(A) Western blots to analyze the protein levels of Taf1, Taf3, Taf4, Tbp, Nanog and Oct4 in control and TAF3 K/D cells (D3, P2 post infection).
(B) qRT-PCR to measure the mRNA levels of Taf3, Nanog and Oct4 in control and TAF3 K/D cells (D3, P2 post infection).
(C) Alkaline phosphatase staining of control and TAF3 K/D cell colonies (R1, P4 post infection) on feeder cells.
(D) Percentages of SSEA1⁺Oct4⁺ cells in wild type, control and TAF3 K/D cell populations (D3, P6 post infection).
(E) Percentages of S phase cells in wild type, control and TAF3 K/D cell populations (D3, analyzed for 6 passages post infection).
(F-G) Confocal fluorescence images of control and TAF3 K/D cells (R1, Passage 4 post infection) co-stained with antibodies against TAF3 and Oct4 (F) or TAF3 and Nanog (G). Nuclei were counterstained with DAPI.
(H) Single cell expression correlation analysis of TAF3 and Nanog in shRNA A treated cells (R1, P4 post infection). Gray values along a random chosen path in DAPI, TAF3 (shRNA A), and Nanog channels of (G) were plotted in the diagram.
Scale Bars, (C) 150 μm, others 50μm.

(A) Difference in expression (log2(anti-TAF3 shRNA A/non-target shRNA)) and statistical significance between TAF3 depleted ES cells and controls for selected genes at three timepoints, ES Cell, EB day 3 and EB day 6.
(B) Tissue bias (red line; Experimental Procedures) calculated between definitive endoderm (E8.0; SM143) and whole brain (E12.5; SM108) SAGE libraries plotted on the difference in expression (defined as in (A); black points) between TAF3 K/D cells and controls. Tissue bias recalculated using randomized gene expression data was plotted in grey.
(C) Overlaps between genes that were differentially expressed in TAF3 K/D EBs at day 6 (EB6) and genes that were differentially expressed in Sox17 over-expressing human embryonic stem cells. P-values (italics) for each intersection were calculated assuming a hypergeometric distribution and a total ortholog count of 14,400. Values significant at the 5% level were plotted in red.

(A) Read accumulation for six ChIP-seq datasets (TAF3, TBP, Pol II, H3K4me3, CTCF and Oct4) at the Gata4 locus. Vertical axis was from 3 to 50 reads for all factors. TAF3 dependent expression changes were plotted below genes as for Figure 4. Peaks referenced in the text were boxed and numbered as in panels D-F. Active promoters were denoted with dashed boxes.
(B) As for (A) at the Lefty1 locus.
(C) Binding correlations (Spearman coefficient) between all pairs of factors considered in this study based on enrichment values calculated using the MACS model at 125K genomic loci. Negative correlations were set to zero for display.
(D) Clustering of TAF3 bound regions into four classes using the first three principal components of binding variation computed from all factors in (C).
(E) Summary statistics for the four TAF3 binding classes in (D). All values except the number of regions in each class (first column) are median values for the whole class. Enrichment was calculated as for (C).
(F) Z-score association between TAF3 binding classes from (D) and genes whose expression was significantly affected by TAF3 depletion. Significant values were underlined; Z-score between +/-1 (P-value > 0.6); Z-scores > 2 (P-value < 0.05); Z-scores > 3 (P-value < 0.001). See Experimental Procedures for details of Z-score calculation.

(A) Western blots to examine TAF3 levels in different cell types and tissues.
(B) TAF3 is selectively reduced upon ES cell differentiation (EB day 1, 2, 4, 6, 8 and 10).

(A-C) qRT-PCR to measure the expression of lineage-specific markers in control and TAF3 K/D samples.
(D) H&E staining of EB sections from control and TAF3 K/D EBs (day 10). (E-F) Confocal fluorescence images of control and TAF3 K/D EBs (day 10) stained with antibody against GATA4 (E) or Afp (F).
(G) Live-cell fluorescence images of control and TAF3 K/D EBs (day 8, 46C).
(H) Confocal fluorescence images of control and TAF3 K/D EBs stained with antibody against Tubb3. EBs (day 4) were cultured on Laminin coated slides for 6 additional days.
(I) Confocal fluorescence images of control and TAF3 K/D EBs (day 10) co-stained with antibodies against TAF3 and GATA4.
(J) Gray values along a path from outside to inside of the EB (non-target) in DAPI, TAF3, and GATA4 channels of (I) were plotted in the diagram.
Scale Bars, 50 μm.

(A) Flow cytometric analysis of definitive endoderm (CXCR4⁺c-kit⁺) and mesoderm (CXCR4⁺Flk-1⁺) cells after the directed in vitro differentiation. Experiments were conducted with or without Activin A.
(B) qRT-PCR to examine the relative expression of endoderm (Sox17, Foxa2) and mesoderm (Mesp1) markers in Activin A treated samples.

(A) Western Blots to examine the relative enrichment of cohesin (Smc1a and Smc3), mediator subunit (med12) and CTCF by TAF3 immunoprecipitation using ES cell nuclear extract. B + E, Benzonase and Ethidium bromide treatment that eliminates DNA-mediated interactions.
(B) and (C) Mapping the CTCF-interacting domain of TAF3. Flag-HA tagged TAF3, RFP or truncated TAF3 proteins were over-expressed in 293T cells with CTCF and were then purified with anti-Flag resin. TAF3 and CTCF enrichments were analyzed by western blots. (C) Coomassie Staining of the purified Flag-HA-TAF3(501-730aa)/CTCF complex.
(D) ChIP analysis to examine the relative TAF3 enrichment at 6 core promoters bound by TAF3 and 11 distal sites bound by TAF3/CTCF in control and CTCF K/D cells.
(E) 3C experiments to assess long-range interactions between the distal TAF3/CTCF/cohesin site and regions around the core promoter of Mapk3 in control (blue) and TAF3 K/D (red) cells. ▲, anchor point.
(F) Western blots to examine TAF3 and CTCF protein levels in control, TAF3 K/D (shRNA A), CTCF K/D (shRNA #1) and TAF3 & CTCF K/D cells (72 hours post infection).
(G) qRT-PCR analysis to measure Mapk3 expression in control, TAF3 K/D, CTCF K/D and TAF3 & CTCF K/D cells.
(H) TAF3 interacts with CTCF and mediates regulatory DNA looping to specify endoderm differentiation.