RNA Polymerase II transcription independent of TBP in murine embryonic stem cells

  1. James ZJ Kwan
  2. Thomas F Nguyen
  3. Anuli C Uzozie
  4. Marek A Budzynski
  5. Jieying Cui
  6. Joseph MC Lee
  7. Filip Van Petegem
  8. Philipp F Lange
  9. Sheila S Teves  Is a corresponding author
  1. Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Canada
  2. Department of Pathology and Laboratory Medicine, University of British Columbia, Canada
  3. Michael Cuccione Childhood Cancer Research Program, BC Children’s Hospital Research Institute, Canada

Abstract

Transcription by RNA Polymerase II (Pol II) is initiated by the hierarchical assembly of the pre-initiation complex onto promoter DNA. Decades of research have shown that the TATA-box binding protein (TBP) is essential for Pol II loading and initiation. Here, we report instead that acute depletion of TBP in mouse embryonic stem cells has no global effect on ongoing Pol II transcription. In contrast, acute TBP depletion severely impairs RNA Polymerase III initiation. Furthermore, Pol II transcriptional induction occurs normally upon TBP depletion. This TBP-independent transcription mechanism is not due to a functional redundancy with the TBP paralog TRF2, though TRF2 also binds to promoters of transcribed genes. Rather, we show that the TFIID complex can form and, despite having reduced TAF4 and TFIIA binding when TBP is depleted, the Pol II machinery is sufficiently robust in sustaining TBP-independent transcription.

Editor's evaluation

This important study employs auxin-induced degradation to provide compelling evidence that the TATA-binding protein (TBP) is not required for ongoing RNA polymerase II transcription nor heat-shock or retinoic acid-induced transcription in murine embryonic stem cells, but is essential for RNA polymerase III transcription. Furthermore, the study shows that TBP-independent TFIID complexes are assembled and present at the transcription start sites of polymerase II-transcribed promoters. This work will be of broad general interest to the regulation of gene expression field.

https://doi.org/10.7554/eLife.83810.sa0

Introduction

Transcription is the first step in gene expression, and it begins with the stepwise assembly of general transcription factors (GTFs) and RNA Polymerase II (Pol II) to form the pre-initiation complex (PIC) at the promoters of genes (Matsui et al., 1980; Reinberg et al., 1987; Buratowski et al., 1989). First, the GTF TFIID, composed of the TATA-box binding protein (TBP) and 13–14 evolutionarily conserved TBP-associated factors (TAFs), binds onto gene promoters (Reinberg et al., 1987; Buratowski et al., 1989; Dynlacht et al., 1991; Cormack and Struhl, 1992; Tanese et al., 1991), which triggers the recruitment of TFIIA and TFIIB (Reinberg et al., 1987; Buratowski et al., 1989). Pol II is then loaded onto the promoter along with TFIIF (Reinberg et al., 1987; Buratowski et al., 1989; Buratowski et al., 1991). The binding of TFIIE and TFIIH completes the PIC, upon which the enzymatic activity of TFIIH unwinds DNA to form a transcription bubble, allowing for Pol II elongation and the incorporation of nucleotides into the nascent RNA (Reinberg et al., 1987; Buratowski et al., 1989; Rimel and Taatjes, 2018).

Decades of research have revealed a central role for TBP in the transcription initiation process. TBP was first discovered to recognize and bind to the TATA-box (TATAAA consensus sequence) in gene promoters in vitro (Nakajima et al., 1988; Hoey et al., 1990; Buratowski et al., 1988), though subsequent studies show that TBP binds to active gene promoters in vivo regardless of the presence of a TATA sequence (Kim and Iyer, 2004; Pugh and Venters, 2016). Early studies in Saccharomyces cerevisiae expressing TBP mutations show that the loss of functional TBP led to decreased GTF and Pol II recruitment and transcription in vivo (Klages and Strubin, 1995; Kim et al., 1994). More recently, acute TBP depletion in yeast cells via the anchor-away system led to a rapid, genome-wide loss of Pol II and GTF occupancy (Petrenko et al., 2017; Petrenko et al., 2019). In addition to facilitating Pol II transcription, TBP is also involved in at least two other RNA polymerases found in eukaryotes, and conditional inactivation of TBP in yeast led to a global decrease in both RNA Pol I and III transcriptional activity (Cormack and Struhl, 1992; Kwon and Green, 1994; Schröder et al., 2003; Shen et al., 1998; White et al., 1992). In vitro experiments using purified proteins from yeast and human cells also confirm the role of TBP in the three main RNA polymerases (Goodrich and Tjian, 1994; Burley and Roeder, 1996). However, mouse embryos harboring a homozygous TBP knock-out, which were viable up to ~40 cell stage of embryogenesis, still displayed normal levels of Pol II transcription (Martianov et al., 2002). Furthermore, acute depletion of TBP in mouse embryonic stem cells (mESCs) showed that global transcription levels remained unaffected (Teves et al., 2018), raising the question of how Pol II transcription occurs in the absence of TBP in these cases.

Several potential mechanisms may facilitate TBP-free Pol II transcription. One, TBP may be required only for the initial round of activation, but ongoing transcription may no longer needed for TBP. The impairment of the reactivation of genes following mitosis when TBP is depleted (Teves et al., 2018) would support this mechanism. Two, metazoan-specific TBP paralogs could functionally replace TBP. In some species, TBP paralogs have been found to bind onto genomic loci distinct from TBP and regulate cell type-specific genes (Akhtar and Veenstra, 2009; Kedmi et al., 2014). Furthermore, TBP-related factors TRF1 and TRF2, first discovered in Drosophila, are also able to substitute for TBP in TATA transcriptional systems (Crowley et al., 1993; Hansen et al., 1997; Rabenstein et al., 1999; Holmes and Tjian, 2000). A third vertebrate-specific paralog, TRF3, mediates transcription of oocyte-specific genes through a non-canonical complex with TFIIA/TFIIB (Yu et al., 2020). Three, a TAF-containing complex could facilitate Pol II transcription without TBP. Indeed, several TAF complexes have been identified, including the TBP-free TAF-containing complex (TFTC) isolated from human cells that can support in vitro Pol II transcription without TBP or TFIID (Wieczorek et al., 1998). TFTC was subsequently shown to correspond to the SAGA (Spt-Ada-Gcn5 acetyltransferase) complex, a conserved eukaryotic co-activator with histone acetyltransferase activity (Cheon et al., 2020), which has been shown to act somewhat redundantly with TFIID in yeast (Donczew et al., 2020).

In this study, we examined the role of TBP in transcription initiation in mESCs using an acute depletion system. We report that, inconsistent with a requirement for TBP in Pol II-dependent transcription, rapid depletion of TBP has no global effect on nascent RNA levels and Pol II chromatin occupancy. In contrast, tRNA transcription and Pol III chromatin occupancy are severely impaired upon depletion of TBP. We found that gene induction via the heat shock (HS) response and retinoic acid (RA)-mediated differentiation occurs normally upon TBP depletion. We also report that although the TBP paralog TRF2 binds to active gene promoters, it does not functionally replace TBP in Pol II transcription. Lastly, we show that the TFIID complex can form when TBP is depleted, although subunits have altered DNA occupancy. We find that TFIIA also has reduced binding when TBP is depleted. Taken together, these results point to a TBP-independent mechanism for Pol II transcription.

