Retinoids and glucocorticoids target common genes in hippocampal HT22 cells

Authors

  • Julie Brossaud,

    1. INRA, Nutrition et Neurobiologie Intégrée, UMR1286, Bordeaux, France
    2. Université de Bordeaux, Nutrition et Neurobiologie Intégrée, UMR 1286, Bordeaux, France
    Search for more papers by this author
    • These authors equally participated in this work.
  • Hélène Roumes,

    1. INRA, Nutrition et Neurobiologie Intégrée, UMR1286, Bordeaux, France
    2. Université de Bordeaux, Nutrition et Neurobiologie Intégrée, UMR 1286, Bordeaux, France
    Search for more papers by this author
    • These authors equally participated in this work.
  • Marie-Pierre Moisan,

    1. INRA, Nutrition et Neurobiologie Intégrée, UMR1286, Bordeaux, France
    2. Université de Bordeaux, Nutrition et Neurobiologie Intégrée, UMR 1286, Bordeaux, France
    Search for more papers by this author
  • Véronique Pallet,

    1. INRA, Nutrition et Neurobiologie Intégrée, UMR1286, Bordeaux, France
    2. Université de Bordeaux, Nutrition et Neurobiologie Intégrée, UMR 1286, Bordeaux, France
    Search for more papers by this author
  • Anabelle Redonnet,

    1. INRA, Nutrition et Neurobiologie Intégrée, UMR1286, Bordeaux, France
    2. Université de Bordeaux, Nutrition et Neurobiologie Intégrée, UMR 1286, Bordeaux, France
    Search for more papers by this author
    • These authors equally participated in this work.
  • Jean-Benoît Corcuff

    Corresponding author
    1. Université de Bordeaux, Nutrition et Neurobiologie Intégrée, UMR 1286, Bordeaux, France
    • INRA, Nutrition et Neurobiologie Intégrée, UMR1286, Bordeaux, France
    Search for more papers by this author
    • These authors equally participated in this work.

Address correspondence and reprint requests to Jean-Benoît Corcuff, Université de Bordeaux, Nutrition et Neurobiologie Intégrée, UMR 1286, F-33076 Bordeaux, France. E-mail: jean-benoit.corcuff@u-bordeaux2.fr

Abstract

Vitamin A metabolite retinoic acid (RA) plays a major role in the aging adult brain plasticity. Conversely, chronic excess of glucocorticoids (GC) elicits some deleterious effects in the hippocampus. We questioned here the involvement of RA and GC in the expression of target proteins in hippocampal neurons. We investigated proteins involved either in the signaling pathways [RA receptor β (RARβ) and glucocorticoid receptor (GR)] or in neuron differentiation and plasticity [tissue transglutaminase 2 (tTG) and brain-derived neurotrophic factor (BDNF)] in a hippocampal cell line, HT22. We applied RA and/or dexamethasone (Dex) as activators of the pathways and investigated mRNA and protein expression of their receptors and of tTG and BDNF as well as tTG activity and BDNF secretion. Our results confirm the involvement of RA- and GC-dependent pathways and their interaction in our neuronal cell model. First, both pathways regulate the transcription and expression of own and reciprocal receptors: RA and Dex increased RARβ and decreased GR expressions. Second, Dex reduces the expression of tTG when associated with RA despite stimulating its expression when used alone. Importantly, when they are combined, RA counteracts the deleterious effect of glucocorticoids on BDNF regulation and thus may improve neuronal plasticity under stress conditions. In conclusion, GC and RA both interact through regulations of the two receptors, RARβ and GR. Furthermore, they both act, synergistically or oppositely, on other target proteins critical for neuronal plasticity, tTG and BDNF.

Abbreviations used
BDNF

brain-derived neurotrophic factor

Dex

dexamethasone

GC

glucocorticoids

GRE

glucocorticoids responsive element

GR

glucocorticoid receptor

MR

mineralocorticoid receptor

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide

RARE

retinoic acid responsive element

RA

retinoic acid

RAR

retinoic acid receptor

RXR

retinoic X receptor

tTG

tissue transglutaminase

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling

Retinoic acid (RA) plays a major role in adult brain plasticity through various processes such as neurite outgrowth and neuronal differentiation but, unfortunately, retinoid signaling decreases with age (Etchamendy et al. 2001; Maden 2007; Mingaud et al. 2008; Shearer et al. 2012). RA main effects are mediated by the formation of heterodimers of RA receptors (RARα, β, or γ) and retinoid-X receptor (RXRα, β, or γ) which bind RA-responsive elements in the regulatory region of target genes. On the other hand, glucocorticoids (GC) elicit harmful cellular and behavioral effects on the hippocampus (Alfarez et al. 2002; Krugers et al. 2006; Sousa et al. 2008). These effects are mediated by GC binding to glucocorticoid or mineralocorticoid receptors (GR and MR, respectively) eliciting receptors nuclear translocation (Bamberger et al. 1996). Following activation, GR forms homodimers that interact with target genes GC responsive elements. Interestingly, both RA and GC regulate genes involved in the cerebral plasticity and memory processes. Moreover, in neuronal cells simultaneous activation of RA and GC signaling pathways results in synergic or opposite effects, suggesting that their pathways interact. In thymocytes, RA enhances GC-induced cell death by improving GC-induced transcriptional activity (Toth et al. 2011). In skeletal muscle, RA strongly inhibits the expression of a GC-induced gene by inhibiting GR transactivation (Aubry and Odermatt 2009). In hepatocytes, dexamethasone (Dex) enhances the RA-dependent increase of RARβ expression (Yamaguchi et al. 1999). Conversely, in human myeloma cells, Dex inhibits RA-dependent induction of tissue transglutaminase (tTG) (Lefebvre et al. 1999). Despite the known involvement of RA and GC in neuronal plasticity, there is no data concerning possible interactions of their pathways in the hippocampus.

