Fig. 8. Sox2 regulates GFAP expression and directly interacts with upstream regulatory DNA sequences of the GFAP gene in vitro and in neural cells chromatin. (A) Sox2 overexpression in differentiating cells represses endogenous GFAP expression. Double immunofluorescence (confocal microscopy) of day 9-differentiated cells transduced with Sox2-expressing lentivirus (Sox2-GFP; left) or control lentivirus (GFP; right) at day 1 (d1) or 4 (d4), with antibodies against GFP (green, revealing Sox2-IRES-GFP, or GFP for control virus), and the astroglial marker GFAP (red). Sox2-lentivirus-transduced cells show no, or very little, GFAP expression, whereas strongly GFAP-positive cells in the same field are Sox2-GFP-negative (left). By contrast, in cells transduced with control virus, GFP and GFAP colocalize within most cells. (B) Double immunofluorescence for GFAP and astrocytic markers S-100 (left) or connexin 43 (CX43; right) (Nagy and Rash, 2000) in differentiation day 9 cells; not transduced (nt) or day 1 transduced with Sox2-GFP-expressing lentivirus (d1). Virtually all cells positive for GFAP co-express S-100 or CX43 in non-transduced cells. In Sox2-transduced cells, numerous cells can be seen which have low or absent GFAP expression; and are positive for S-100 (left) or for CX43 (right), confirming their astroglial identity. (C) Putative Sox2-binding sites within a 0.6 kb region (0.6GFAP) just upstream to a previously investigated 2.5 kb GFAP promoter/enhancer. The sequence highlights the Sox2 consensus sequences investigated (red). Gfap is the oligonucleotide used in EMSA experiments in E; MutGfap is its mutated version (nucleotide substitutions in green). CDS: coding sequence. (D) Co-transfection experiments in P19 cells. Activity of a luciferase reporter gene driven by the 0.6 GFAP region linked to a TK minimal promoter (0.6GfapTK), or by the TK promoter only (TK), when co-transfected with Sox2 expression vector, or control `empty' vector (as indicated). Asterisk indicates a statistically significant difference (paired t-test, P<0.005). Results are average of n=4 transfections in duplicate. (E) EMSA with probes (indicated below the panels) encompassing the Sox2 consensus binding sites in the 0.6 GFAP region (Gfap), or the same probe mutated as in 8B (MutGfap), or a control probe carrying a Sox2-binding site from an Oct4 gene enhancer (Oct4). Nuclear extracts (P19; SOX2/COS, COS cells transfected with Sox2 expression vector; COS, untransfected COS cells), and competitor oligonucleotides with the molar excesses used for the competition experiments in the right panel, are indicated above the figure. (F) ChIP with anti-SOX2 antibodies of the 0.6 Gfap region in P19 and E12.5 spinal cord cell chromatin, compared with control SRR2 (which is bound by Sox2 in P19, but not in E12.5 spinal cord cell chromatin) (Miyagi et al., 2006) or nestin (bound by Sox2 in P19 and E12.5 spinal cord cell chromatin) (Tanaka et al., 2004; Miyagi et al., 2006) regulatory regions. The anti-Sox2 antibody precipitates both GFAP and SRR2 chromatin in P19 cells, but only GFAP chromatin in spinal cord cells, as expected. Antibodies are indicated above the panels; cell types and amplified DNA regions are indicated below the panels. Arrowheads indicate the positions of PCR bands corresponding to amplified target regions. Low-intensity diffused bands at the bottom are non-reacted primers. Results are representative of three experiments. unrel, unrelated control antibody against SV40 large-T antigen; Input chrom, input chromatin (not immunoprecipitated) - a positive control for the PCR reaction.