Glucocorticoid treatment of astrocytes results in temporally dynamic transcriptome regulation and astrocyte-enriched mRNA changes in vitro

Abstract

While general effects of glucocorticoids are well established, the specific cellular mechanisms by which these hormones exert tissue-dependent effects continue to be elaborated. Diseases that demonstrate altered glucocorticoid signaling have been associated with alterations in astrocytes, yet relatively little is known about the effects of glucocorticoids upon this cell type. We have analyzed mRNA expression patterns following glucocorticoid treatment of mouse primary astrocyte cultures. Microarray analysis of cortical astrocyte cultures treated with dexamethasone over an eight-point, 24 h time course identified 854 unique genes with ≥twofold change in mRNA expression at one or more time points. Clustering analysis associated subsets of these mRNA expression changes with gene ontology categories known to be impacted by glucocorticoids. Numerous mRNAs regulated by dexamethasone were also regulated by the natural ligand corticosterone; all of the mRNAs regulated ≥twofold by corticosterone were substantially attenuated by cotreatment with the glucocorticoid receptor antagonist RU486. Of the mRNAs demonstrating ≥twofold expression change in response to both glucocorticoids, 33 mRNAs were previously associated with glucocorticoid regulation, and 36 mRNAs were novel glucocorticoid targets. All genes tested by qPCR for glucocorticoid regulation in cortical astrocyte cultures were also regulated by glucocorticoids in hippocampal astrocyte cultures (18/18). Interestingly, a portion of glucocorticoid-regulated genes were astrocyte enriched; the percentage of astrocyte-enriched genes per total number of regulated genes was highest for the early time points and steadily decreased over the time course. These findings suggest that astrocytes in vitro may initially deploy cell type-specific patterns of mRNA regulatory responses to glucocorticoids and subsequently activate additional cell type-independent responses.

the brain is composed of a number of different cell types, resulting in complex molecular processes that underlie both normal brain function and disease pathology. While studies have historically attributed numerous functions to neurons, glial cells (e.g., astrocytes, oligodendrocytes, microglia) are increasingly identified as active participants in a range of brain functions (e.g., Refs. 17, 30). For example, astrocytes are now known to be intricate contributors to many fundamental brain processes, including energy metabolism (5), metabolite transport (25), the blood brain barrier (1), synapse regulation (20), intercellular communication and gliotransmission (54), and injury response (6). In an effort to understand the transcriptional underpinnings of these intercellular distinctions, recent efforts in brain cell-specific transcriptome analyses have identified subsets of mRNAs that are enriched in a given cell type, including astrocytes (8). These results reinforce the notion that cell type context may be an important factor in understanding cellular mechanisms that contribute to defined neural circuits and related brain functions. How cell types defined by these differences in basal gene expression transcriptionally adapt to specific stimuli may inform our understanding of both normal and disease conditions.

The molecular distinctions between neurons and glial cell types are gaining interest as additional disease states are associated with glial pathology. Some single-gene diseases have been associated with specific glial alterations, e.g., Alexander's disease with astrocytes (44), myelination disorders with oligodendrocytes (23), but recent results implicate a glial component in many complex disorders as well, e.g., psychiatric disorders (reviewed in Ref. 13); schizophrenia (68, 72), bipolar disorder (55), major depression (12). Changes in astrocyte density and size in distinct brain regions of the postmortem brains of patients with depression imply that distinct subpopulations of astrocytes are disrupted in the disorder (46, 56). Postmortem brain transcriptional profiling studies have also identified astrocyte-associated mRNAs as being altered in specific brain regions in depression (2, 4), further suggesting the relevance of transcriptional control by cell type in this disorder.

Even with growing interest in glial biology in the face of increasing evidence of disease alteration, relatively little is known about effects of common transcriptional regulators in astrocytes. Given the differences in baseline gene expression between cell types, the effects of transcriptional regulators may also vary as a function of cell type. If cell-specific transcriptional alterations contribute to disease states associated with astrocytes, then discerning how known pathological alterations influence gene transcription in a cell type-dependent fashion may enhance our understanding of disease mechanisms. For example, in the case of major depression, a well-established physiological alteration is the clinical presentation of hypercortisolemia, i.e., elevated levels of cortisol/glucocorticoids in the blood (e.g., Ref. 28), reviewed in Refs. 26, 50.

Glucocorticoids are a class of steroid hormones that mediate the effects of the hypothalamic-pituitary-adrenal axis. This system is commonly associated with stress signaling and the “fight-or-flight” response (reviewed in Refs. 11, 16). Glucocorticoids act as powerful transcriptional regulators that signal through two types of receptors, the mineralocorticoid receptor (MR) and the glucocorticoid receptor (GR). MR is a high-affinity receptor associated with mechanisms involving lower glucocorticoid concentrations (e.g., circadian rhythmicity of cortisol). GR is a modestly lower-affinity receptor that responds to higher concentrations of glucocorticoids, such as levels associated with stressful experiences. Hypercortisolemia is thought to particularly impact GR-mediated regulation. Alterations in glucocorticoid signaling are thought to contribute to some disorders, e.g., major depression (41), Alzheimer's disease (27), and be causative for others, e.g., Cushing's syndrome (51). How stress hormones influence astrocyte gene expression at the transcriptome level and how these steroid-mediated transcription patterns change across time remains largely uncharacterized.

A small number of mRNAs have been identified individually as glucocorticoid-sensitive in astrocytes, e.g., glutamine synthetase (34), glial fibrillary acidic protein (GFAP) (60), suggesting unique impact of glucocorticoids by cell type. These mRNAs have been identified using two experimental strategies: 1) in vitro studies that use astrocyte cultures to examine glucocorticoid regulation of individual mRNAs (predominant strategy, e.g., Ref. 34) and 2) in vivo studies that manipulate stress steroid levels and identify known glial-enriched mRNAs as steroid-regulated (e.g., Ref. 52). However, neither of these approaches has addressed the temporal patterning of mRNA expression changes at the transcriptome level. These approaches were thus not able to investigate associations between multiple mRNAs regulated by glucocorticoids or previously uncharacterized mRNA targets of glucocorticoid regulation in astrocytes.

Here we have characterized global mRNA expression changes induced by glucocorticoid treatment in astrocytes in vitro. Primary mouse astrocyte cell cultures were treated with glucocorticoids in a series of experiments designed to determine GR-mediated mRNA regulation over a 24 h time course. We used the synthetic GR-selective agonist dexamethasone to identify temporally regulated mRNAs. Similarly, we demonstrate that all of the mRNAs regulated ≥twofold by the natural ligand corticosterone were attenuated with cotreatment of the GR-selective antagonist RU486. Understanding glucocorticoid transcriptional regulation in astrocytes may provide additional insights into cell-specific effects of glucocorticoids in the brain.