We questioned the involvement of RA and GC in the expression of target genes and proteins involved either in the signaling pathways or in neuron differentiation and plasticity. The receptors involved were investigated: RAR, RXR, GR, and MR. Indeed, for instance, the expression of RARβ is RA-dependent through a positive feedback loop (Ballow et al. 2003; Latasa and Cosgaya 2011). Two target proteins were also chosen to investigate neuronal plasticity. First, the tissue transglutaminase 2 (tTG) was chosen because of its involvement in cell survival and differentiation, and its dependence on RA and GC control (Lefebvre et al. 1999; Campisi et al. 2008; Fahey et al. 2009; Garabuczi et al. 2013; Hodrea et al. 2012). Second, brain-derived neurotrophic factor (BDNF) was chosen because GC down-regulates BDNF expression in the hippocampus (Murakami et al. 2005; Duman and Monteggia 2006), and conversely, RA enhances BDNF expression in midbrain cells (Katsuki et al. 2009).

Our results confirm the involvement of RA- and GC-dependent pathways in our neuronal cell model. First, both pathways regulate the transcription and expression of their own and reciprocal receptors. Second, upon simultaneous activation, Dex did not always oppose RA action in the same way. Dex reduced the expression of tTG when associated with RA, whereas it stimulated its expression when used alone. Conversely, Dex reduced BDNF expression, both when used alone and combined with RA.

Materials and methods

Cell cultures

Immortalized brain cell lines that retain parental cell characteristics have been generated from neuron/glial precursors, astrocytes, and microglia (Lendahl and McKay 1990). HT-22 cells are immortalized mouse hippocampal-neuronal precursor cells that were subcloned from their parent HT-4 cells and keep the parents' characteristics (Liu et al. 2009a). We choose this cell line rather than the other main hippocampal line, H19-7, of rat origin because we wished to more easily relate to prior work with mice in the laboratory. The clone of HT22 cells was kindly provided by Dr. E. Maronde (Frankfurt am Main, Germany) (Benz et al. 2010). To allow cells to expand, they were grown under 5% CO2 at 37°C in (Dubelcco's modified Eagle medium, Life Technologies, Van Allen Way Carlsbad, CA, USA) with pyruvate supplemented with 10% fetal bovine serum (Life Technologies) and 1% streptomycin sulfate/phenoxypenicilinic acid (stock solution 50 mg/mL and 10 000 UI/mL, respectively). The culture medium was changed every other day (including during experiments).

The cells were seeded in plates (density: 1500 cells/cm² except for immunocytochemistry) and cultured for 4 days in Dubelcco's modified Eagle medium with antibiotics and pyruvate supplemented with 10% charcoal-depleted fetal bovine serum. Except for the dose–response experiments (10−10 to 10−6 M), treatments consisted in all-trans RA (final concentration 10−6 M), Dex (final concentration 10−6 M), or RA+Dex (final concentrations 10−6 M for both). RA and Dex stock solutions (Sigma Aldrich, St. Louis, MO, USA; 17.5 and 10 mmol/L, respectively) were diluted into ethanol:dimethylsulfoxide (50 : 50) (VWR International, West Chester, PA, USA) added as vehicle in controls.

MTT assay

Cell viability was determined using a colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Briefly, 50 μL of MTT (0.5 mg/mL) were added in each well and incubated for 2 h at 37°C. The supernatants were removed and the formazan crystals were allowed to dissolve in 200 μL dimethylsulfoxide. The absorbance value was determined at 595 nm using a microplate spectrophotometer (Victor Multilabel plate reader; Perkin Elmer, Waltham, MA, USA).

In situ cell death detection

In situ nuclear DNA fragmentation was measured according to a method based on 3′OH end labeling of DNA breaks with deoxyuridine terminal deoxynucleotidyl transferase. Apoptosis was quantified with an in situ cell death detection kit (Roche, Boulogne-Billancourt, France). The cells were then washed with phosphate buffer saline (PBS) and peroxydases were blocked with 3% H2O2 in methanol. The cells were fixed (paraformaldehyde, 4%, 15 min) and permeabilized (1%Triton-X100 in PBS, 3 min). Two negative and one positive controls were included in each experiment. For the negative controls, fixed and permeabilized cells were incubated with the kit label solution instead of Terminal deoxynucleotidyl transferase dUTP Nick End Labeling (TUNEL) reaction mixture (label solution + enzyme solution). For the positive control, cells were incubated with DNaseI recombinant, grade I (30 U/mL in Tris-HCl pH7.5 50 mM, MgCl2 10 mM, bovine serum albumin 1 mg/mL) (10 min). TUNEL reaction mixture was added (60 min, 37°C, dark humid atmosphere). After rinsing, cells were analyzed using an epifluorescence microscope (Nikon H600L, Champigny sur Marne, France).