Results

TBP is not required for global Pol II transcription

We have previously generated homozygous knock-in of the minimal auxin-inducible degron tag at the N-terminus of TBP (mAID-TBP) in mESCs (C64 cells) (Teves et al., 2018), which enables rapid and acute depletion of the tagged protein upon the addition of indole-3-acetic acid (IAA) (Figure 1A; Holland et al., 2012). Treatment of C64 cells with 6 hr of IAA (+IAA) resulted in no major effects on cell viability (Figure 1—figure supplement 1A) with over ~90% depletion of mAID-TBP as detected by immunofluorescence (Figure 1—figure supplement 1B), by western blot of whole cell lysates (Figure 1B) and of subcellular fractions, including chromatin-bound fractions (Figure 1—figure supplement 1C). We also detected a large decrease in bound TBP at Gapdh and Hspa1a as measured by ChIP-qPCR (Figure 1—figure supplement 1D). To further assess the degree of acute depletion of chromatin-bound TBP genome-wide, we performed spike-in controlled Cleavage Under Targets and Tagmentation (CUT&Tag), a high-resolution chromatin profiling technique (Kaya-Okur et al., 2019), in two independent replicates after 6 and 16 hr of TBP depletion. After spike-in normalization, CUT&Tag analysis of TBP under control conditions shows high enrichment of TBP at the transcription start sites of genes (TSSs) (Figure 1C–D), with high reproducibility across replicates compared to the nonspecific IgG negative control (Figure 1—figure supplement 1E–G). After 6 and 16 hr of IAA treatment, we observed a large depletion of bound TBP (Figure 1C–D, Figure 1—figure supplement 1E), and when quantified at the promoters of genes (herein defined as –250 bp to TSS) relative to control conditions, we observed over ~80% decrease (Figure 1E). Pearson correlation confirms the difference between conditions, as control samples correlated more strongly with one another compared to IAA-treated samples (Figure 1—figure supplement 1G). Therefore, using orthogonal approaches, we confirm efficient depletion of TBP.

Figure 1 with 3 supplements see all
Global Polymerase II (Pol II)-mediated transcription is TATA-box binding protein (TBP)-independent in mouse embryonic stem cells (mESCs).

(A) Schematic of indole-3-acetic acid (IAA)-mediated degradation of mAID-TBP. (B) Western blot analyses of whole cell lysates for control and IAA-treated C64 mESCs blotting for ɑ-TBP, with ɑ-Tubulin used as a loading control. Cells were incubated with DMSO (control) or 500 μM IAA for 6 hr unless otherwise stated. (C) Gene browser tracks of Actb (top) and Gapdh (bottom) for cleavage under targets and tagmentation (CUT&Tag) analyses of TBP (orange) and Pol II (green), and strand-specific reads from NET-seq data (blue) in control or IAA-treated C64 mESCs. Alternative promoters are indicated. (D) Genome-wide average plots (top) and heatmaps (bottom) arranged by decreasing Pol II occupancy of TBP CUT&Tag (left), Pol II CUT&Tag (middle), and NET-seq (right) in a 2 kb window surrounding the transcription start site (TSS) of all genes for control and IAA-treated C64 mESCs. (E) Normalized read counts of TBP (left) and Pol II (right) CUT&Tag signal for control vs. IAA-treated C64 mESCs in the promoter (–250 to TSS) region of each gene are displayed as a scatter plot. (F) Differential gene expression (DGE) analysis of Pol II CUT&Tag (left) and NET-seq analysis (right) in control vs. IAA-treated C64 mESCs.

Figure 1—source data 1

Full uncropped membrane scans of immunoblot analyses with α-TBP control and indole-3-acetic acid (IAA)-treated C64 whole cell lysates without and with annotations.

Relevant bands for Figure 1B are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 1B.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig1-data1-v2.zip
Figure 1—source data 2

Full uncropped membrane scans of immunoblot analyses with α-Tubulin for control and indole-3-acetic acid (IAA)-treated C64 whole cell lysates without and with annotations.

Relevant bands for Figure 1B are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 1B.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig1-data2-v2.zip

To investigate the immediate effects of TBP depletion in mESCs, we measured Pol II occupancy using CUT&Tag under control and 6 hr of TBP depletion, and normalized the signal using the ChIPseqSpikeInFree method (Jin et al., 2020; Figure 1C and D) and two independent spike-in controls (Figure 1—figure supplement 2A). The normalization methods gave concordant results, hence subsequent analyses for Pol II CUT&Tag was performed using the ChIPseqSpikeInFree method unless stated otherwise. We observed that in control samples, Pol II binds at high levels on the promoters and gene bodies of Gapdh and Actb (Figure 1C, Figure 1—figure supplement 2B). After 6 hr of IAA treatment, Pol II occupancy at the Gapdh and Actb loci remained largely unchanged. We then plotted the genome-wide occupancy of Pol II surrounding the TSS for all genes as heatmaps (Figure 1D, Figure 1—figure supplement 2A), and also displayed the normalized read counts for Pol II at the promoters of genes with and without IAA treatment as a scatter plot (Figure 1E). Overall, we observed no global changes in genome-wide Pol II occupancy after TBP depletion. Prolonged IAA treatment (16 hr) also showed no global effects on Pol II occupancy (Figure 1—figure supplement 2C). Indeed, Pearson correlation analysis using 10 kb bins of all Pol II CUT&Tag replicates show high correlation coefficients (0.85–0.97) across both control and IAA-treated samples regardless of treatment times, showing no global differences in Pol II occupancy (Figure 1—figure supplement 2D).

To directly measure Pol II activity, we performed NET-seq (native elongating transcript sequencing) analysis in control and TBP-depleted C64 mESCs. NET-seq captures newly transcribed RNA from elongating Pol II at single nucleotide resolution through extensive fractionation (Mayer et al., 2015). After normalization with spike-in controls, we detected comparable intronic and exonic signals in Gapdh and Actb genes for both control and IAA-treated samples (Figure 1C, Figure 1—figure supplement 2B). We then displayed normalized NET-seq reads as heatmaps in a 2 kb region surrounding the TSS for all genes (Figure 1D, Figure 1—figure supplement 2E). Consistent with the Pol II CUT&Tag data, we observe no global differences in nascent RNA levels between control and IAA-treated cells. Using k-means clustering, we identified genes with the largest decrease in TBP occupancy after IAA treatment (Figure 1—figure supplement 3A). Despite these genes experiencing the largest depletion of TBP upon IAA treatment, Pol II occupancy and nascent RNA levels remained unchanged (Figure 1—figure supplement 3A). We also observed no difference in Pol II occupancy and NET-seq signal for promoters containing or lacking a defined TATA-box sequence upon TBP depletion (Figure 1—figure supplement 3B). Taken together, our Pol II CUT&Tag and NET-seq data show that acute TBP depletion does not lead to genome-wide disruption of Pol II transcription in mESCs.

We next performed differential gene expression (DGE) analysis using the edgeR Bioconductor package (Robinson et al., 2010) to identify genes with statistically significant differences in Pol II occupancy (CUT&Tag) or activity (NET-seq) between control and IAA-treated samples (Figure 1F). Consistent with the global average plots and heatmaps, DGE analysis of both Pol II and NET-seq data in control vs. IAA-treated samples show that ~99.9% of genes do not display significant changes upon TBP depletion (black dots in Figure 1F). However, 29 genes showed significant downregulation with approximately one- to fourfold decrease in Pol II occupancy upon IAA treatment (Supplementary file 1). Gene browser tracks for Pol II occupancy and NET-seq signal on Rab26 and Eif4a2 genes confirm a decrease upon IAA treatment (Figure 1—figure supplement 3C). However, prolonged auxin treatment (16 hr) was also analyzed using DGE analysis and only two genes showed a significant downregulation in Pol II occupancy (Figure 1—figure supplement 2C), suggesting that some of the identified downregulated genes at 6 hr may be false positives. Therefore, in contrast to the global effects of TBP depletion in yeast cells, these findings show that only a select number of genes may be affected upon TBP depletion in mouse ESCs.