MATERIALS AND METHODS

Reagents.

Dulbecco's modified Eagle's medium (DMEM, 11960-044), fetal bovine serum (FBS, Cat. #16000-044), antibiotic-antimycotic (Cat. #15240-062), charcoal-stripped FBS (CS-FBS, Cat. #12676-029), trypsin (Cat. #15090-046), and Hanks' balanced salt solution (HBSS, Cat. #14170-112) were obtained from Invitrogen (Gibco). Dextrose (glucose, Cat. #BP350-1) and Trypan blue (Cat. #SV3008401) were obtained from Fisher Scientific. Dexamethasone (Cat. #D-2915), corticosterone (Cat. #C2505), and mifepristone (RU486, Cat. #M8046) were obtained from Sigma Aldrich. Standard cell culture medium was 10% FBS in DMEM, 1:100 dilution of antibiotic-antimycotic, and 33 mM glucose.

Astrocyte cell culture.

Adult C57B/6 mice were obtained from Charles River. All animal procedures at the University of Michigan were approved by the University Committee on Use and Care of Animals and monitored by the Unit for Laboratory Animal Medicine. Mice were bred to obtain postnatal brain tissue. Primary astrocyte cell culture was obtained by a protocol derived from previous studies (61). Brains were extracted from P0-P2 mice (average 6 mice/litter, meninges removed, placed in cold HBSS), and cortex and hippocampus were dissected using a dissecting microscope. Tissue dissections were pooled across animals and manually triturated. The tissue was then treated with a 0.05% trypsin solution in HBSS for ∼15 min to chemically release the cells from connective tissues. The trypsin was neutralized using 10% FBS/DMEM medium, and cells were pelleted via centrifugation. The supernatant was removed, and the cells were washed using 10% FBS/DMEM medium and centrifuged again. The resulting supernatant was then removed, and the cells were resuspended in 10% FBS/DMEM. Sample cell density was determined with a hemocytometer and Trypan blue dye. Cells were then plated on uncoated plastic six-well dishes (Costar 3516) at 250,000–500,000 cells/well in 2 ml 10% FBS/DMEM medium. Cells were grown in an incubator at 37°C/5% CO₂. All media were removed 24 h after plating, and fresh media were then applied. For subsequent medium replacement, 50% of media was changed every 2–3 days. Cells were grown for 9–11 days in vitro until near confluence and then used for experiments. Cell culture derived by this protocol yielded 97% GFAP-positive cells (data not shown), implicating the majority of cells as GFAP+ astrocytes.

RNA sample generation: glucocorticoid and antagonist in vitro treatments.

On the day prior to glucocorticoid treatment, all media were removed and replaced with 10% CS-FBS/DMEM to limit residual serum steroid effects. For experiments, cells were treated with 50% fresh medium to a final concentration of 1) 100 nM glucocorticoids (corticosterone or dexamethasone), 2) 100 nM glucocorticoids + 500 nM antagonist (corticosterone + RU486), 3) 500 nM antagonist (RU486), or 4) vehicle (solution containing identical components to treatment group in a given experiment except for steroid/antagonist). For dexamethasone experiments, cells were exposed to a specific treatment for one of eight different lengths of time (0.25, 1, 2, 4, 6, 12, 18, 24 h). For corticosterone/RU486 experiments, cells were exposed to a specific treatment for 2, 4, or 6 h. For all treatment conditions examined, samples for the 2, 4, and 6 h treatments were generated as triplicate samples per group (i.e., 3 glucocorticoid treated, 3 vehicle treated); samples at other time points were generated as duplicate samples (i.e., 2 glucocorticoid treated, 2 vehicle treated). Total RNA samples were collected using TRIzol reagent per manufacturer's protocol (Invitrogen). RNA concentrations were obtained using a Nanodrop spectrophotometer (Thermo Scientific).

Microarray studies.

RNA samples were prepared using Illumina TotalPrep RNA Amplification kit (500 ng/sample, Ambion AMIL1791). Biotinylated RNA samples (1.5 μg/sample) were then applied to MouseWG-6 v2.0 Expression Beadchips (Illumina BD-201-0202) and processed per manufacturer's recommendations. Microarrays were scanned on a Illumina BeadArray Reader. Microarray data were analyzed using Illumina GenomeStudio software (version 1.1.1). GenomeStudio analyses were performed using quantile normalization with an Illumina custom error model and referenced to the corresponding vehicle group. Multiple testing corrections were applied using a Benjamini and Hochberg false discovery rate with differential expression defined as P < 0.05; P values were obtained through the GenomeStudio software. Subsequent calculations (e.g., fold-changes) were derived using Microsoft Excel. For the corticosterone studies, mRNAs were considered to be regulated by corticosterone and attenuated by RU486 by meeting one of the following statistical criteria: 1) corticosterone treatment P < 0.05 and corticosterone + RU486 cotreatment P > 0.05 or 2) corticosterone treatment P < 0.05 and corticosterone-RU486 cotreatment P < 0.05 but corticosterone-RU486 cotreatment resulted in ≥50% reduction in the magnitude of fold-change compared with corticosterone treatment. Regulated Illumina probes were manually analyzed using the National Center for Biotechnology Information (NCBI) gene database to update annotations and identify single genes associated with multiple regulated probes (http://www.ncbi.nlm.nih.gov/gene); for probes that targeted the same gene, the probe containing the highest magnitude of changes was used for downstream analyses. For Figure 1, if a cell type marker had multiple probes, the probe with the highest average signal across all samples was used and then averaged per group per time point for analysis (16 total values). All microarray data are deposited on the NCBI Gene Expression Omnibus database [accession numbers: GSE39270 (corticosterone), GSE39272 (dexamethasone)].

Fig. 1.

Cell culture system specifically expresses astrocyte mRNA cell markers. A: plot of average mRNA expression for subset of astrocyte cell markers among sample groups from 8 time points in Dexamethasone (Dex) time course. B: plot of average mRNA expression for subset of cell markers associated with additional brain cell types (neurons, oligodendrocytes, microglia). Bars for each mRNA are colored according to the cell type association. Error bars display SE.

Gene ontology analysis.

We conducted gene ontology (GO) analyses with the DAVID software database (15, 64) using Illumina probe ID notations for specified sets of mRNAs. In an effort to identify more specific GO terms (vs. broad), we performed analyses using default settings for terms of the GO biological processes (BP_FAT) and GO molecular function categories (MF_FAT).

Hierarchical clustering analysis.