Real-time PCR analyses

The cells were lysed with an extraction kit (Trizol reagent; Invitrogen, Van Allen Way Carlsbad, CA, USA). RNA concentration was determined by spectrophotometry on a NanoDrop ND-1000 spectrophotometer (Labtech, Palaiseau, France). RNA integrity was verified using the RNA-6000 NanoLabChip kit combined with a 2100 Bioanalyser (Agilent Technologies, Les Ulis, France). RNA integrity numbers (RIN) were greater than 8 indicating a good and comparable RNA quality across samples. RNA (1 μg) was reverse transcribed to cDNA using ImPromII reverse transcriptase (Promega, Charbonnières, France) (Bonnet et al. 2008). Total RNA mixed with RNAsin (Promega) and DNase (Roche) were incubated at 37°C (15 min). OligodT and random primers (Promega) were added and incubated at 75°C (10 min). The reverse transcriptase reaction was performed at 42°C (60 min, final volume 20 μL).

Real-time quantitative PCR was performed using the LightCycler® 480II system (Roche). To detect target genes amplification products, a LC480 SYBER-GREEN I Master was used. PCR was performed in microtiter plates in a final volume of 20 μL containing 1X LC480 SYBER-GREEN I Master solution, 0.5 μM of each primer and 5 ng of cDNA. Specific primers were as follows: GAACATCATCCCTGCATCCA forward and CCAGTGAGCTTCCCGTTCA for Glyceraldehide 3 phosphate deshydrogenase (GAPDH); GCTGGGCAAGTACACTACGAAC and GGCGAACTCCACAGTCTTAATG for RARα; CAGCTGGGTAAATACACCACGAA and GGGGTATACCTGGTACAAATTCTG for RARβ; CCCAAGGATGCTGATGAAAATC and GCCCTTTCTGCTCCCTTAGTG for RARγ; CCATCTTTGACAGGGTGCTAACA and ATCTGCATGTCACGCATCTTAGAC for RXRα; CCTGAAGATGTGAAGCCACC and CGTTGACGCTCCTCCTGAAC for RXRβ; CTCTGGTGAAACACATCTGTGCC and GGGGTATACCTGGTACAAATTCTGA for RXRγ; GTGGAAGGACAGCACAATTACCT and GCGGCATGCTGGACAGTT for GR; GCCGTGGAAGGACAACACA and CCTAAGTTCATGCCGGCTTG for MR; AACAGCAACCTGCTCATCGAGTAC and TTCGCTCTTCTCCTGTGGTGTGGG for tTG; AACCATAAGGACGCGGACTTG and TTGACTGCTGAGCATCACCC for BDNF. All these primers were generated from the respective mRNA sequences obtained from the National Center for Biotechnology Information (NCBI) gene bank. The following program was used: initial denaturation step for 10 min at 95°C, amplification for 45 cycles (10 s denaturation at 95°C, 6s annealing at 62°C, 10 s polymerization at 72°C), and melt-curve analysis (5 s at 95°C, 1 min at 65°C and 97°C, 0.1°C/s). The specificity and identity of the amplified products were verified as follows: (i) the melting curve analysis showed a single melting peak after amplification, and (ii) amplified products for each gene were verified by sequencing with the Big Dye Terminator v1.1 (Applied Biosystems, Foster City, CA, USA). The GAPDH housekeeping gene was used as reference gene for relative quantification, and sample equality was verified by analyzing the expression of GAPDH. The results were normalized by the ratio of the relative concentration of the target to that of GAPDH in the same sample. Quantification of the data was performed using the LightCycler480 Relative Quantification software (version 1.5). To compensate for differences in target and reference gene amplification efficiency, either within or between experiments, this software provides a calibrator–normalized relative quantification including a PCR efficiency correction. Therefore, the results are expressed as the target/reference ratio divided by the target/reference ratio of the calibrator. In our case, the calibrator was chosen among the cells in control conditions. We verified that the expression level of the reference gene GAPDH was unaffected by our different treatments.

Western blots

Protein extraction, electrophoresis, and transfer were performed as described (Mingam et al. 2008). The cells were washed with ice-cold PBS, scraped off, and centrifuged (285 g, 5 min, 4°C). The cell pellets were crushed in a lysis buffer [Tris-HCl pH7.5 20 mM, EDTA 1 mM, MgCl2 5 mM, Dithiothreitol (DTT) 1 mM, NaOV 1 mM, NaF 1 mM, and a protease mix inhibitor (Sigma-Aldrich P2714)] on ice (incubation 30 min). The samples were centrifuged (13 000 g, 20 min, 4°C). Protein concentration was assessed by bicinchoninic acid protein assay (Uptima, Montluçon, France). Proteins (40 μg) were loaded on Sodium dodecyl sulfate polyacrylamide gel electrophoresis gels (SDS-PAGE gels) (10%), transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, MA, USA) and incubated overnight with primary and 1 h with appropriate secondary antibodies. Primary antibodies were diluted as follows: rabbit anti-actin (Sigma Aldrich, A2066) 1 : 2500; mouse anti-RARβ (Santa Cruz Biotechnology, Dallas, Texas, USA, sc-56864) 1 : 500; rabbit anti-GR (Santa Cruz Biotechnology, sc-1004) 1 : 10 000; mouse anti-tTG 2 (Abcam, Cambridge, UK, ab2386) 1 : 500; rabbit anti-BDNF (Abcam, ab108383) 1 : 5000. Secondary horseradish peroxidase-conjugated antibodies were diluted as follows: donkey anti-mouse (Jackson Immunoresearch, Westgrove, PA, USA) 1 : 5000; donkey anti-rabbit (Jackson Immunoresearch) 1 : 5000. The blots were developed using Western Lighting Chemiluminescence Reagent Plus (PerkinElmer). Actin was used as a housekeeping gene. The density of the band on the membranes was quantified with a Syngene (Saint Quentin en Yvelines, France) detection system.