TBP is required for RNA Pol III transcription of tRNA genes

Previous studies have shown that TBP is required for initiation of the three main eukaryotic RNA polymerases (Cormack and Struhl, 1992). Whereas Pol II transcribes protein coding genes, Pol III transcribes all of the tRNAs, the 5S ribosomal RNA, 7SK, and 7SL RNAs, the U6 spliceosomal RNA, and other groups of non-coding RNAs that are critical for cellular functions (Roeder and Rutter, 1970; Weinmann and Roeder, 1974; Price and Penman, 1972; Zieve et al., 1977; Reddy et al., 1987; Wolin and Steitz, 1983; Martignetti and Brosius, 1995). Given our findings for Pol II, we next asked if TBP is required for Pol III transcription in mESCs. Treatment of C64 cells with 6 hr of IAA reduced global TBP protein levels, but protein levels of the RPC7 subunit of Pol III do not change, as confirmed by western blot of C64 whole cell lysates (Figure 2—figure supplement 1A), suggesting that Pol III protein levels are unaffected by TBP depletion. To investigate whether the chromatin occupancy of Pol III is affected, we performed CUT&Tag analysis of Pol III with and without IAA (Figure 2A–B, Figure 2—figure supplement 1B), and validated TBP depletion on tRNA genes in these samples (Figure 2—figure supplement 1D). Pol III binds at high levels on tRNA genes LeuCAA and GlnCTG in control samples, and upon TBP depletion, we observed a major decrease in Pol III occupancy at these tRNA genes and across each individual replicate (Figure 2A, Figure 2—figure supplement 1B). We also detected Pol III CUT&Tag signal on 5S ribosomal RNA, 7SK, and 7SL RNAs (Figure 2—figure supplement 1C), but the effects upon TBP depletion is less consistent, likely due to the highly repetitive nature of these genes. We then plotted the Pol III occupancy in a 2 kb window surrounding the TSS for all tRNAs in control and TBP-depleted cells and observed a global decrease in Pol III occupancy across all tRNAs upon TBP depletion (Figure 2B, Figure 2—figure supplement 1D). We also plotted the normalized read count values for Pol III on the promoters of tRNAs with and without IAA treatment as a scatter plot (Figure 2C) and observed that most points fall below the diagonal, confirming a major effect of TBP depletion on Pol III binding at tRNA genes.

Figure 2 with 1 supplement see all
TATA-box binding protein (TBP) is required for Pol III transcription of tRNAs in mouse embryonic stem cells (mESCs).

(A) Gene browser tracks of tRNA65-LeuCAA (top) and tRNA64-GlnCTG/tRNA1010-LeuCAA (bottom) for cleavage under targets and tagmentation (CUT&Tag) analyses of TBP (orange), Pol III (brown), and NET-seq data (blue) in control or indole-3-acetic acid (IAA)-treated C64 mouse embryonic stem cells (mESCs). (B) Average plots (top) and heatmaps arranged by decreasing occupancy (bottom) for TBP CUT&Tag (left), Pol III CUT&Tag (middle), and NET-seq (right) in a 2 kb window (CUT&Tag) or 50 bp window (NET-seq) surrounding the transcription start site (TSS) of all tRNAs. (C) Normalized read counts of Pol III CUT&Tag (top) and NET-seq signal (bottom) for control vs. IAA-treated mESCs in either the promoter (–250 bp to TSS) region of each tRNA gene for CUT&Tag or 50 bp window surrounding the TSS of all tRNAs for NET-seq.

NET-seq can also capture the activity of elongating Pol III (Mayer et al., 2015). We therefore analyzed NET-seq signals on tRNA genes in control and TBP-depleted mESCs. We detected high signals for tRNA genes LeuCAA and GlnCTG in control cells, and observed a two- to threefold decrease upon IAA treatment (Figure 2A, Figure 2—figure supplement 1B). We also detected subtle changes in the NET-seq signal on 5S ribosomal RNA, 7SK, and 7SL RNA genes (Figure 2—figure supplement 1C), with the caveat that the highly repetitive nature of the reads may obscure the real effects. We displayed normalized NET-seq reads as heatmaps in a 50 bp region surrounding the TSS for all tRNAs, and as a scatter plot with normalized read counts, and observed a similar decrease of signal for all tRNAs upon TBP depletion (Figure 2B–C, Figure 2—figure supplement 1D). Therefore, TBP is required for the transcription of tRNAs by Pol III in mESCs, consistent with previous studies in yeast and in vitro systems (Cormack and Struhl, 1992; White et al., 1992; Han et al., 2018; Wang and Stumph, 1995). Importantly, the consistent role of TBP in Pol III transcription contrasts strikingly with the TBP-independent Pol II transcription in mESCs.

TBP is not required for gene activation

Previously, we have shown that depletion of TBP in mESCs specifically during mitosis led to impaired reactivation of Pol II genes as cells enter G1 phase (Teves et al., 2018). Based on these observations, we hypothesized that a potential mechanism for TBP-independent Pol II transcription is that TBP may be required to activate silenced genes, but that subsequent transcriptional events would no longer require TBP. To test this hypothesis, we turned to the HS response. A highly conserved protective mechanism to various stressors, the HS response leads to rapid transcriptional induction of HS protein genes (DiDomenico et al., 1982). We exposed C64 cells to HS at 42°C for 30 min (Figure 3A) before collecting cells for Pol II CUT&Tag and NET-seq analyses. Gene browser tracks and heatmaps for CUT&Tag and NET-seq at HS genes show massive increases in occupancy and activity upon HS compared to control samples (Figure 3B, Figure 3—figure supplement 1A–C). DGE analysis of Pol II CUT&Tag data shows that 392 and 374 genes are significantly upregulated and downregulated, respectively, in HS samples compared to control (Figure 3C, Supplementary file 2). Gene ontology (GO) analysis of the upregulated genes showed markers that are typically enriched for HS response, further confirming proper HS induction (Supplementary file 3).

Figure 3 with 1 supplement see all
TATA-box binding protein (TBP) is dispensable for gene activation by heat shock and retinoic acid differentiation in mouse embryonic stem cells (mESCs).

(A) Schematic of heat shock treatments for C64 mESCs. After 6 hr of treatment with auxin (indole-3-acetic acid [IAA]) or DMSO, cells were either collected or subjected to heat shock at 42°C for 30 min before collection. (B) Gene browser tracks of Hspa1a (left) and Hsp90aa1 (right) for cleavage under targets and tagmentation (CUT&Tag) analyses of TBP (orange), Polymerase II (Pol II) (green), and NET-seq data (blue) in control or IAA-treated C64 mESCs. (C–D) Differential gene expression (DGE) analysis of Pol II CUT&Tag in heat shocked vs. control C64 mESCs (C) and in heat shocked + IAA-treated vs. heat shocked C64 mESCs (D). (E) Schematic of retinoic acid (RA) treatment for C64 mESCs. Cells were treated for 16 hr with DMSO or IAA and RA before collection. (F) Gene browser tracks of Stra8 (left) and Myc (right) for CUT&Tag analyses of TBP (orange), Pol II (green) in control or IAA-treated C64 mESCs. (G–H) DGE analysis of Pol II CUT&Tag in RA-treated vs. control C64 mESCs (G) and in RA + IAA-treated vs. RA-treated C64 mESCs (H).

