RETINOIC ACID INDUCED DISRUPTION OF cAMP-INDUCED HUMAN SVG ASTROGLIAL CELL
DIFFERENTIATION: MORHOLOGY AND GLOBAL GENE EXPRESSION EFFECTS.
L D Burgoon1,3, K Y Kwan2,M R Fielden2,3, J E Trosko3,4, T R Zacharewski2,3.
1Dept. of Pharmacology & Toxicology; 2Dept. of Biochemistry & Molecular
Biology; 3Institute for Environmental Toxicology; National Food Safety &
Toxicology Center; 4Dept. of Pediatrics & Human Development, Michigan State
University, East Lansing, MI, USA.
Previous work has shown that treatment of human SVG cells with 5 µM forskolin
(F) and 200 µM 3-isobutyl-1-methylxanthine (iBMX) increases cAMP levels
resulting in differentiation and dramatic morphologic changes. We show that
co-treatment with 0.5 µM retinoic acid (RA) enhances these morphologic
changes as measured by increases in the total number of processes (p = 0.03)
and the length of processes contacting an adjacent cell (LPT; p < 0.01) over
time. At 36h, RA inhibited regression of LPT (p = 0.02) with increases in mean
process length (p = 0.04) and the longest primary process (p <0.01). A custom,
SVG-specific cDNA microarray (2,990 genes) was used to investigate the temporal
changes in gene expression during cAMP induced differentiation. Temporal changes
were identified using two different techniques. First, an incomplete block design,
individual gene mixed-model ANOVA was used for analysis, followed by t-tests
using a step-down Bonferroni adjustment for multiple comparisons. The mixed
model included treatment, time, dye and treatment x time as fixed effects, and
spots as random effects. To identify genes with significant expression changes
during differentiation, least squares means of F/iBMX treated cells were compared
to time-matched vehicle control samples utilizing t-tests. For comparison, a
second technique utilizing Shannon Entropy filtering of normalized microarray
data from treated, vehicle control, and untreated/time 0 h control cells, with
Principal Components Analysis (PCA) of the filtered data to result in the identification
of expression patterns was used. Results from the two divergent approaches were
comparable and identified significant temporal and treatment effects on gene
expression, although some differences were observed.