Quantification of tTG activity

The activity of tTG was quantified on cytosolic protein by detecting the incorporation of [3H]-putrescine (PerkinElmer) into NN-dimethylcasein (Sigma Aldrich) as previously described (Piacentini et al. 1986; Alfos et al. 1996). The cells were washed with ice-cold PBS, scraped off, and centrifuged (285 g, 5 min, 4°C), lysed in an extraction buffer (saccharose 0.25 M, trisma pH7 50 mM, EDTA 1 mM), and centrifuged (105 000 g, 1 h, 4°C). The lysate was mixed with the reaction buffer (Trisma buffer pH 8.3 50 mM, CaCl2 5 mM, DTT 10 mM, NaCl30 mM, NN'-dimethylcasein 3 mg/mL, putrescine 0.2 mM and [3H]-putrescine 1 μCi to 10-300 μg of protein) and incubated for 30 min, at 37°C, in a shaking water bath. Hundred microliter samples were spotted on Whatman 3MM filter paper moistened with 20% trichloroacetic acid (TCA). Free [3H]-putrescine was eliminated by washing with large volumes of cold TCA 5% containing KCI 0.2 M. Filters were transferred into vials with scintillation fluid and radioactivity was quantified using a βcounter.

Enzyme-linked immunosorbent assay of BDNF concentration

BDNF concentrations in the cell culture medium were quantified following manufacturer's instructions with the ChemiKine™ BDNF Sandwich ELISA Kit (Millipore).

Immunolocalization and immunofluorescence quantification

After 4 days treatments (seeding density: 750 cells/cm²), the cells were fixed (paraformaldehyde 4%, 15 min) and permeabilized (Triton X-100 1%, 3 min). Specificity was prevented by incubation (1 h) with PBS/bovine serum albumin (3%). tTG, BDNF, RARβ, and GR were detected using corresponding antibodies diluted as follows: mouse anti-tTG2 (Abcam, ab2386) 1 : 50; rabbit anti-BDNF (Abcam, ab108383) 1 : 100; mouse anti-RARβ (Santa Cruz Biotechnology, sc-56864) 1 : 50; mouse anti-RARβ (Abcam, H00005915-B01P), rabbit anti-GR (Santa Cruz Biotechnology, sc-1004) 1 : 50; incubation time was 3 h. The cells were subsequently incubated with appropriate secondary Alexa fluor antibodies for 1 h (Molecular Probes, Van Allen Way Carlsbad, CA, USA, A11005, A21206, A11005, A21206; 1 : 1000). Next, cells were observed using an epifluorescence microscope (Leica AF DMI6000, Nanterre, France). The results were analyzed and quantified using the Metamorph software (Molecular Device, St. Grégoire, France).

Background was subtracted in each treated image using Metamorph software. All images were thresholded (using inclusive threshold). Cytosolic and nuclear compartments were distinguished by mounting the blades on slides with Vectashield with, 6-diamidino-2-pnenylindole (DAPI) (data not shown). Antibody specificity in immunochemistry was verified with data obtained from a commercial supplier or by previous publications (tTG (Scarpellini et al. 2009), BDNF (commercial supplier of www.labome.com), GR (Mikkonen et al. 2010). We used two antibodies directed against RARβ (one polyclonal and one monoclonal); both elicited similar staining.

Data analysis

Unless otherwise indicated, all data were expressed as mean ± SEM, calculated for at least three independent experiments. The statistical significance of the differences between multiple groups was determined using the non-parametric Kruskal–Wallis' test. When the statistic was associated with a p < 0.05 probability, intergroup comparisons were conducted using the Mann–Whitney U-test.

Results

RA and Dex modify RARβ, GR, tTG and BDNF, mRNA expression

RARα, RARβ, RARγ, RXRα, RXRβ, RXRγ, MR, GR, tTG, and BDNF mRNA were quantified after a 4d treatment with RA or Dex, concentrations from 10−10 to 10−6 M.

RA significantly increased RARβ mRNA expression even at the lowest RA concentration (more than fivefold at 10−6 M) (Fig. 1a, left panel). Dex increased RARβ mRNA expression only for the highest concentrations (10−7 and 10−6 M, more than threefold at 10−6 M) (Fig. 1a, right panel).

Figure 1.

Effects of retinoic acid (RA) and dexamethasone (Dex) on retinoic acid receptor (RAR)β, glucocorticoid receptor (GR), tissue transglutaminase (tTG), and brain-derived neurotrophic factor (BDNF) mRNA expression. RARβ (a), GR (b), tTG (c), BDNF (d) mRNA expressions upon increasing concentrations of RA or Dex (real-time PCR analyses, see Material and Methods). All mRNA were significantly modulated by RA or Dex treatments (10−6 M) except for BDNF in Dex-treated cells. Mean ± SEM from at least three independent experiments. *Significantly different from the control (p < 0.05).

RA did not significantly affect RARα, RXRα, and RXRβ mRNA expression even at 10−6 M but increased RARγ mRNA expression (increase: 54.7 ± 6.3% at 10−6 M) (not shown). Dex did not affect RXRβ mRNA expression but significantly decreased RARα, RARγ, and RXRα mRNA expression at the highest concentration (decreases: 9.5 ± 3.4, 10.9 ± 3.4, and 8.4 ± 5.8%, respectively) (not shown). Finally, RXRγ mRNA was not detectable, regardless of conditions (not shown).