To investigate the effects of TBP depletion on the induction of HS genes, we treated C64 cells with IAA for 6 hr followed by HS for 30 min and performed Pol II CUT&Tag and NET-seq analyses (Figure 3A). We confirmed TBP depletion following IAA treatment on HS genes using CUT&Tag (Figure 3—figure supplement 1A–B). Under HS + IAA conditions, gene browser tracks for Pol II CUT&Tag and NET-seq at HS genes show that these genes are still induced upon TBP depletion (Figure 3B, Figure 3—figure supplement 1A). DGE analysis of Pol II CUT&Tag for HS versus HS + IAA samples shows no significant changes in gene expression (Figure 3D), indicating that global expression changes induced by HS still occur upon TBP depletion. We also plotted Pol II occupancy for the top 275 upregulated HS genes identified from DGE analysis as heatmaps centered at the TSS (Figure 3—figure supplement 1C) and observed no change between HS and HS + IAA conditions. These results suggest that depletion of TBP does not impair Pol II induction of HS genes.

The HS response is a classical gene induction system regulated at the level of paused Pol II elongation (Core and Adelman, 2019), and HS genes may retain a scaffold of PIC components following promoter pausing release. Therefore, we also examined the role of TBP in an orthogonal gene activation system through RA-mediated differentiation of C64 mESCs, which leads to the silencing of pluripotency markers and activation of ectoderm-specific genes (Semrau et al., 2017). We treated C64 cells with RA and IAA for 16 hr (Figure 3E), and performed Pol II CUT&Tag. Gene browser tracks of Stra8 and Cdx1 genes show significant increases in Pol II occupancy after RA treatment (Figure 3F, Figure 3—figure supplement 1D). Additionally, the pluripotency marker Myc displays a large decrease in Pol II occupancy under RA conditions (Figure 3F). DGE analysis of Pol II CUT&Tag revealed 44 upregulated and 88 downregulated genes, confirming proper RA induction (Figure 3G, Supplementary file 4).

We then performed Pol II CUT&Tag in TBP-depleted and RA-treated cells (Figure 3E), and verified TBP depletion using CUT&Tag analysis (Figure 3F, Figure 3—figure supplement 1D–F). Gene browser tracks for Pol II CUT&Tag at the Stra8 gene, a RA-mediated gene, show similar levels of Pol II binding in both RA and RA + IAA conditions (Figure 3F). DGE analysis of Pol II CUT&Tag data for RA and RA + IAA samples (Figure 3H) also showed no significant differences between the two conditions. Finally, we plotted the Pol II occupancy for the 44 significantly upregulated genes extracted from DGE analysis for RA and RA + IAA treatment as heatmaps centered at the TSS (Figure 3—figure supplement 1F). Both RA and RA + IAA conditions display similar levels of Pol II, confirming that depleting TBP does not affect the activation of RA-specific genes. Taken together, these results indicate that TBP is dispensable for both ongoing transcription and gene induction.

The TBP paralog TRF2 is not required for Pol II transcription

As a second potential mechanism for TBP-independent Pol II transcription, we hypothesized that TBP paralogs may functionally replace TBP. The mouse genome contains two known TBP paralogs, the Tbpl1 and Tbpl2 genes, which encode for the proteins TBP-related factor 2 and 3 (TRF2 and TRF3), respectively (Akhtar and Veenstra, 2011). These proteins were shown to facilitate transcription in place of or in conjunction with TBP in specific organisms (Akhtar and Veenstra, 2009; Akhtar and Veenstra, 2011), and binds to active promoters along with TBP and TAF7l in mouse haploid testes cells (Martianov et al., 2016). We observed that unlike TRF2, TRF3 is not expressed in mESCs as measured by Pol II CUT&Tag signal and chromatin-associated RNA-seq from a previous study (Teves et al., 2018; Figure 4—figure supplement 1A), and that TRF2 levels are unaffected by TBP depletion (Figure 4A).

Figure 4 with 1 supplement see all
TATA-box binding protein (TBP) paralog TBP-related factor 2 (TRF2) is expressed in mouse embryonic stem cells (mESCs), but does not functionally replace TBP.

(A) Western blot analyses of whole cell lysates for α-TRF2 and α-H3K27me3 in control and indole-3-acetic acid (IAA)-treated C64 mESCs. (B) Western blot analyses of whole cell lysates for α-TRF2 and α-HA in the TRF2 knock-out cells overexpressing HA-TRF2 (B8HA), TRF2 knock-out cell line (B8), and C64. α-HA shows the TRF2 band in only the B8HA cell line, indicating proper and specific expression of HA-TRF2. (C) Gene browser tracks of Gapdh for cleavage under targets and tagmentation (CUT&Tag) analyses of α-HA (magenta), TBP (orange), and Polymerase II (Pol II) (green), in control or IAA-treated B8HA cell line. (D) Genome-wide average plot (top) and heatmap (bottom) arranged by decreasing HA-TRF2 occupancy of α-HA CUT&Tag in a 2 kb window surrounding the transcription start site (TSS) of all genes for control and IAA-treated B8HA mESCs. (E) Genome-wide average plot (top) and heatmap (bottom) arranged by decreasing Pol II occupancy of α-Pol II CUT&Tag in a 2 kb window surrounding the TSS of all genes for control and IAA-treated B8 mESCs. (F) Differential gene expression (DGE) analysis of Pol II CUT&Tag in control vs. IAA-treated B8 mESCs.

Figure 4—source data 1

Full uncropped membrane scans of immunoblot analyses with α-TRF2 and α-H3K27me3 control and indole-3-acetic acid (IAA)-treated C64 whole cell lysates without and with annotations.

Relevant bands for Figure 4A are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 4A.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig4-data1-v2.zip
Figure 4—source data 2

Full uncropped membrane scans of immunoblot analyses with α-HA control B8HA, B8, and C64 whole cell lysates without and with annotations.

Relevant bands for Figure 4B are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 4B.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig4-data2-v2.zip
Figure 4—source data 3

Full uncropped membrane scans of immunoblot analyses with α-TRF2 control B8HA, B8, and C64 whole cell lysates without and with annotations.

Relevant bands for Figure 4B are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 4B.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig4-data3-v2.zip

To profile TRF2 chromatin binding in mESCs, we overexpressed HA-tagged TRF2 (B8HA mESCs) in a TRF2 knock-out C64 mESCs (B8 mESCs) and confirmed expression via western blot of whole cell lysates (Figure 4B). We then performed spike-in normalized CUT&Tag for HA-TRF2 using the HA antibody with and without TBP depletion, and confirmed TBP depletion in B8HA cells through TBP CUT&Tag analysis (Figure 4C, Figure 4—figure supplement 1B). We found that HA-TRF2 binds to the promoter of Gapdh, and that the levels remain relatively unchanged upon TBP depletion (Figure 4C, Figure 4—figure supplement 1B). The HA signal is also specific, showing no binding in the B8 mESCs (Figure 4C). We then displayed the HA-TRF2 CUT&Tag data as a heatmap and average plot for all genes (Figure 4D). Similar to TBP, TRF2 binds to promoters of all active genes, and occupancy levels remain similar upon IAA treatment. Furthermore, B8HA HA-TRF2 occupancy is positively correlated with TBP and Pol II levels from the C64 cell line (Figure 4—figure supplement 1C–D). Therefore, in contrast to other species, mouse TRF2 has the capacity to bind to the same promoters of active genes as TBP in mESCs.