Hierarchical clustering was performed using k-means clustering in R. Clustering analysis was performed on the set of all probes regulated ≥twofold at any time point by dexamethasone. Cluster number (8 clusters) was chosen based on analyzing the lowest number of clusters that produced a sum of squares statistically indifferent from immediately higher numbers of clusters.

qPCR validation studies.

The same RNA samples used for the microarray studies were also used for qPCR analyses. RNA samples (1 μg) were converted to cDNA using Superscript II via random hexamer priming (Invitrogen). We used ∼50% of each cDNA reaction for Applied Biosystems (ABI) Taqman mRNA qPCR assays in custom Low-Density Array (TLDA) format (ABI #4346799) with Taqman reagents (ABI #4440048). The Taqman arrays were processed on an Applied Biosystems Viia7 instrument. Genes measured were selected based on a series of factors (highest magnitude regulation, known/unknown glucocorticoid regulation, astrocyte enrichment, house-keeping gene expression controls). For validation experiments, genes were measured in technical triplicate per sample. Differential expression analysis was performed by the delta-delta-Ct method (Ref. 39, β-actin as control reference) using Statminer software (Integromics). An average of Ct values from technical replicates was taken as the Ct value for each gene measurement; individual Ct values identified as outliers via the Grubbs outlier test were excluded from downstream analyses. Specific Taqman mRNA assays used in this study are for the following targets: Actb (Mm01205647_g1), Adora2b (Mm00839292_m1), Aldh1l1 (Mm00550947_m1), Atp6v1b2 (Mm00431987_m1), Ch25h (Mm00515486_s1), Egr2 (Mm00456650_m1), Fkbp5 (Mm00487401_m1), Folh1 (Mm00489655_m1), Foxo1 (Mm00490672_m1), Gap43 (Mm00500404_m1), Gfap (Mm01253033_m1), Gja1 (Mm00439105_m1), Gjb6 (Mm01317508_m1), Glul (Mm00725701_s1), Klf9 (Mm00495172_m1), Mapk4 (Mm00554001_m1), Mertk (Mm00434920_m1), Pdk4 (Mm01166879_m1), Per1 (Mm00501813_m1), Phlda1 (Mm00456345_g1), Sgk1 (Mm00441380_m1), Slc1a2 (Mm00441457_m1), Slc1a3 (Mm00600697_m1), Sult1a1 (Mm01132072_m1), Syn2 (Mm00449780_m1), Txnip (Mm00452393_m1), Wnt7a (Mm00437354_m1).

Glucocorticoid regulation literature searching and astrocyte mRNA enrichment analyses.

Previously reported glucocorticoid mRNA regulation was assigned based on analysis of the literature for each gene (using PubMed and Google Scholar search tools, terms used: gene symbol and/or probe ID + “glucocorticoids, corticosteroids, corticosterone, dexamethasone, prednisone”). Astrocyte mRNA enrichment analyses were performed using in vivo astrocyte enrichment gene expression data from “Supplemental Table 4” from Cahoy and colleagues (8).

RESULTS

Astrocyte cell cultures robustly express astrocyte markers and do not express markers for other brain cell types.

Previous studies have utilized these cell culture protocols for studying astrocytes in vitro (61, 63). To confirm that our cell culture protocol successfully selected for astrocytes, we examined the mRNA expression of cell-type markers associated with different brain cell types (Fig. 1). Across all groups of the dexamethasone time course, the cell culture model expressed a number of astrocyte-specific cell markers [e.g., GFAP, connexin-43 (Gja1), GLAST (Slc1a3), aquaporin-4 (Aqp4), Aldh1l1], while exhibiting limited or no expression above background of cell markers associated with other brain cell types, i.e., connexin-47 (Cx47), myelin-associated glycoprotein (Mag), myelin oligodendrocyte glycoprotein (Mog), myelin basic protein (Mbp) (oligodendrocytes); neurofilament (Nef3), neuronal K-Cl cotransporter (KCC2/Slc12a5), synaptosomal-associated protein 25 (Snap25), neuron-specific β-tubulin (Tuj1/Tubb3) (neurons); Cd11b/Itgam, Cd45/Ptprc (microglia). These results suggest that the vast majority of cells in this culture system are indeed astrocytes. In addition, nearly identical gene expression patterns of these cell-type mRNA markers were detected in individual groups of samples (i.e., vehicle treated or glucocorticoid treated; data not shown), suggesting that glucocorticoid treatment did not affect the cell-type character of the culture system.

Dexamethasone dynamically regulates mRNAs in astrocytes over time in vitro.

To establish the transcriptome profile of primary glucocorticoid receptor regulation in astrocytes in vitro, primary astrocyte cell cultures were treated with the GR-selective synthetic agonist dexamethasone (100 nM) at 8 different time intervals spanning a 24-h time course (i.e., 0.25, 1, 2, 4, 6, 12, 18, 24 h). Microarray analyses were then performed on the resulting RNA samples to identify glucocorticoid-sensitive mRNAs in this model system (Supplemental Table S1).¹ Treatment of primary astrocyte cell cultures with the synthetic GR-agonist dexamethasone resulted in dynamic gene expression changes over a 24 h period of exposure (total 886 probes for 854 unique genes regulated ≥twofold, Fig. 2A, Supplemental Table S2a/b). The total number of mRNAs whose expression level changed ≥twofold increased with the duration of steroid exposure. In addition, the overall directionality of dexamethasone-mediated changes varied across the time course (e.g., majority of genes whose expression changed at early time points were upregulated by dexamethasone). These findings were also highly reproducible in a replicate microarray study of triplicate samples from the 2, 4, and 6 h time points based on the same statistical criteria for differential expression at similar magnitudes of regulation (≥97% overlap of mRNAs regulated ≥twofold in replicate data with mRNAs regulated ≥1.5-fold in time course data, data not shown). qPCR studies confirmed the observed microarray findings for all mRNAs tested for glucocorticoid regulation (18/18 mRNAs, Fig. 2B). The expression level of GR mRNA was similar in all data, suggesting that GR levels were consistent across astrocyte cultures (data not shown). Five astrocyte-associated genes (Slc1a3, Slc1a2, Gfap, Gja1, Aldh1l1) were not altered due to glucocorticoid treatment, indicating that observed mRNA changes were not generic to all astrocyte mRNAs (Fig. 2B).

Fig. 2.

Dex dynamically regulates specific mRNAs in astrocytes in vitro over 24 h time course. A: plot of number of genes that show >2-fold regulation by dexamethasone over time (P < 0.05). B: comparison of linear fold-change mRNA expression values of microarray data with qPCR data for select mRNAs regulated by Dex treatment (vs. vehicle). Data ordered based on magnitude of mRNA expression change in 2 h qPCR data. HR, hours; boldface = P < 0.05; italics = P > 0.05.