RA significantly decreased GR mRNA expression (decrease: 14.3 ± 3.3%) (Fig. 1b, left panel). Dex also decreased GR mRNA expression at the highest concentrations (10−7 and 10−6 M; decreases: 21.9 ± 8.5% at 10−6 M) (Fig. 1b, right panel). RA and Dex both increased MR expression (increases: 39.8 ± 8.0 and 45.5 ± 5.6%, respectively, at 10−6 M) (data not shown).

RA significantly increased tTG mRNA expression in a dose-dependent manner (10 to 100-fold) (Fig. 1c, left panel). An increased tTG mRNA expression was also observed with Dex, but only for the highest concentrations (10−7 and 10−6 M, about three-fold for 10−6 M Dex) (Fig. 1c, right panel).

RA significantly increased BDNF mRNA expression (about two-fold at 10−6 M), but Dex did not modify BDNF mRNA expression, whatever concentration was used (Fig. 1d).

RARβ, GR, and tTG mRNA were quantified after different durations of treatment (RA and/or Dex 10−6 M; 0.5 h, 1 h, 3 h, 1 day, and 4 day) (Fig. 2). As BDNF is a late gene, its expression was quantified at 24 h and 96 h). Some specific effects of RA alone versus Dex or RA+Dex were only seen after 96 h for RARβ and BDNF. On the basis of these results, a 96 h treatment and 10−6 M RA and Dex concentrations were selected to stimulate their respective pathways in our cell model.

Figure 2.

Time-responses of retinoic acid receptor (RAR)β (a), glucocorticoid receptor (GR) (b), tissue transglutaminase (tTG) (c), and brain-derived neurotrophic factor (BDNF) (d) mRNA expression quantified by real-time PCR analyses. Effects of retinoic acid (RA) (10−6 M; black), dexamethasone (Dex) (10−6 M; light gray) or RA+Dex (dark gray) on retinoic acid receptor (RAR)β, GR, tTG, and BDNF mRNA expression. In this study, 96 h treatments best highlighted the specific effects of the three treatments on BDNF and RARβ expressions. Results are mean values ± SEM from at least three independent experiments. *Significantly different from the control (cells treated with vehicle); #significantly different from Dex treatment effect; ¤ significantly different from RA treatment. (p < 0.05).

HT22 cells viability, apoptosis and necrosis

After 4 day treatments by RA (10−6 M) and/or Dex (10−6 M), a MTT assay was performed to evaluate the proportion of viable cells under the different conditions. There was no significant difference of cell viability between control and RA- or Dex- or RA+Dex-treated cells (Fig. 3a).

Figure 3.

Quantification of apoptotic and necrotic cells after treatment with retinoic acid (RA) or dexamethasone (Dex). Results of the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay on RA (10−6 M; black) Dex (10−6 M; light gray) or RA+Dex (dark gray) treated cells. No significant difference of cell viability was evidenced between control and treated cells (a). Apoptotic and necrotic cells detection (white arrows) after treatment with RA (10−6 M) (b). In situ nuclear DNA fragmentation was measured by the Terminal deoxynucleotidyl transferase dUTP Nick End Labeling (TUNEL) technique. The image is representative of at least three separated sets of culture. Scale bar: 100 μm. The percentage of TUNEL-positive cells was determined in triplicate counts of 200 cells treated by RA (10−6 M) or Dex (10−6 M) (c). There was no significant difference between the rate of TUNEL-positive cells in control and RA- and/or Dex-treated cells. Mean ± SEM from at least three independent experiments.

Cell apoptosis and necrosis were evaluated by TUNEL after the same treatments (Fig. 3b and c). Control TUNEL-positive cells corresponded to 2.6 ± 0.6% total cells. There was no significant difference between the percentage of TUNEL-positive cells in control and RA- and/or Dex-treated cells.

Modulation of RARβ expression, abundance, and localization by RA or Dex treatment

As RARβ was the most affected RA receptor by RA or Dex (Fig. 1a) and as RARβ is a major mediator of RA (Ballow et al. 2003; Latasa and Cosgaya 2011), we focused on their regulation of RARβ.

The RARβ mRNA expression was significantly increased by RA, Dex, or RA+Dex although the increase elicited by Dex alone (about threefold) was significantly lower than the one observed with RA or RA+Dex (Fig. 4a).

Figure 4.

Consequences of retinoic acid (RA) or glucocorticoid pathways activations in retinoic acid receptor (RAR)β regulation. Quantifications of RARβ expression (a), by real-time RT-PCR and abundance by immunofluorescence quantification on plated cells (b), after 4d treatments with RA (10−6 M) or dexamethasone (Dex) (10−6 M). Dex increased in RARβ mRNA and protein expression but less than with RA or RA+Dex. RA-Dex led to an intermediate level of RARβ abundance. Mean ± SEM from at least three independent experiments. *Significantly different from the control, # from Dex-treated cells; ¤ from RA-treated cells (p < 0.05). Immunocytochemical localization of RARβ (c) in untreated and RA- and/or or Dex-treated cells: for all conditions, RARβ was concentrated around the nucleus with a decreasing gradient toward the cell periphery. The images are representative of at least three separated sets of culture. Scale bar: 20 μm.