To test whether TRF2 can functionally replace TBP, we generated a TRF2 knock-out cell line with the entire coding sequence removed (B8 cells). We validated the knock-out with western blot of whole cell lysates and qRT-PCR (Figure 4B, Figure 4—figure supplement 1E–F). We then performed qRT-PCR to quantify the intronic and exonic transcripts of Gapdh in B8 cells with and without IAA treatment (Figure 4—figure supplement 1G). We observed no differences in both intronic and exonic RNA levels for Gapdh despite the combination of TBP knockdown and TRF2 knock-out, suggesting that nascent transcription of Gapdh is unaffected. To determine the effects on Pol II upon knocking out TRF2, we performed Pol II CUT&Tag on the B8 cells. We then performed DGE analysis of the Pol II CUT&Tag in the TRF2 knock-out B8 cells versus the control C64 mESCs (Figure 4—figure supplement 1H). DGE shows only 13 upregulated genes and 8 downregulated genes, suggesting that knock-out of TRF2 has no global consequences. To further investigate if TRF2 and TBP are functionally redundant in mESCs, we performed CUT&Tag for TBP and Pol II on TRF2 knock-out B8 cells with and without TBP depletion (Figure 4E, Figure 4—figure supplement 1I–J). We confirmed proper depletion of TBP in the B8 cell line (Figure 4—figure supplement 1I), and Pearson correlation analysis using 10 kb bins of Pol II CUT&Tag shows good concordance between replicates (Figure 4—figure supplement 1K). DGE analysis shows that 100 and 97 genes were downregulated and upregulated, respectively, upon TBP depletion in the B8 cells, confirming that Pol II occupancy is not affected globally upon TBP depletion in the TRF2 knock-out cells (Figure 4F). Additionally, we performed a cell viability assay under prolonged IAA treatment when either TBP, TRF2, or both are absent (Figure 4—figure supplement 1L). Both B8 and C64 cells displayed a similar death curve when treated with IAA, suggesting that the combinatorial knock-out of TRF2 with TBP depletion does not accelerate cell death. We conclude that TRF2 is not functionally redundant to TBP and that Pol II transcription may be TBP family-independent in mESCs.

The TFIID complex forms upon TBP depletion

A third potential mechanism for TBP-independent Pol II transcription is that TAF-containing TFIID complexes could function when TBP is depleted. To examine the composition of TFIID upon TBP depletion, we performed co-immunoprecipitation followed by mass spectrometry (IP-MS) with α-TAF4 in control and IAA-treated cells (Figure 5—figure supplement 1A–B). TAF4 is a member of the core TFIID subcomplex (Wright et al., 2006), and is also present in other TAF-containing complexes like TFTC/SAGA (Wieczorek et al., 1998). We confirmed TBP depletion, TAF4 expression, and TAF4 IP in control and IAA-treated cells by western blot analysis (Figure 5A). From our TAF4 IP-MS analysis, we observed an 80% decrease of co-precipitated TBP in IAA-treated cells compared to control cells (Figure 5B), although we observed a greater decrease by western blot analyses of whole cell lysates (Figure 5A). In the TAF4 IP-MS of control cells, we identified peptides for all of the TAF subunits of the TFIID complex, as well as some peptides for TAF paralogs like TAF4b and TAF9b. Importantly, in the IAA-treated cells, we detected similar levels of enrichment for the identified TAF subunits (Figure 5C, Figure 5—figure supplement 1A–B and D). Statistical analyses of replicates show no significant differences in the enrichment levels of TFIID members by TAF4 IP-MS in control vs. IAA-treated cells. These results suggest that the full TFIID complex can form even when TBP is depleted. Furthermore, the nuclear TFIID complex does not dissociate into smaller subcomplexes when TBP is depleted, as gel filtration analysis from control and TBP-depleted nuclear lysates showed no change in TAF4 elution patterns. TAF4 is only detected at the earliest fractions, indicative of the high molecular weight of the full TFIID complex, in both control and TBP-depleted nuclear lysates (Figure 5—figure supplement 1C). Although the gel filtration only provides information on the nuclear complex, we observed no changes in relative stoichiometries of the TAF subunits, as measured by the TAF4 IP-MS (Figure 5—figure supplement 1D), suggesting that TBP depletion has no effect on any TAF4 subcomplexes formed in the cytoplasm. Taken together, these results indicate that, even under TBP-depleted conditions, the full TFIID complex can form, and that any subcomplexes containing TAF4 are unaffected.

Figure 5 with 1 supplement see all
The TFIID complex forms upon TATA-box binding protein (TBP) depletion.

(A) Top: Western blot analyses for α-TAF4, α-TBP, and α-Tubulin of control and indole-3-acetic acid (IAA)-treated C64 whole cell extracts used in the TAF4 IP-MS analyses. Input represents 5% of the protein samples used in the TAF4 IPs. Bottom: Western blot analyses for α-TAF4 and α-TBP of TAF4 IPs from control and IAA-treated C64 whole cell extracts. (B–C) Boxplots showing abundance of TBP (B) and the TBP-associated factors (TAFs) (C) from α-TAF4 IP-MS analyses of control (blue) and IAA-treated (red) C64 mouse embryonic stem cells (mESCs). Protein levels were normalized to levels in α-IgG pull-downs from the same cells. Two asterisks represent p≤0.01 using standard two-tailed t-test. Error bars represent standard deviation of n=4.

Figure 5—source data 1

Full uncropped membrane scans of immunoblot analyses with α-TAF4 for control and indole-3-acetic acid (IAA)-treated C64 whole cell extracts without and with annotations.

Input represents 5% of the protein samples used in the TAF4 IPs. Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in analyses. Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 5A.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig5-data1-v2.zip
Figure 5—source data 2

Full uncropped membrane scans of immunoblot analyses with α-TBP for control and indole-3-acetic acid (IAA)-treated C64 whole cell extracts without and with annotations.

Input represents 5% of the protein samples used in the TAF4 IPs. Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in analyses. Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 5A.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig5-data2-v2.zip
Figure 5—source data 3

Full uncropped membrane scans of immunoblot analyses with α-Tubulin for control and indole-3-acetic acid (IAA)-treated C64 whole cell extracts without and with annotations.

Input represents 5% of the protein samples used in the TAF4 IPs. Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in analyses. Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 5A.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig5-data3-v2.zip
Figure 5—source data 4

Full uncropped membrane scans of immunoblot analyses with α-TAF4 for TAF4 IPs from control and indole-3-acetic acid (IAA)-treated C64 whole cell lysates without and with annotations.

Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 5A.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig5-data4-v2.zip
Figure 5—source data 5

Full uncropped membrane scans of immunoblot analyses with α-TBP for TAF4 IPs from control and indole-3-acetic acid (IAA)-treated C64 whole cell lysates without and with annotations.