Hierarchical clustering analysis reveals subsets of dexamethasone-regulated genes associated with specific cellular functions.

To determine temporal patterns of gene expression changes in astrocytes due to glucocorticoid treatment, we performed a clustering analysis on the set of all mRNAs regulated ≥twofold by dexamethasone using the k-means method. Based on sum of squares analysis, the k-value that best explained the temporal variation using the least amount of clusters was an eight-cluster analysis. Of the eight clusters, five of the cluster averages/centers were upregulated (clusters 1, 2, 3, 4, 5), while three were downregulated (clusters 6, 7, 8) by the 24 h time point (Fig. 3A). Some clusters exhibited average mRNA changes that peaked at earlier time points (e.g., 4 h, 6 h) and tailed off at later time points (clusters 3, 4, 6). Other clusters exhibited more delayed glucocorticoid-mediated changes (e.g., not changed until ≥12 h) (clusters 2, 7, 8). GO analyses were then performed on the genes in each cluster using the DAVID software. GO analysis of these clusters revealed distinct cellular targets of glucocorticoid regulation based on the duration of steroid exposure. Of the eight clusters examined, the most significantly associated biological processes in four of them were highly enriched for specific cellular functions; the other four clusters were not highly correlated with any one type of biological process. The four enriched clusters were associated with 1) apoptosis (cluster 4, increase of negative regulation), 2) cofactor and ion binding (cluster 5, positive correlation), 3) DNA replication (cluster 7, negative correlation), and 4) cell cycle (cluster 8, negative correlation) (Figs. 3, B and C; genes listed in Supplemental Table S3).

Fig. 3.

Clustering analysis of Dex gene expression time course data yields distinct sets of genes that contain similar gene ontology (GO) categories. A: table of number of genes per cluster and cluster centers (average log₂ value) over the time course. B: graph of 4 GO term-associated cluster centers from k-means cluster analysis. C: table of graphed clusters and their corresponding most significant GO terms.

Dexamethasone-sensitive mRNAs also regulated by natural ligand corticosterone via the GR.

To determine if dexamethasone-regulated mRNAs were similarly regulated by the natural ligand of the GR, we treated primary astrocyte cultures with corticosterone (100 nM) for an abbreviated time course, targeting the time points of initial regulation in the dexamethasone time course (i.e., 2, 4, 6 h). Corticosterone treatment altered the expression level of numerous genes at these three time points (2 h, 28 mRNAs; 4 h, 112 mRNAs; 6 h, 73 mRNAs) (Fig. 4A). To determine if the observed mRNA expression differences were mediated through the GR, we treated parallel cultures with both corticosterone and RU486 (500 nM), a GR-selective antagonist. All 141 unique mRNAs with ≥twofold expression changes induced by corticosterone were attenuated by cotreatment with RU486 at one or more of the time points (Fig. 4A, Supplemental Table S4). qPCR data of a subset of corticosterone-regulated/RU486-attenuated mRNAs matched the directionality and magnitude of the microarray data (18/18 mRNAs, Fig. 4B). Comparison of corticosterone-regulated/RU486-attenuated mRNAs demonstrating ≥twofold expression change with mRNAs regulated ≥twofold by dexamethasone at similar time points revealed substantial but incomplete overlap (Fig. 4C), identifying more genes regulated ≥twofold by dexamethasone than corticosterone. The set of corticosterone-regulated/RU486-attenuated mRNAs as defined by statistical criterion but with the twofold change criterion removed (i.e., corticosterone P < 0.05, corticosterone + RU486 P > 0.05) contained a greater number of mRNAs regulated ≥twofold by dexamethasone (Fig. 4C). Based on this latter comparison, ∼60% of mRNAs regulated ≥twofold by dexamethasone were also regulated by corticosterone and attenuated by RU486. mRNAs that were regulated ≥twofold by corticosterone and dexamethasone and whose corticosterone regulation was attenuated by RU486 were primarily upregulated by glucocorticoid treatment (Fig. 4D). Of these 69 mRNAs regulated by both glucocorticoids, 33 have been previously associated with glucocorticoid regulation (Table 1), whereas 36 demonstrated novel glucocorticoid regulation (Table 2).

Fig. 4.

mRNAs regulated by synthetic glucocorticoid in primary astrocyte cultures are also regulated by natural ligand and the regulation is attenuated by GR antagonist. A: plot of number of genes that show >2-fold mRNA expression change due to corticosterone (Cort) that is attenuated by RU486 cotreatment (Cort/RU) (P < 0.05). B: table of qPCR validation data on select Cort-sensitive mRNAs. Genes ordered by magnitude of fold-change due to 2 h Cort exposure. C: proportional Venn diagram of overlap of genes regulated >2-fold by Dex at 2, 4, or 6 h vs. genes regulated by Cort (P < 0.05) and attenuated by RU486 (P > 0.05) at those same time points. D: graph characterizing genes regulated by both Dex and Cort ≥2-fold and also attenuated by RU486 in terms of previous literature reporting and direction of regulation. Regulated by Cort = P < 0.05, attenuated by RU = P > 0.05. FC, fold-change; boldface = P < 0.05, italics = P > 0.05.

Table 1.

Glucocorticoid-sensitive genes with reported GR-mediated regulation

Gene Symbol	Definition	Direction	Neural Reported?
Ada	adenosine deaminase	upregulated
Bcl6	B-cell leukemia/lymphoma 6	upregulated
Cbs	cystathionine beta-synthase	upregulated	+
Ccnd2	cyclin D2	downregulated
Errfi1	ERBB receptor feedback inhibitor 1	upregulated	+
Fam107a	family with sequence similarity 107 member A	upregulated	+
Fkbp5	FK506 binding protein 5	upregulated	+
Gadd45 g	growth arrest and DNA-damage-inducible 45 gamma	upregulated	+
Hdc	histidine decarboxylase	upregulated	+
Hspa1a	heat shock protein 1A	upregulated
Ipas/Hif3a	inhibitory PAS domain AKA Hypoxia-inducible factor 3	upregulated	+
Klf9	Kruppel-like factor 9	upregulated	+
Lcn2	lipocalin 2	upregulated	+
Lims2	LIM and senescent cell antigen like domains 2	upregulated
Map 3k6	mitogen-activated protein kinase kinase kinase 6	upregulated
Map k4	mitogen-activated protein kinase 4	upregulated
Mertk	c-mer proto-oncogene tyrosine kinase	upregulated	+
M gll	monogly ceride lipase	upregulated
Pdk4	pyruvate dehy drogenase kinase, isozyme 4	upregulated	+
Per1	period homolog 1	upregulated	+
Phlda1	pleckstrin homology-like domain family A member 1	downregulated
Pim3	proviral integration site 3	upregulated
Prodh	proline dehy drogenase	upregulated
Rhou	Ras homolog gene family member U	upregulated
S3-12	plasma membrane associated protein S3–12	upregulated
Sesn1	sestrin 1	upregulated
Sgk1	serum/glucocorticoid regulated kinase 1	upregulated	+
Sphk1	sphingosine kinase 1	upregulated
Sult1a1	sulfotransferase family 1A phenol-preferring member 1	upregulated	+
Tsc22d3	TSC22 domain family member 3	upregulated	+
Txnip	thioredoxin interacting protein	upregulated
Zbtb16	zinc finger and BTB domain containing 16	upregulated

GR, glucocorticoid receptor.

