparison with the mouse genomic DNA obtained from C57Bl/6J mice and available under NCBI nucleotide bank Accession no. BX530055. Different fragments were cut with Pst1, HindIII ( for pdcx 249bp) and Apa1 restriction enzymes (New England Biolabs, Ipswich, MA, USA) and cloned into the pEGFP1 vector (BD/Clontech, Heidelberg, Germany). Integrity of the putative Dcx gene regulatory sequences of Dcx gene, pdcx-(1 kbp, 1.2 kbp, 2 kbp)-eGFP was confirmed by sequencing (Génome Express, Meylan, France).

(RT)-PCR

PCR products were analyzed on agarose gel (Eurogentec) and quantified using Image Master 1D Prime v3.01 program (Amersham). The results are shown as the relative amount of the mRNA of interest relative to the Gapdh housekeeping gene mRNA.

Site-directed mutagenesis

Ex vivo electroporation

Plasmids were prepared using the EndoFree Plasmid Kit (Qiagen, Hilden, Germany). E14.5 pregnant mice were anesthetized by gas and euthanized by cervical dislocation. The embryonic chain was removed from the mother, the embryos were isolated from the amniotic sac and decapitated. DNA was microinjected into lateral ventricles of isolated embryonic mouse heads placed in ice-cold L-15 medium supplemented with 3% glucose (1 M), 2.6% sodium bicarbonate (1 M) and 1% penicillin/streptomycin (100×). For DNA microinjection (using the FemtoJet apparatus, Eppendorf AG, Hamburg, Germany), 75-mm glass capillary tubes were pulled and tips were broken. Plasmid solutions were stained with Fast Green solution (0.05%) to monitor injection sites. Electroporations were performed on whole heads (skin and skull intact) using an ECM 830 electroporator (BTX) and the following parameters: five 50 ms long pulses separated by 1s long intervals at 50 V. After pulse delivery, the embryo heads were immersed in ice-cold L-15 supplemented with 3% glucose (1 M), 2.6% sodium bicarbonate (1 M) and 1% penicillin/streptomycin, brains were extracted and transferred into liquid 3% low melting agarose (37°C; Bio-Rad Laboratories, Hercules, CA, USA) and incubated on ice for 1 h. Embedded brains were cut coronally (250 μm) with a vibratome (VT1000S, Leica). Brain slices were transferred and maintained in organotypic slice cultures on sterilized culture plate inserts (0.4-μm pore size; Millicell-CM, Millipore Billerica, MA, USA). Brain slices were maintained in semi-dry conditions in wells containing Neurobasal medium supplemented with 2% B27, 1% N2, 1% penicillin/streptomycin. After 2 or 3 days, slices were fixed for 30 min in 4% PFA at 4°C and incubated overnight in sucrose solution (20%). The following day, slices were sectioned using a cryostat at 14 micron-thickness (LEICA CM3050S) and stained for immunofluorescence as previously described.

Statistical analysis

Statistical analyses were performed by one-way ANOVA followed by Dunnett's post hoc tests, using a GraphPad Prism program (GraphPad, San Diego, CA, USA). Each experiment was performed in triplicates and repeated on at least three different occasions. Individual comparisons are expressed as mean ± SD.