Relevant bands for Figure 5A are highlighted with a red rectangle. L=ladder, x=lanes not used in Figure 5A.

https://cdn.elifesciences.org/articles/83810/elife-83810-fig5-data5-v2.zip

Effects of TBP depletion on DNA binding of TAF subunits and TFIIA

Next, we examined the DNA-binding profiles of certain TAF subunits. TAF1 is the largest subunit of TFIID, interacts with TBP through an N-terminal TBP-binding sequence, and contains multiple DNA-binding domains (Curran et al., 2018). We performed TAF1 CUT&Tag in control and TBP-depleted C64 mESCs. TAF1 binds at the promoter regions of Actb and Gapdh at similar levels in control and IAA-treated cells (Figure 6A, Figure 6—figure supplement 1A). We then plotted the genome-wide occupancy of TAF1 in a 2 kb window surrounding the TSS for all genes as heatmaps (Figure 6B, Figure 6—figure supplement 1B), and displayed the normalized read counts for TAF1 on the promoter of all genes with and without IAA treatment as a scatter plot (Figure 6C). From the heatmaps and scatter plots, we observed no major changes in TAF1 occupancy after TBP depletion. Pearson correlation analysis using 10 kb bins or surrounding the TSS (-/+1000 bp) shows that the TAF1 control and IAA samples correlate with one another (Figure 6—figure supplement 1C). Furthermore, k-means clustering for changes in TBP binding upon IAA treatment shows that TAF1 binding does not change for genes that experience the largest depletion of TBP (Figure 6—figure supplement 3A, cluster 3). Scatter plot analysis of TAF1 signal at each promoter relative to Pol II levels show that genes do not decrease in either TAF1 or Pol II occupancy upon TBP depletion (Figure 6—figure supplement 3B). Therefore, TAF1 binding is unaffected by TBP depletion.

Figure 6 with 3 supplements see all
Effects of TATA-box binding protein (TBP) depletion on the DNA binding of TAF1, TAF4, and TFIIA.

(A) Gene browser tracks of Actb (top) and Gapdh (bottom) for cleavage under targets and tagmentation (CUT&Tag) analyses of TAF1 (blue), TAF4 (red), and TFIIA (gray). TBP signal (orange) from Figure 1 is replotted for comparison. (B) Genome-wide average plots (top) and heatmaps (bottom) arranged by decreasing Polymerase II (Pol II) occupancy for TAF1 (left), TAF4 (middle), and TFIIA (right) CUT&Tag in a 2 kb window surrounding the transcription start site (TSS) of all genes for control and indole-3-acetic acid (IAA)-treated C64 mouse embryonic stem cells (mESCs). (C) Normalized read counts of TAF1 (left), TAF4 (middle), and TFIIA (right) CUT&Tag for control vs. IAA-treated C64 mESCs at the promoter (–250 bp to the TSS) region for all genes.

We next analyzed TAF4 occupancy by performing CUT&Tag in control and TBP-depleted C64 mESCs. We observed that TAF4 binds on the promoters of Gapdh and Actb in control cells (Figure 6A, Figure 6—figure supplement 1A); however, after TBP depletion, we observed a moderate decrease of TAF4 occupancy at these loci. We plotted TAF4 occupancy in a 2 kb window surrounding the TSS of all genes in control and TBP-depleted cells as heatmaps and observed a similar decrease in TAF4 occupancy across all genes upon TBP depletion (Figure 6B, Figure 6—figure supplement 1D). We also plotted the normalized TAF4 read counts on the promoter of all genes with and without IAA treatment as a scatter plot (Figure 6C) and observed that most points fall below the diagonal with about 65% decrease in signal. Pearson correlation analysis using 10 kb bins or surrounding the TSS (-/+1000 bp) shows that the TAF4 control samples correlate more strongly with one another than with the IAA-treated, confirming that TBP depletion does indeed affect TAF4 occupancy (Figure 6—figure supplement 1D). Furthermore, when we performed k-means clustering of the TAF4 CUT&Tag data, we observed that genes with the largest decrease in TBP occupancy after IAA treatment also experienced the greatest decrease in TAF4 occupancy (Figure 6—figure supplement 3A). Scatter plot analysis of TAF4 relative to Pol II shows that genes that experience a decrease in TAF4 occupancy upon TBP depletion do not decrease in Pol II occupancy (Figure 6—figure supplement 3B). Therefore, unlike TAF1, TAF4 binding is affected by TBP depletion.

Previous studies have shown that TBP mediates the interaction between TFIID and TFIIA (Kraemer et al., 2001). Indeed, in the TAF4 IP-MS, we detected high levels of TFIIA-γ peptides under control conditions that significantly decreased in the IAA-treated conditions (Figure 6—figure supplement 2A), confirming that TBP mediates this interaction. The decrease in interaction is not due to altered protein levels, as measured by whole cell lysate western blot (Figure 6—figure supplement 2A). To determine the effects of TBP depletion on TFIIA binding to promoters, we performed CUT&Tag using an antibody that detects the second subunit of TFIIA, with high concordance between replicates (Figure 6—figure supplement 2B–D). We observe that TFIIA binds to the promoters of Gapdh and Actb in control conditions, and that this occupancy is decreased upon IAA treatment (Figure 6A, Figure 6—figure supplement 2B). We then plotted TFIIA occupancy in a 2 kb window surrounding the TSS of all genes in control and IAA-treated cells as a heatmap and observed a global decrease in occupancy upon TBP depletion (Figure 6B, Figure 6—figure supplement 2C). Scatter plot analysis of normalized reads on the promoters of all genes in control versus TBP-depleted cells show about 60% decrease in overall signal (Figure 6C). Pearson correlation analysis using 10 kb bins or surrounding the TSS (-/+1000 bp) show that TFIIA IAA-treated samples correlate more closely with one another than with the control samples and vice versa (Figure 6—figure supplement 2D). Additionally, k-means clustering of the TFIIA CUT&Tag data shows a similar trend to that of TAF4, where genes that experience the greatest decrease in TBP depletion (cluster 3) also show the largest reduction in TFIIA signal (Figure 6—figure supplement 3A). Scatter plot analysis of TFIIA signal relative to Pol II shows that although the occupancy of TFIIA decreases, Pol II occupancy is unaffected (Figure 6—figure supplement 3C). A similar analysis between TFIIA and TAF4 (Figure 6—figure supplement 3C) shows that genes experiencing a decrease in TAF4 occupancy also decrease in TFIIA binding.

Lastly, we plotted TBP, Pol II, TAF4, TAF1, TFIIA CUT&Tag, and our NET-seq data as heatmaps (top) and average plots (bottom) arranged by decreasing TBP occupancy and observe that TBP occupancy correlates strongly with transcription levels (Pol II and NET-seq data) and with the other GTFs in control conditions, as expected (Figure 6—figure supplement 3D). Furthermore, highly bound TBP genes show the greatest reduction in TFIIA and TAF4 occupancy when TBP is depleted, but experience no differences in Pol II and TAF1. Taken together, our data suggest that TBP depletion disrupts TFIID-TFIIA association and negatively affects TFIIA and TAF4 binding to promoters. However, despite these effects on TAF4 and TFIIA, the Pol II machinery remains sufficiently robust in supporting Pol II transcription.

Discussion

In this study, we took advantage of a drug-inducible degradation system to deplete TBP rapidly and examined the mechanistic role of TBP in Pol II transcription in mESCs. We found that TBP is largely not required for ongoing Pol II transcription. TBP is, however, required for transcription of tRNA genes by Pol III. We tested several possible mechanisms for TBP-independent Pol II transcription, and found that TBP is not required for gene induction, and that the TBP paralog TRF2 does not functionally replace TBP for Pol II transcription when TBP is depleted. Additionally, the TFIID complex still forms even when TBP is acutely depleted, but that its association with TFIIA is significantly decreased. Although TAF1 binding was unaffected, TAF4 and TFIIA occupancy levels decreased upon TBP depletion. Despite these changes, Pol II activity remains unaffected, suggesting a robustness in the Pol II machinery that compensates for TBP depletion.