Table 2.

Glucocorticoid-sensitive genes demonstrating novel GR-mediated regulation

Gene Symbol	Definition	Direction
1810011O10Rik	RIKEN cDNA 1810011O10 gene	upregulated
9930004G02Rik	RIKEN cDNA 9930004G02 gene	upregulated
Accn1	amiloride-sensitive cation channel 1 neuronal (degenerin)	upregulated
Acsl3	acyl-CoA synthetase long-chain family member 3	upregulated
Axud1	AXIN1 up-regulated 1	upregulated
Bcat1	branched chain aminotransferase 1 cytosolic	upregulated
C030009J22Rik	RIKEN cDNA C030009J22 gene	upregulated
Chk	choline kinase alpha	upregulated
Cldn2	claudin 2	upregulated
E230024B12Rik	RIKEN cDNA E230024B12 gene	upregulated
Fam46b	family with sequence similarity 46, member B (Fam46b)	upregulated
Folh1	folate hydrolase 1	upregulated*
Galntl4	UDP-N-acetyl-alpha-d-galactosamine:polypeptide N-acetylgalactosaminyltransferase-like 4	upregulated
Gjb6	gap junction protein beta 6	upregulated
Gm22	gene model 22	upregulated
Got1l1	glutamic-oxaloacetic transaminase 1-like 1	upregulated
Hspb1	heat shock protein 1	upregulated
Kcnt1	potassium channel subfamily T member 1	upregulated
LOC385279	LOC385279	upregulated
Lpin3	lipin3	upregulated*
Lrrc8	leucine-rich repeat containing 8A	upregulated
Mical2	microtubule associated monoxygenase, calponin and LIM domain containing 2	upregulated
Papss2	3′-phosphoadenosine 5′-phosphosulfate synthase 2	upregulated*
Pex11a	peroxisomal biogenesis factor 11a	upregulated
Pgm5	phosphoglucomutase 5	upregulated
Rassf4	Ras association (RalGDS/AF-6) domain family member 4	upregulated
Rbpms	RNA binding protein gene with multiple splicing	upregulated
Scel	sciellin	upregulated
Slc10a6	solute carrier family 10 (sodium/bile acid cotransporter family) member 6	upregulated
Slc24a4	solute carrier family 24 (sodium/potassium/calcium exchanger) member 4	upregulated
Slc25a33	solute carrier family 25 (pyrimidine nucleotide carrier) member 33	upregulated
Smox	spermine oxidase	upregulated
Spsb1	SplA/ryanodine receptor domain and SOCS box containing 1	upregulated
Tcfcp2l1	transcription factor CP2-like 1	upregulated
Tilz3c	TSC22-related-inducible leucine zipper 3c	upregulated
Tmem200a	transmembrane protein 200A	downregulated

^*Previously reported as glucocorticoid insensitive.

mRNAs regulated by glucocorticoids in cortical astrocyte cultures are also regulated in hippocampal astrocyte cultures.

To determine if observed glucocorticoid-mediated mRNA changes were specific to astrocytes derived from cortical brain regions, additional qPCR experiments were conducted on RNA samples isolated from primary astrocyte cell cultures derived from hippocampus and treated with dexamethasone. All mRNAs tested for dexamethasone regulation by qPCR in cortical astrocyte cultures were also regulated by dexamethasone in hippocampal astrocyte cultures (18/18 mRNAs, P < 0.05) (Fig. 5).

Fig. 5.

A portion of mRNAs regulated by glucocorticoids in cortical astrocyte cultures is also regulated by glucocorticoids in hippocampal astrocyte cultures. qPCR data of select mRNAs regulated by Dex treatment in cortical astrocyte cultures that are also hippocampal astrocyte cultures. Genes ordered based on 2 h FC values. Boldface = P < 0.05, italics = P > 0.05.

Glucocorticoid-regulated mRNAs include sets of astrocyte-enriched genes.

A portion of the mRNAs regulated by glucocorticoids in these data was known to be specifically expressed in astrocytes (e.g., Glul, Gjb6; Fig. 6A). In an effort to determine how glucocorticoids might modulate unique astrocyte cell functions, mRNAs regulated by glucocorticoids were compared with mRNAs reported to be enriched in astrocytes, thus described as “astrocyte enriched” (8). On average, the astrocyte cultures expressed ≥60–70% of astrocyte-enriched mRNAs (≥1.5-fold enrichment), which represented 13–14.5% of the detected mRNAs across samples (data not shown). Many genes identified as glucocorticoid sensitive in these data are reported to be astrocyte enriched (overall percentage 23.4%, Fig. 6B). To investigate the functional implications of these findings, glucocorticoid-regulated, we analyzed astrocyte-enriched genes for GO term association (Fig. 6C). The most significant “biological processes” term associated with the broadest enrichment category (≥1.5-fold enrichment) was related to cell cycle. The biological processes terms most associated with greater astrocyte-enrichment levels (≥4-fold, ≥10-fold) included 1) sulfur metabolic processes (genes upregulated), 2) nitrogen compound biosynthesis (genes upregulated), and 3) oxidation reduction (genes upregulated) (Table 3). Among glucocorticoid-sensitive mRNAs, the percentage of astrocyte-enriched mRNAs as a function of all glucocorticoid-regulated mRNAs was higher for shorter durations of glucocorticoid exposure compared with longer durations over the time course (e.g., 2 h = 42.4%, 24 h = 22.5%); this percentage consistently trended downward over the entire time course, reaching a value at the 24-h time point that was slightly below the overall percentage for the entire time course (Fig. 6D).

Fig. 6.