Immunofluorescence quantification on plated cells showed that RA and Dex increased RARβ compared to control (increases: 123.6 ± 16.6% and 19.5 ± 4.0%, respectively) (Fig. 4b). RA+Dex led to an intermediate increase in RARβ abundance (34.7 ± 6.1%).

The subcellular localization of RARβ investigated by immunocytochemistry showed that in control as well as RA- and RA+Dex-treated cells RARβ was mainly present around the nucleus with a decreasing gradient toward the cell periphery (Fig. 4c). This gradient was lower in Dex-treated cells where the cytoplasmic localization in these cells was more homogenous (Fig. 4c).

Modulation of GR expression, abundance and localization by RA or Dex treatment

The GR mRNA expression was decreased by RA, Dex and RA+Dex (decreases: 18.3 ± 4.7, 20.9 ± 0.1 and 24.9 ± 1.2%, respectively) (Fig. 5a). Moreover, the RA+Dex-dependent decrease was significantly greater than that because of RA or Dex treatments alone.

Figure 5.

Consequences of retinoic acid (RA) or glucocorticoid pathways activations in glucocorticoid receptor (GR) regulation. Quantification of GR: mRNA expression by real-time RT-PCR (a) and abundance by western blots (b), after 4 day treatments with RA (10−6 M) or dexamethasone (Dex) (10−6 M). RA+Dex treatment led to a greater decrease of GR mRNA level than RA or Dex treatments. GR abundance was significantly decreased in cells treated with Dex or RA+Dex. Mean ± SEM from at least three independent experiments. *Significantly different from the control, # from Dex-treated cells; ¤ from RA-treated cells (p < 0.05). Immunocytochemical localization of GR (c) in untreated and RA- and/or Dex-treated cells: in control cells and in cells RA-treated, GR was cytoplasmic. It translocated into the nucleus with Dex and RA+Dex. The images are representative of at least three separated sets of culture. Scale bar: 20 μm.

The GR abundance was quantified by Western blots (Fig. 5b). It was not affected by RA, but significantly decreased by Dex and RA+Dex when compared to control (decreases: 41.3 ± 5.8 and 25.2 ± 10.6%, respectively). There was no statistically significant difference between the effect of Dex and RA+Dex.

The GR subcellular localization showed that in control or RA-treated cells, GR displayed a perinuclear localization (Fig. 5c). In Dex- and RA+Dex-treated cells, GR was mainly translocated into the nucleus (Fig. 5c).

Effects of RA and Dex treatments on tTG expression, abundance, activity, and localization

The tTG mRNA expression was increased by RA (Fig. 6a) and similar results were obtained in the presence of RA+Dex (more than 100- and 70-fold, respectively). Dex increased tTG mRNA expression compared to control, but significantly less than RA or RA+Dex (eight-fold increase compared to the control).

Figure 6.

Implication of retinoic acid (RA) and glucocorticoid pathways in tTG regulation. Quantification of tTG: mRNA expression by real-time RT-PCR (a), abundance by western blots (b) and activity (c) after 4 day treatments with RA (10−6 M) or dexamethasone (Dex) (10−6 M). All treatments significantly increase tTG expression, abundance, and activity. The RA-induced increase was higher than that induced by RA+Dex, itself higher than that induced by Dex. Mean ± SEM from at least three independent experiments. *Significantly different from the control, # from Dex-treated cells; ¤ from RA-treated cells (p < 0.05). Immunocytochemical localization of tTG (d) in untreated and RA- and/or Dex-treated cells: in control cells, RA- and RA+Dex-treated cells, tTG was concentrated around the nucleus. In Dex-treated cells, tTG was uniformly distributed throughout the cells. The images are representative of at least three separated sets of culture. Scale bar: 20 μm.

The tTG abundance was significantly increased by RA (variation of 516.1 ± 51.6%) (Fig. 6b). The increase induced by Dex was more moderate but still different from control (variation 162.3 ± 23.5%). Finally, RA+Dex led to a significant intermediate increase (variation of 270.7 ± 39.9%).

The tTG activity evaluated by incorporation of [3H]-putrescine was significantly increased by RA and significantly less by RA+Dex (15699 ± 6803 and 7306 ± 225%, respectively) (Fig. 6c). The increased tTG activity observed with Dex was significant although smaller than with RA (521 ± 172%).

The tTG subcellular localization showed that in control, RA- and RA+Dex-treated cells, tTG was mainly concentrated around the nucleus with a decreasing gradient toward the cell periphery (Fig. 6d). In Dex-treated cells, tTG was localized differently; it was distributed throughout the cells (Fig. 6d).

Effects of RA and Dex treatment on BDNF expression, abundance, secretion, and localization

The BDNF mRNA expression was only modulated by RA and RA+Dex (increases: 179.7 ± 10.8 and 144.3 ± 16.2% for RA- and RA+Dex-treated cells, respectively) (Fig. 7a). Interestingly, the increase in BDNF expression was significantly higher with RA alone than with RA+Dex. Dex tended to but did not influence BDNF mRNA levels.

Figure 7.

Implication of retinoic acid (RA) and glucocorticoid pathways in brain-derived neurotrophic factor (BDNF) regulation. Quantification of BDNF: mRNA expression by real-time RT-PCR (a), secretion (b). BDNF expression was only modulated with RA or RA+dexamethasone (Dex) treatments which increase BDNF secretion. Mean±SEM from at least three independent experiments. *Significantly different from control, # from Dex-treated cells; ¤ from RA-treated cells (p < 0.05). Immunocytochemical localization of BDNF (c) in untreated and RA or Dex-treated cells: in control cells, BDNF had perinuclear localization. Treatment with RA led to BDNF migration to the cell extremities. In Dex- and RA+Dex-treated cells, BDNF was less perinuclear than in controls and more diffuse in the cytoplasm. The images are representative of at least three separated sets of culture. Scale bar: 20 μm.