Previous studies have examined the role of individual subunits of the TFIID complex in mammalian systems. For instance, knockdown of the TAF5 subunit in human ESCs led to reduced expression of the pluripotency genes (Pijnappel et al., 2013), and in mESCs, acute degradation of TAF12 led to a global decrease in gene expression (Sun et al., 2021). These studies are consistent with a more prominent role for TFIID. However, a major limitation inherent to the mAID system is the incomplete degradation of the protein, resulting in ~10–20% remaining TBP protein. A previous study has quantified the absolute number of TBP molecules in mESCs to 99,111±29,125 molecules (Cattoglio et al., 2019), suggesting ~10,000–20,000 remaining TBP molecules after depletion. From the Pol II CUT&Tag and NET-seq data, ~70% of Refseq TSSs have detectable signals, which results in ~27,185 active genes (from 38,837 total). Based on these numbers, even the highest estimated end of remaining TBP is insufficient to account for one molecule per active TSS. These numbers do not consider Pol II expression that occurs at enhancers or the ribosomal genes transcribed by Pol I. Furthermore, the TBP requirement for Pol III activity in mESCs provides confirmation that the acute TBP depletion system in mESCs has the expected consequences, and emphasizes the lack of effect on Pol II transcription when TBP is depleted. However, these results do not provide information on the functional role of TBP when recruited to Pol II promoters. One possibility is that Pol II transcription has evolved to have multiple redundancies such that removal of one factor (TBP) may not fully disrupt the whole system. Another possibility is that the transient nature of the embryonic cell state, with its unique chromatin landscape, enables Pol II transcription without TBP. Regardless, at the level of TBP depletion that we observe, we conclude that in mESCs Pol II transcription can occur independently of TBP.

We initially hypothesized that the TBP paralog TRF2 could functionally replace TBP in our depletion system. TRF2 is a more distant TBP paralog compared to the other paralogs, sharing only ~40% amino acid identity with the core domain of TBP (Dantonel et al., 1999). In Drosophila, TRF2 fails to bind to DNA that contains the canonical TATA box. Instead, TRF2 from various species such as Drosophila, worms, frogs, fish, and flies has been proposed to activate transcription for TATA-less promoters (Akhtar and Veenstra, 2011; Isogai et al., 2007). Previous studies have also shown that mouse TRF2 binds to active promoters in haploid testis cells along with TBP and TAF7l (Martianov et al., 2016), but the germline may have unique regulatory mechanisms. Our CUT&Tag results show that TRF2 binds to all active promoters in mESCs, and its level of binding correlates strongly to both TBP and Pol II (Figure 4—figure supplement 1C–D). One possible explanation for species-specific differences in TRF2 functions is the variability in the TRF2 core domain across species. Unlike TBP, whose core domain is over 80% conserved across species, the core domain of TRF2 only shares 40–45% amino acid conservation between metazoans (Akhtar and Veenstra, 2011; Dantonel et al., 2000). Although TRF2 has been historically seen to govern specific subsets of genes distinct from TBP, our results show that TRF2 in mESCs may have the potential to substitute for TBP during transcription initiation for mESCs Pol II-transcribed genes. It was thus surprising that depletion of TBP in the TRF2 knock-out mESCs had no global effect on Pol II transcription (Figure 4E, Figure 4—figure supplement 1E), suggesting that Pol II transcription in mESCs may not only be TBP-independent, but TBP family-independent.

We have shown that the full TFIID complex can form upon TBP depletion and that TFIID does not dissociate into smaller subcomplexes when TBP is depleted, suggesting a resiliency in the TFIID complex. Several studies point to a modular assembly for the full complex, which may allow for such resiliency. For instance, human TAF8 and TAF10 form as a heterodimer in the cytosol (Kamenova et al., 2019), with TAF2 joining the subcomplex to promote the nuclear import and incorporation into human TFIID (Kamenova et al., 2019; Soutoglou et al., 2005; Trowitzsch et al., 2015). Similarly, human TAF5, TAF6, and TAF9 have also been shown to form in the cytosol as a subcomplex to promote the incorporation of TAF9 into TFIID (Antonova et al., 2018). It is possible that the modular nature of the assembly of TFIID allows for a robust structure even when TBP molecules are severely limiting. Furthermore, TAF7 has been shown to dissociate from the rest of TFIID after transcription initiation and acts as a regulator of RNA translation by binding onto RNAs and delivering them to polysomes for translation in the cytosol (Gegonne et al., 2006; Cheng et al., 2021). This finding suggests a more dynamic TFIID complex with subunits associating and dissociating in flux, which may allow for built-in redundancy and robustness when a subunit is missing. Such a dynamic TFIID complex would also be consistent with the different effect on TAF1 and TAF4 DNA-binding profiles when TBP is depleted, especially if the absence of TBP leads to changes in TAF-DNA residence times. In contrast, TBP in Pol III initiation forms the TFIIIB complex with only two other factors: Brf2 and Bdp1 (Gouge et al., 2017). It is intriguing to speculate that the greater number of TAFs in TFIID (13 or more) might allow for more redundancy than the two other subunits of TFIIIB, providing a potential mechanism to the observed requirement for TBP in Pol III but not for Pol II transcription.