Certain mRNAs regulated by Dex are astrocyte enriched. A: table of genes found to be regulated ≥2-fold by Dex and Cort in both microarray and qPCR experiments and their reported enrichment in astrocytes relative to other neural cell types. Genes arranged by magnitude of enrichment. *Cahoy et. al., (8). B: proportional Venn diagram of overlap of total number of unique Dex-regulated mRNAs with 3 subsets of astrocyte-enriched mRNAs. C: table containing most statistically significant GO category associated with 3 astrocyte-enriched mRNA subsets of Dex-regulated mRNAs in B. D: plot of percentage of astrocyte-enriched mRNAs as a function of total mRNAs regulated ≥2-fold by Dex over time.

Table 3.

GO associations of glucocorticoid-sensitive astrocyte-enriched genes

Illumina ID	Symbol	Gene Name	Astrocyte Enrichment*	Direction
GO term: sulfur metabolism
ILMN_1253588	Papss2	3′-phosphoadenosine 5′-phosphosulfate synthase 2	32.7	upregulated
ILMN_2750558	Mamdc2	MAM domain containing 2	16.2	downregulated
ILMN_1230318	Cbs	cystathionine beta-synthase	23.4	upregulated
ILMN_2669416	Mgst1	microsomal glutathione S-transferase 1	19.2	upregulated
ILMN_2776008	Gstk1	glutathione S-transferase kappa 1	4.7	upregulated
GO term: nitrogen biosynthetic process
ILMN_2722716	Atp1a2	ATPase, Na+/K+ transporting, alpha 2 polypeptide	29.8	upregulated
ILMN_1230318	Cbs	cystathionine beta-synthase	23.4	upregulated
ILMN_2636666	Prodh	proline dehy drogenase	29.4	upregulated
ILMN_1219442	Slc7a2	solute carrier family 7 (cationic amino acid transporter, y+ system), member 2	21.2	upregulated
ILMN_1243741	Rora	RAR-related orphan receptor alpha	4.8	upregulated
ILMN_2644496	Glul	glutamate-ammonia ligase (glutamine synthetase)	8.5	upregulated
GO term: oxidation reduction
ILMN_1222765	Adhfe1	alcohol dehydrogenase, iron containing, 1	22.6	upregulated
ILMN_1222734	Aldh1a1	aldehyde dehydrogenase family 1, subfamily A1	11.7	upregulated
ILMN_1214498	Cyp2d22	cytochrome P450, family 2, subfamily d, polypeptide 22	16.8	upregulated
ILMN_1251504	Cyp4v3	cytochrome P450, family 4, subfamily v, polypeptide 3	13.6	upregulated
ILMN_2766037	Dio2	deiodinase, iodothyronine, type II	56	upregulated
ILMN_1250917	Fmo1	flavin containing monooxygenase 1	19.7	upregulated
ILMN_2636666	Prodh	proline dehydrogenase	29.4	upregulated
ILMN_2644092	Aass	aminoadipate-semialdehyde synthase	6.1	upregulated
ILMN_1258158	Aldh6a1	aldehyde dehydrogenase family 6, subfamily A1	4.4	upregulated

GO, gene ontology.

^*Cahoy et al., (8).

DISCUSSION

Cell-specific mechanisms in the brain have long been of interest in an effort to understand normal neural processes and disease pathology. The goal of these studies was to characterize glucocorticoid-regulated mRNA regulation in astrocytes in vitro. Previous studies had identified genes regulated by glucocorticoids in astrocytes, but 1) mRNA regulation was observed on an individual gene basis and 2) few published reports have looked extensively at patterns of mRNA changes due to glucocorticoid exposure at time points less than 24 h. To extend current knowledge of GR-mediated transcriptional regulation to shorter durations of glucocorticoid exposure, we treated cortical astrocyte cultures with the synthetic GR agonist dexamethasone over eight time points from 15 min to 24 h. We chose a glucocorticoid concentration relevant to stress mechanisms (100 nM) to parallel 1) concentrations commonly used in other in vitro gene expression studies and 2) the range of corticosterone concentrations observed in the brain as measured by microdialysis following acute stress in rodents (e.g., Refs. 18, 59, 70).

Over the 24 h time course, we identified 854 unique genes that were regulated more than twofold at one or more time points by dexamethasone compared with vehicle control treatment. In an effort to validate the observed transcriptional regulation, we performed qPCR experiments on samples at the earliest time points that contained large numbers of regulated mRNAs (i.e., 2 h, 4 h, 6 h). All of the genes tested for glucocorticoid sensitivity by qPCR showed significant differences similar in magnitude and direction to the microarray data (18/18 mRNAs). We also observed that mRNAs previously reported to be glucocorticoid-regulated in astrocytes were regulated in manner consistent with the literature, e.g., Glul (37), Glt-1/Slc1a2 (76), MAO-B (10), Ndrg2 (69), GDH (29), with some exceptions, e.g., Fgf2 (43), GFAP (60). The exceptions may be the result of a number of factors, e.g., methods of cell preparation, species differences (e.g., mouse vs. rat), including primary vs. secondary effects of glucocorticoids based on length of glucocorticoid exposure in culture, e.g., GFAP glucocorticoid regulation sensitive to time of exposure (60). The overall consistency of our results with previous findings supports the validity of this data in terms of characterizing glucocorticoid regulation in astrocytes in vitro.

The dynamic temporal gene expression regulation by glucocorticoids observed over 24 h raises the possibility of coordinated pathway regulation. To examine relationships between the regulated mRNAs, we performed 1) a clustering analysis using eight clusters to group mRNAs according to glucocorticoid-mediated expression changes over time and 2) GO analyses on the mRNAs in each cluster. The clustering analysis identified novel temporal patterns of regulation among the glucocorticoid-regulated mRNAs. Clusters of mRNAs varied by the magnitude and rate of glucocorticoid-induced activation or repression (Fig. 3). Our results are similar to spatial patterns observed in previous reports characterizing time-dependent mRNA expression changes mediated by glucocorticoids in other cell culture systems (33, 57). From the GO analysis, the most statistically enriched biological processes changed between time points over the 24-h time course; different pathways were affected at earlier time points compared with later time points. Among the eight clusters, specific clusters associated with distinguishable sets of similar GO terms (cluster 4 = increased gene expression related to negative regulation of apoptosis, cluster 5 = increased gene expression related to cofactor binding and ion binding, cluster 7 = decreased gene expression related to DNA replication, cluster 8 = decreased gene expression related to cell cycle regulation). Negative regulation of apoptosis, DNA replication, and cell cycle pathways has often been associated with glucocorticoid regulation; glucocorticoids are known to inhibit proliferation in many cell types, including astrocytes (14). Cofactors and ions associated with mRNAs represented in the binding-related cluster included zinc (Bcl6b, Dpep1, Trp63, SceI, Cpm, Gm22, Lims2, Zhx3), pyridoxal phosphate (Cbs, Got1l1, Agxt2l1), and divalent cations (calcium: Dlk1, S100a7a, Arsj, Galntl2; magnesium: Pgm5, Acsl3, Ppm1k; iron: Cdo1, Cyp2d22). Together, these findings document that specific pathways in astrocytes are targeted by glucocorticoids across time in vitro.