The intracellular BDNF abundance assessed by Western blots was similar regardless of the treatment (data not shown). As BDNF is a secreted protein, the effect of these treatments was evaluated on the secretion of BDNF from these cells (Fig. 7b). RA and RA+Dex significantly increased BDNF secretion (157.8 ± 16.2 and 134.3 ± 5.4%, respectively). Conversely, Dex significantly decreased BDNF secretion when compared to controls (70.7 ± 5.4%).

The BDNF subcellular localization showed that in control cells, BDNF was peripheral to the nucleus in secretory vesicles (Fig. 7c). After RA, BDNF was localized in cells extensions (Fig. 7c). After Dex and RA+Dex treatments, the distribution of BDNF was less perinuclear than in control cells, more diffuse in the cytoplasm (Fig. 7c).

Discussion

We investigated genes involved in neuronal plasticity that can be used to decipher interactions between retinoids and glucocorticoids signaling pathways in hippocampal cells. As unavoidable when using cell lines in experiments, data will be best confirmed with in vivo studies. We used cells from the HT22 line because of its mouse hippocampal origin, which were of interest to us in relation with prior work on memory and aging in mouse (Etchamendy et al. 2001; Mingaud et al. 2008), and HT22 known sensitivity to glucocorticoids (Behl et al. 1997). The use of in vitro-grown fetal cells would not be suitable here because of the major role played by RA in neuronal differentiation (Maden 2007; Rhinn and Dolle 2012).

We first investigated a primary level of interaction between the retinoids and glucocorticoids signaling pathways by questioning the reciprocal influence of their ligands on the nuclear receptors. The up-regulated expression of RARβ, in agreement prior studies of neuronal or non-neuronal cells (Ballow et al. 2003; Coste and Labbe 2011; Latasa and Cosgaya 2011), is explained by the presence of a RA-responsive element in the RARβ gene (De The et al. 1990). This RA-induced increase of RARβ expression was reduced by Dex (although not significantly for RARβ mRNA). This contrasts with another study where Dex enhanced the RA-dependent increase in RARβ mRNA (Yamaguchi et al. 1999). Many negative interactions may occur, including a reduced half-life of RARβ mRNA induced by RA or Dex (George et al. 1998). Alternatively, the presence/absence of co-repressors or co-activators of RARβ and GR may explain the inverse effects of the combination of RA and Dex. Moreover, gene regulation in the brain subserves specific functions different from those of other cells. Finally, we observed RARβ in the cytoplasm as described in breast and neuroblastoma cells (Sommer et al. 1999; Dey et al. 2007; Masia et al. 2007). This suggests non-genomic pathways through: phosphatidylinositide-3-kinase (Ohashi et al. 2009), calmodulin kinase (Liu et al. 2009b), MAP kinase (Okamoto et al. 2000). This may be the basis for cross talks between RA-stimulated kinase cascades and genomic pathways to regulate myriads of proteins including the receptors themselves (Rochette-Egly 2005; Rochette-Egly and Germain 2009; Boldizsar et al. 2010; Galliher-Beckley et al. 2011).

Synaptic reorganization occurs after simultaneous MR and GR activation by endogenous glucocorticoids. Conversely, exclusive GR activation triggers neuronal abnormalities more deleterious than the dual activation (Oliveira et al. 2006; Sousa et al. 2008). Thus, we focused our attention on GR: RA and Dex decreased GR expression; the combined treatment elicited a more pronounced effect. A Dex-dependent GR expression decrease in the hippocampus was known (Reul et al. 1989; Numakawa et al. 2009). Conversely, data on down-regulation in HT22 cells seems less established. Wang et al. (2002) found that in their HT22 clone “GR levels remained relatively unchanged” after 1 μM Dex for 3 days (4 days here) contrasting with a major decrease in a hepatoma cell line. From the 1st to the 4th day in our conditions, this decrease reached a significant level but was more modest: about 40% on day 4. This could be in agreement with a negligible Dex-induced degradation of the GR protein (Wang et al. 2002) and a modest but continuous decrease of GR mRNA. A stable presence of GR in the nucleus under prolonged Dex treatment is found both by Wang et al. and ourselves. Wang et al. reported that the lack of major disappearance of GR could be related to proteasome-ubiquitine activity (Wang and DeFranco 2005). Whether this could explain these small differences between our subclones of HT22 is unknown.

Interestingly, the Dex-stimulated pathway has here a more prominent action on GR expression when co-stimulated by RA: this is a different result from the one described above for RARβ. Thus, functionally, RA could contribute to reduce the deleterious effects of GR activation through amplification of GC-elicited GR negative feedback loop. As expected, Dex-induced GR translocation into the nucleus (Sarabdjitsingh et al. 2010).

Thus, retinoids and glucocorticoids signaling pathways interact by at least reciprocally influencing the expression of their receptors, RARβ and GR. We then investigated whether other genes expressed in neurons could be targeted to and differently regulated by these pathways. We focused on genes involved in neuronal plasticity: tTG and BDNF.

