Fig 2: Impaired DNA Looping at TFIIIC-Looped Genes by TFIIIC Depletion or AEs Deletion Abrogates the Reactivation of Gene Expression upon Serum Re-exposure(A) Changes in specific intra-TADs contacts made by TFIIIC in T47D treated with siGTF3C5 or siCTRL in the absence of serum. Data are the log2 FC of observed versus expected (95% CI) of Hi-C data. The changes in siGTF3C5 versus siCTRL show significant (PPr = 0.99) decrease of total intra-TADs contacts compared to mock-transfected (–S) versus siCTRL.(B) Hi-C matrix heatmaps of the UHRF1 locus for siCTRL and siGTF3C5 cells in the absence of serum. Color scale bar is reported with red representing values from 200 and above, white values equal to 100, and blue values equal to 0. The looping between the AE bound by TFIIIC and its respective targets (PLIN4/5 and UHRF1) is indicated by a box when detected and an arrow when lost. Note that even in the absence of serum the Alu-FEM1A loops back to the PLIN4/5 locus upon siGTF3C5 (as reported for the condition of presence of serum in Figure 4B).(C) Venn diagram showing the overlap between the TFIIIC-looped and the serum-dependent genes compared to the regions showing compartment switch (Figure S4B). Note the higher overlap of the regions showing compartment changes with the serum-dependent genes (Fisher’s exact test).(D) RNA-seq expression analysis in T47D cells of TFIIIC-activated genes after serum re-exposure for the indicated times in conditions of siCTRL and siGTF3C5. Expression data are reported as RPM (**p < 1x10-12, two-tailed paired t test in comparisons for each time-point). The comparison between 0 to 3 h in siCTRL cells is shown to highlight the rapidity of gene activation after serum addition (***p < 1x10-20). GTF3C5-depleted levels were maintained during the time course (Figure S6B).(E) Schematic representation of the CRISPR-Cas9 approach to delete the TFIIIC-bound AE located between the PLIN4/5 and UHRF1 loci in chromosome 17. The wild-type (WT) and the deleted alleles are shown. The targeted AE is shown as a purple box, the position of the guide RNAs (gRNA1 and gRNA2) is marked with triangles, and the primers used for the screen (Figure S6C) are indicated with blue arrows. Arrows indicate the chromatin interactions in +S (red) and –S (black) conditions based on Figure 4B.(F) UHRF1 expression by qRT-PCR in T47D parental cells (WT) and AE-deleted (Clone11) in the absence of serum (mean ± SEM of biological duplicates with data in WT cells normalized to 1; one-tailed t test).

Fig 3: TFIIIC controls activity-dependent genes transcription and SINE acetylation.(A) Representative images of cortical neurons transfected with a GFP expression vector and either control or Gtf3c5 siRNA. Neurons were transfected at DIV2, after three days exposed to 50 mM KCl for 45 minutes or left untreated, and subjected to quantitative RNA-FISH analysis of c-Fos mRNA. Maximal z-projections of confocal stacks of transfected cells are shown. In GFP-expressing cells, c-Fos mRNA ribonucleoparticles (red) were detected by FISH (reconstructed cell edges are shown in green). Nuclei were stained with DAPI (blue). Ribonucleoparticles showing low levels of c-Fos mRNA are indicated by arrowheads. (B) Quantitative analysis of RNA-FISH experiments. For each neuron, total fluorescence intensity of all c-Fos mRNA particles was calculated. Average and s.e.m. of at least 25 cells per condition are shown (*, P<0.05; **, P<0.01; ***, P<0.001, two-way ANOVA). (C, D, E and F) Mouse primary cortical neurons were infected with lentiviral particles driving the expression of a short hairpin RNA targeting either firefly luciferase (shLUC, as a negative control) or Gtf3c5 (shGtf3c5). 4 days later cells were stimulated with either 50 mM KCl for 45 minutes or left untreated, then subjected to RNA extraction and cDNA synthesis, followed by qRT-PCR analysis of c-Fos and Gadd45b pre-mRNA (C, E) and mRNA (D, F). Silencing of Gtf3c5 was sufficient to drive a significant enhancement of activity-dependent transcription, as assessed both at the level of pre-mRNA (C, E) and fully processed RNA (D, F). (G) Primary cortical neurons were infected with lentiviral particles driving the expression of a short hairpin RNA targeting either firefly luciferase (shLUC, as a negative control) or Gtf3c5 (shGtf3c5). 4 days later cells were stimulated with either 50 mM KCl for 45 minutes or left untreated, then subjected to ChIP using either H3K9K14ac or histone H3 antibodies, followed by qPCR. Lentiviral-mediated silencing of Gtf3c5 enhanced histone H3 acetylation at c-Fos RSINE1, Gadd45b B1F and not at c-Fos TSS and Gapdh B4. Histograms show the ratio of immunoprecipitation efficiency between H3K9K14ac and H3 antibodies relative to total chromatin input (average and s.e.m. of 5 experiments are shown; *, P<0.05, Student's t-test).

Fig 4: Architectural proteins associate with SMARCAD1 in mouse pluripotent ESCs. (a) Experimental set-up. (b) Western blot analysis of endogenous immunoprecipitates (IPs) from 46C nuclear extracts; 6% of nuclear extract was used as input. The IPs were performed in the presence of ethidium bromide and benzonase to exclude associations mediated by nucleic acids. SMARCAD1 runs as a doublet in these cells. The lower band in the TFIIIC63 panel corresponds to the heavy chain of the antibody used in the IP.