To extend the observed synthetic glucocorticoid regulation to the endogenous ligand and further verify the GR-based mechanism of regulation on target mRNAs in astrocytes, we conducted microarray experiments on cortical astrocyte cell culture samples treated with 1) the endogenous ligand corticosterone and 2) corticosterone coadministered with RU486 (mifepristone), a GR-selective antagonist. Cotreatment of corticosterone and RU486 attenuated the regulation of almost all of the mRNAs regulated ≥twofold by corticosterone alone, suggesting they were regulated via GR. RU486 acted as an antagonist to corticosterone action for these robustly glucocorticoid-regulated mRNAs, consistent with reports involving glucocorticoid-antiglucocorticoid coadministration (e.g., Refs. 3, 7, 35). Although RU486 has been reported to have agonistic properties under select circumstances (e.g., Refs. 9, 53), we did not observe significant agonist activity by RU486 in these experiments (i.e., RU486 treatment alone). RU486 agonist activity has been suggested to be cell-type dependent (45). Although RU486 action has not been explicitly studied in astrocytes, the combination of GR concentration (74), corepressor expression (32, 42), and other cellular factors in astrocytes may not permit substantial RU486 agonist activity in this context.

Because both dexamethasone and corticosterone signal through the GR, we predicted the two steroids would regulate mRNAs in our astrocyte culture system in a similar manner. Previous comparative in vitro studies have found similar regulation between the two steroids for select mRNA targets (e.g., Ref. 66). Compared with dexamethasone treatment, corticosterone treatment regulated a smaller number of mRNAs ≥twofold at 2, 4, and 6 h of steroid exposure [absolute number of regulated genes: 341 (dexamethasone) vs. 141 (corticosterone)]. While a portion of the discrepancy may be due to experimental variation and astrocyte heterogeneity, this difference in the number of mRNAs regulated ≥twofold by these two glucocorticoids may be due in part to the higher GR affinity of dexamethasone vs. corticosterone (24, 35, 58). Differential responses to time-limited treatments of dexamethasone vs. corticosterone have been documented for other mRNAs (e.g., Refs. 19, 31, 40). In our experiments, the set of genes statistically regulated by corticosterone (P < 0.05) without the fold-change criteria included additional genes regulated ≥twofold by dexamethasone (133 additional genes, Fig. 4). These results are consistent with the interpretation that dexamethasone is a more robust regulator of initial mRNA expression changes than corticosterone at specific time points within our time course. The equivalence of dexamethasone and corticosterone regulation observed in previous studies may be due to longer duration of glucocorticoid exposures (e.g., 24+ h) that may overcome temporal, pharmacological distinctions between these molecules. Based on our findings, future temporal studies comparing regulation by dexamethasone and corticosterone may yield more precise comparisons of regulated mRNAs by matching physiological criteria other than absolute steroid concentration (e.g., specified IC₅₀ or EC₅₀ to GR).

We then compared the sets of 1) mRNAs regulated by dexamethasone with 2) mRNAs regulated by corticosterone and attenuated by RU486 cotreatment at any of the 2, 4, or 6 h time points. We found 69 mRNAs regulated more than twofold by both glucocorticoids that meet this criteria. Among these mRNAs, 33 had previously been associated with glucocorticoid regulation (31 upregulated, 2 downregulated; Table 1) and 36 displayed novel glucocorticoid regulation (35 up, 1 down; Table 2). Our findings extend the observed glucocorticoid regulation in other systems to astrocytes; among the mRNAs known to be regulated by glucocorticoids, all of the mRNA changes in this study were in the same direction as previously reported (data not shown). Both previously reported and the unreported glucocorticoid-regulated mRNAs may have unique actions in astrocytes; their functional roles in astrocytes remain to be investigated.

Comparisons of astrocyte gene expression between brain regions are infrequent despite the growing evidence for astrocyte heterogeneity and brain region-dependent astrocyte differences associated with both mRNA regulation and specific diseases (75, 76). Most astrocyte cell culture studies have derived cells in a brain region-independent manner (e.g., whole brain or hemispheres), yet astrocytes derived from different brain regions have been reported to have different gene expression profiles (73). To determine if the observed gene expression changes were specific to cortical astrocyte cultures, we measured the glucocorticoid sensitivity of select mRNAs in hippocampal astrocyte cultures. All of the genes measured were regulated by glucocorticoids in both cortical cultures and hippocampal cultures, indicating that these glucocorticoid effects occur in astrocytes derived from multiple brain regions (Fig. 5). This observed regulation could result from common mechanisms for robustly regulated mRNAs (i.e., directly mediated by GR), although we cannot exclude the possibility that astrocytes in vitro undergo cumulative cellular changes that disrupt or remove brain-region dependent differences. Whether the glucocorticoid regulation of mRNAs in astrocytes characterized in our experiments occurs in all astrocytes regardless of brain region source remains an open question. Additional experiments will be needed for comprehensive transcriptional comparisons of glucocorticoid regulation between astrocyte cell cultures derived from different brain regions (e.g., basal gene expression concordance for all mRNAs by brain region source, examining lower fold-change expression changes).

Some mRNAs regulated by glucocorticoids in our experiments have been previously reported to have enriched expression in astrocytes compared with other brain cell types (i.e., neurons, oligodendrocytes) (8) (Fig. 6, A and B). Transcriptional regulation of these astrocyte-enriched mRNAs may have particular functional significance for astrocytes. Many mRNAs previously associated with glucocorticoid regulation as well as mRNAs displaying novel regulation are also astrocyte-enriched, representing 205 of the 854 unique mRNAs regulated by glucocorticoids (24.0%). In addition, multiple mRNAs (Atp6v1b2, Gap43, Syn2) reported to be both 1) enriched in neurons (8) and 2) regulated by glucocorticoids (21, 65) were not regulated in our astrocyte culture system (data not shown), further suggesting that glucocorticoids act in a cell-specific manner among brain cell types.