The former, tTG, plays a role in neuronal differentiation and function (Facchiano et al. 2010). The stimulation of tTG expression and function under RA or Dex treatments is consistent with results obtained in hepatoma cells (Fukuda et al. 1994). Interestingly, here, the RA+Dex treatment was reduced tTG expression and function compared with RA alone. Conversely, in hepatoma cells, the RA+Dex treatment increased both RA and Dex responses. However, results similar to ours were obtained in myeloma cells (Lefebvre et al. 1999). As described above for the receptors, it is possible that depending on the origin of the cells, factors – different from the receptors and targeted by RA and Dex – could be involved.

We observed a more diffuse cytoplasmic expression in Dex-treated cells compared to controls or RA-treated cells. This localization is consistent with descriptions of cytosolic tTG in various cells (Aeschlimann et al. 1995; Akimov and Belkin 2001; Scarpellini et al. 2009). Several studies have shown that RA stimulates the expression of tTG and elicits apoptosis despite being up-regulated in various malignant tissue (Chen and Mehta 1999; Pasquali et al. 1999; Ou et al. 2000; Fok et al. 2006; Verma et al. 2006; Yuan et al. 2007). The RA and Dex concentrations used here were used previously (Lefebvre et al. 1999; Yamaguchi et al. 1999; Ballow et al. 2003; Numakawa et al. 2009; Latasa and Cosgaya 2011) and were not deleterious despite massive increase of tTG. We observed neither modification of cell proliferation nor apoptosis possibly because tTG among other roles could act as a pro- or anti-apoptotic factor depending on multiple factors (cell type, localization).

The second target protein we investigated, BDNF, is a neurotrophin, essential to neuronal growth and differentiation (Lindvall et al. 1994; Korte et al. 1998; Ernfors and Bramham 2003). Changes in BDNF levels in the hippocampus are associated with pathological conditions, such as Alzheimer's disease and depression (Nagahara et al. 2009; Yulug et al. 2009). Here, in control HT22 cells, BDNF is cytoplasmic as described in hippocampal neurons (Tang et al. 2010; Park and Loh 2011). Dex decreases BDNF expression in colon carcinoma (Kino et al. 2010), cortical neurons (Numakawa et al. 2009) or in hippocampus (Murakami et al. 2005; Duman and Monteggia 2006; Wei et al. 2010). BDNF secretion was clearly reduced by Dex in HT22. Conversely, RA-stimulated BDNF secretion is in agreement with results found in midbrain neurons (Katsuki et al. 2009; Kurauchi et al. 2011). This RA-induced secretion was reduced by Dex to be intermediate between RA-stimulated and Dex-inhibited. Thus, here, RA and Dex have opposite effects on the secretion of a neurotrophin. As seen for GR, RA may minimize the potentially deleterious effect of a GC-decrease in the neurotrophin BDNF in the hippocampus. This restoration of normal BDNF levels may improve neuronal plasticity in stress condition. This result can be related to the fact that RA administration restores adult rat hippocampal plasticity (Bonnet et al. 2008). As during aging, there are decreased bioavailability of RA and increased GC levels, RA supplementation may represent a potential therapeutic tool to minimize the deleterious effects on memory and hippocampal plasticity during aging.

Interestingly there may be some relationships between tTG activity and neurotransmitter secretion, here BDNF. Indeed, tTG inhibits BDNF processing in primary rat neurons and tTG inhibitors stimulate catecholamine release from rat brain synaptosomes (Borrell-Pages et al. 2006). However, despite an increased tTG in Dex-treated cells, BDNF secretion was decreased. Multiple mechanisms must be involved including the above-mentioned variations of tTG targets. Thus, it would be of interest to correlate the effect of RA and Dex on phenotypic changes related to neuronal plasticity. In vivo and in vitro studies investigating cell phenotype, and gene involved in neurite growth, synapse genesis, etc., are being carried out presently in the laboratory.

In this study, we studied genes that could be used to investigate interactions between retinoids and glucocorticoids signaling pathways in a hippocampal cell line to better understand the mechanisms that occur during brain aging. A first level of interaction is modifications in the amount of molecules participating in the pathways (Grummer and Zachman 1998; Yamaguchi et al. 1999). Indeed, here, RARβ and GR are regulated by their own and reciprocal ligands, generating a closed feedback loop. A second level of interaction is direct or indirect (e.g., via co-factors) relationships between these receptors. This second level co-exists here as targets not involved in the pathways (involved in neuronal plasticity) were differentially affected by RA and Dex. A direct mechanism has been described; RAR/RXR heterodimers interact with a ligated GR, resulting in an enhanced transcriptional activity of GR (Toth et al. 2011). Alternatively, indirect mechanisms have been described through the regulation of the expression of proteins (Aubry and Odermatt 2009), action of co-factors of transcription (Wang et al. 2004), kinases pathways (Rochette-Egly 2005; Rochette-Egly and Germain 2009). This neuronal model could thus be valuable to understand GC-RA interactions in the hippocampus during cerebral plasticity aging.

Acknowledgements

Many thanks to Dr. Maronde for the gift of HT22 cells. We thank Dr. Lambert (UMR 5248-France) for providing imagery on the Leica AF DMI6000 and for the use of the Metamorph software and Mrs M. Schuler for the English language corrections.

This study was supported by the Conseil Régional d'Aquitaine. J Brossaud was the recipient of a grant from Bordeaux' CHU.

Conflict of interest

The authors declare that they have no conflict of interest.

Ancillary