What is the functional output of regulating these astrocyte-enriched mRNAs? GO analyses of mRNAs from the entire time course data set that were enriched in astrocytes ≥fourfold and ≥10-fold both identified the terms “sulfur metabolism” (Papss2, Mamdc2, Cbs, Mgst1, Gstk1), “nitrogen compound biosynthetic process” (Atp1a2, Cbs, Prodh, Slc7a2, Rora, Glul), and “oxidation reduction” (Adhfe1, Aldh1a1, Cyp2d22, Cyp4v3, Dio2, Fmo1, Prodh, Aass, Aldh6a1) as significantly enriched processes (Fig. 6C, Table 3). The regulated mRNAs associated with nitrogen biosynthetic processes and oxidation reduction encode for proteins associated with a range of cellular processes, including nitric oxide metabolism (e.g., Slc7a2, Rora) and amino acid metabolism and transport (e.g., Cbs, Prodh, Slc7a2, Glul, Aass, Aldh6a1). These pathways are associated with the cellular stress response, a reaction to glucocorticoid stimulation that is known to increase reactive oxygen species and cellular metabolic load (16).

We also observed striking basal and temporal trends in the percentage of glucocorticoid-regulated mRNAs that were astrocyte enriched. First, the percentage of glucocorticoid-regulated mRNAs that are astrocyte enriched was significantly higher for all lengths of glucocorticoid exposure than the average percentage for all detected mRNAs (range across all time points: 13–14.5%). All astrocyte-enriched percentages for glucocorticoid-regulated mRNAs were >1.5× greater and up to 3× greater than the astrocyte-enriched percentages for all expressed mRNAs. The difference in percentages implies that astrocyte-enriched mRNAs are on average more likely to be regulated by glucocorticoids compared with all mRNAs expressed in our astrocyte cultures. Our dataset is one of the first to demonstrate empirically a bias in transcriptional regulation in astrocytes toward astrocyte-enriched mRNAs at the level of the transcriptome in vitro. Second, we noticed that astrocyte-enriched mRNAs represented a greater percentage of glucocorticoid-regulated genes at earlier time points compared with later time points in the time course. This percentage steadily declined over the entire time course (e.g., for 1.5-fold enriched mRNAs, 42.4% at 2 h → 22.5% at 24 h). This temporal trend suggests an additional bias of glucocorticoid regulation toward astrocyte-enriched mRNAs in initial astrocyte responses to steroid exposure.

Generalizability of these findings to other brain cell types and additional transcriptional regulators is an intriguing potential extension of these findings. Glucocorticoids are thought to act in a cell-type dependent fashion based upon a number of factors, including 1) basal gene expression patterns (e.g., genes expressed only in astrocytes based on chromatin structure and gene availability/regulation) and 2) the compliment of transcriptional regulators present in the cell (e.g., cell-dependent transcription factor expression and interactions with GR in astrocytes) (54a, 67, 71). We have added empirical evidence suggesting that cell-enriched mRNAs may direct or guide initial cellular responses to glucocorticoids. Based on our findings, understanding the response of a specific cell type to a given signaling cascade may be enhanced by 1) knowing which mRNAs are enriched in that cell type relative to other cells in a given tissue and 2) investigating the regulation of cell type-enriched mRNAs. We currently do not know how the observed glucocorticoid regulation in astrocytes compares with glucocorticoid regulation in other brain cell types (e.g., neurons, oligodendrocytes), although there are data available on glucocorticoid mRNA regulation in hippocampal slices and a neuronal cell line (48, 49). As additional cell type-enriched expression data become available, comparing and contrasting our findings with glucocorticoid regulation in other cell types, both in the brain and in other tissues, may increase our understanding of the mechanisms of cell type-specific glucocorticoid regulation.

There are caveats to consider in interpreting these results. First, we used stringent criteria to define differential expression. mRNAs were considered differentially expressed in the microarray experiments when they were changed more than twofold by both dexamethasone and corticosterone vs. vehicle treatment. While this conservative criterion increases the likelihood of identifying authentic glucocorticoid regulation by limiting the identification of false positives, it also excludes mRNAs that are regulated at a lower magnitude by dexamethasone, corticosterone, or RU486. In addition, some genes were statistically regulated by one glucocorticoid and not the other; such genes may be authentic glucocorticoid targets but would not be identified as such based on these criteria (e.g., qPCR validation of Adora2b and Wnt7a). Second, all of these gene expression experiments were conducted on astrocytes in vitro. This approach isolated the glucocorticoid response in astrocytes from other cell types, but it does not assess the effects in vivo. As with many primary culture systems, cultured astrocytes are reported to have different gene expression profiles compared with astrocytes in vivo (8, 22). Additional experiments will be needed to determine how the glucocorticoid regulation in astrocytes in vitro translates to the in vivo environment and corresponding physiological influence. Third, astrocytes in the brain are regulated in the context of intercellular interactions; how glucocorticoids affect astrocytes and these mRNAs in their interactions with neurons and other glial cell types in the intact brain remains to be investigated. In addition, although a subset of regulated mRNAs are reportedly astrocyte enriched, these mRNAs may be expressed in other cell types. We are interested in knowing if the glucocorticoid regulation of these mRNAs observed in astrocytes also occurs in a similar manner in other cell types. Fourth, while the time course of regulation is consistent with primary transcriptional regulation, we have not demonstrated that GR directly modulates the transcription of these genes but rather that mRNA levels change. Future studies can be designed to assess whether GR binds the promoter domains near these the genomic origins of these glucocorticoid-regulated mRNAs. Finally, the associations mentioned in terms of astrocyte enrichment and gene ontology analyses are derived from transcriptome-level data sets. Individual gene connections and pathway enrichment must be further investigated to confirm the neurological relevance of these statistical connections.

In summary, we have characterized the responsiveness of mRNAs in astrocytes to glucocorticoids in vitro. To our knowledge, our study is the most extensive reported investigation of the temporal patterning of glucocorticoid-mediated mRNA regulation at the transcriptome level in a specific glial cell type. We find that 1) glucocorticoids regulate a specific set of mRNAs in astrocytes in vitro, 2) the observed dynamic gene expression regulation contains associations with specific biological pathways, 3) glucocorticoids regulate specific mRNAs in astrocyte cultures derived from multiple brain regions, and 4) a subset of glucocorticoid-regulated mRNAs is enriched in astrocytes. Together, these data add to the knowledge of glucocorticoid-mediated gene expression regulation and further enhance our understanding of glucocorticoid signaling and stress biology in the brain.

GRANTS

Funds used to support this work include National Institutes of Health Grants T32-EY-017878 (B. S. Carter), T32-MH-014279 (B. S. Carter), and R01-DA-025973 (R. C. Thompson).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: B.S.C. and R.C.T. conception and design of research; B.S.C. performed experiments; B.S.C., F.M., and R.C.T. analyzed data; B.S.C. and R.C.T. interpreted results of experiments; B.S.C. prepared figures; B.S.C. drafted manuscript; B.S.C. and R.C.T. edited and revised manuscript; B.S.C., F.M., and R.C.T. approved final version of manuscript.