Previous studies have identified the presence of other non-canonical initiation complexes, suggesting a high level of robustness in the machinery to accommodate Pol II transcription. For example, TAF paralogs are often expressed and used in a cell type-specific manner (Freiman, 2009), including TAF4b in oocytes and TAF9b in motor neuron differentiation (Herrera et al., 2014; Liu et al., 2011; Freiman et al., 2001; Pointud et al., 2003). In Xenopus embryos, knockdown of TBP and the two paralogs showed recruitment of Pol II to the promoter and active transcription for a subset of genes (Gazdag et al., 2016). In yeast, transcription activation can occur in the absence of TFIIA, and in vitro TFIIA was shown not to be needed for basal transcription (Chou et al., 1999; Sayre et al., 1992). Additionally, gel mobility-shift analysis and fractionated cross-substitution experiments using Drosophila and HeLa systems showed that TFIIA was not required for complexes to be transcriptionally competent (Tyree et al., 1993; Wampler et al., 1990). Our results provide additional evidence on the robustness of the Pol II machinery. Despite the changes in DNA binding of TAF4 and TFIIA when TBP is depleted, Pol II transcription remains unaffected. Perhaps in the same way that TFIID could behave as a dynamic complex, the entire PIC may also function as a dynamic hub instead of acting as a monolithic complex. Previous studies using single molecule imaging and tracking provide support for this concept. For example, the average residence time of TBP is on the order of minutes whereas TFIIB is on the order of seconds (Teves et al., 2018; Zhang et al., 2016). Such a dynamic system may provide sufficient robustness to facilitate Pol II transcription when TBP is depleted.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Cell line (Mus musculus, male)JM8.N4 mouse ES cellsKOMP repositoryRRID: CVCL_J962Parental cell line used for all genetic manipulations
Cell line (Mus musculus, male)mAID-TBP KI C94https://doi.org/10.7554/eLife.35621C94Endogenous knock-in of the minimal auxin inducible degron to N-terminus of TBP in JM8.N4 cells
Cell line (Mus musculus, male)Halo-Pol II C64https://doi.org/10.7554/eLife.35621C64Endogenous knock-in of the HaloTag to C-terminus of Rbp1, largest subunit of Pol II, in mAID-TBP C94 cells
Cell line (Mus musculus, male)B8This paperEndogenous knock-out of the Tbpl1 gene from Halo-Pol II C64 cells – see Materials and Methods
Cell line (Mus musculus, male)B8HAThis paperOverexpressed HA-tagged TRF2 in B8 cells – see Materials and Methods
Cell line (Drosophila melanogaster)S2Gift from Dr. Eric JanSpike-in of 20% (by cell number) in CUT&Tag and qRT-PCR
Transfected construct (Mus musculus)pTRF2-HAThis paperHA tagged TRF2 cDNA expressing plasmid – see Materials and Methods
Transfected construct (Mus musculus)pCas9-mCherry-TRF2-1This paper; backbone:
cat #64324 (Addgene)
Targets Tbpl1 for knockout - see Materials and Methods
Transfected construct (Mus musculus)pCas9-mCherry-TRF2-2This paper; backbone:
cat #64324 (Addgene)
Targets Tbpl1 for knockout - see Materials and Methods
Antibodyα-TBP (Mouse, monoclonal)Abcamcat:ab51841Western blot (1:3000) ChIP (5 μg) CUT&Tag (1 μg) IF (1:100)
Antibodyα-Pol II (Rpb1 CTD)Cell Signaling Technologiescat:D1G3KCUT&Tag (1 μg)
Antibodyα-H3K27me3 (Rabbit, monoclonal)Cell Signaling Technologiescat:C36B11Western blot (1:7000) CUT&Tag (1 μg)
Antibodyα-HA (Rabbit, polyclonal)Epicyphercat:13–2010Western blot (1:1000) CUT&Tag (1 μg)
Antibodyα-Tubulin (Rabbit, polyclonal)Abcamcat:ab6046Western blot (1:7000)
Antibodyα-Pol III RPC32 (mouse monoclonal)Santa Cruz Biotechnologycat:sc-48365Western blot (1:2000)
Antibodyα-TAF4 (mouse monoclonal)Santa Cruz Biotechnologycat:sc-136093Western blot (1:2000) CUT&Tag (1 μg) Co-IP (3 μg)
Antibodyα-H3.3 (mouse monoclonal)Abnovacat:H00003021-M01Western blot (1:5000)
Antibodyα-TRF2 (mouse monoclonal)Gift from Dr. László ToraWestern blot (1:3000)
Antibodyα-TFIIA-γ (mouse monoclonal)Santa Cruz Biotechnologycat:sc-374483Western blot (1:3000)
Antibodyα-TFIIA-γ (rabbit polyclonal)Santa Cruz Biotechnologycat:sc-25365CUT&Tag (1 μg)
AntibodyIRDye 800CW α-Mouse IgG (Goat polyclonal)Licorcat:926–32210Western blot (1:20000)
AntibodyIRDye 800CW α-Rabbit IgG (Goat polyclonal)Licorcat:925–32211Western blot (1:20000)
AntibodyIRDye 680RD α-Mouse IgG (Goat polyclonal)Licorcat:926–68070Western blot (1:20000)
AntibodyIRDye 680RD α-Rabbit IgG (Goat polyclonal)Licorcat:926–68701Western blot (1:20000)
Antibodyα-Mouse IgG (Rabbit, polyclonal)Abcamcat:ab46540ChIP (5 μg) CUT&Tag (1:100)
Antibodyα-TAF1 (Rabbit, monoclonal)Cell Signaling Technologiescat:D6J8BCUT&Tag (1 μg)
Antibodyα-Rabbit IgG (Guinea Pig, polyclonal)Antibodies-onlinecat:ABIN101961CUT&Tag (1:100)
AntibodyAlexa Fluor 594 α-Mouse IgG (Goat, polyclonal)Invitrogencat:A-11005IF (1:100)
Antibodyα-Mouse IgG (mouse, monoclonal)Santa Cruz Biotechnologycat:sc-2025Co-IP (3 μg)
Sequence-based reagentBiotinylated oligonucleotides for NET-seq librariesThis paperSee Materials and Methods and Supplementary file 5 for full list
Sequence-based reagentqRT-PCR oligonucleotidesThis paperSee Supplementary file 6 for full list
Sequence-based reagentChIP-qPCR oligonucleotidesThis paperSee Supplementary file 7 for full list
Peptide, recombinant proteinpA-Tn5Gift from Dr. Steven HenikoffDiluted to 1 x in each reaction
Peptide, recombinant proteinpAG-Tn5Epicyphercat:15–1017Diluted to 1 x in each reaction

Cell lines

Request a detailed protocol

For all cell lines used in this study, the parental line is JM8.N4 mouse ES cells, purchased from KOMP repository, RRID: CVCL_J962. C64 is a CRISPR-Cas9 genetically modified JM8.N4 cell line containing mAID-TBP knock-in obtained as previously described (Teves et al., 2018). B8 is a CRISPR-Cas9 genetically modified C64 cell line with a Tbpl1 complete gene knock-out. The cell lines have been authenticated by STR profiling, and tested negative for mycoplasma contamination.

Cell culture

Request a detailed protocol

ES cells were cultured on 0.1% gelatin-coated plates in ESC media KnockOut DMEM (Corning) with 15% FBS (HyClone), 0.1 mM MEM non-essential amino acids (Gibco), 2 mM GlutaMAX (Gibco), 0.1 mM 2-mercaptoethanol (Sigma-Aldrich) and 1000 units/ml of ESGRO (Chem-icon). ES cells were fed daily, cultured at 37°C in a 5% CO2 incubator, and passaged every 2 days by trypsinization. For endogenously tagged mAID-TBP C64 and B8 cells, TBP degradation was performed by addition of IAA at 500 μM final concentration to a confluent plate of cells for 6 or 16 hr. HS was performed at 42°C in a 5% CO2 incubator for 30 min. For HS and IAA treatment, cells were first treated with 6 hr of auxin followed by an additional 30 min before being collected. RA (Sigma-Aldrich R2625-100MG) treatment was performed at 37°C in a 5% CO2 incubator at 0.25 μM for 16 hr. Cells treated with RA and IAA were incubated for 16 hr at 0.25 μM and 500 μM, respectively.

Cellular fractionation

Request a detailed protocol

Cell pellets (15 million) were lysed with Buffer A (0.1% Triton X-100, 10 mM HEPES pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10% glycerol, 1 mM DTT, 1× protease inhibitor mix) and incubated on ice for 8 min. Resulting nuclei were collected by centrifugation at 3000 × g at 4°C. The supernatant was collected as the cytoplasmic fraction, and the nuclei were resuspended and washed in Buffer B (3 mM EDTA, 0.2 mM EGTA, 1 mM DTT, 1× protease inhibitor mix) twice. The nuclei were lysed with Buffer B+1% SDS on ice for 15 min before being mixed with 4× SDS (8% SDS, 200 mM Tris-HCl pH 6.8, 40% glycerol, 50 mM EDTA, 4% beta-mercaptoethanol, 0.04 % wt/vol bromophenol blue) in preparation for western blot.

Generation of Tbpl1 knock-out B8 cell line with CRISPR-Cas9

Request a detailed protocol

Two Guide RNAs (gRNAs) targeting the entire locus of Tbpl1 was designed using UC Santa Cruz CRISPR Guide RNA Design tool (http://crispor.tefor.net) and cloned into the pU6-(BbsI)_CBh-Cas9-T2A-mCherry (Addgene #64324). mESCs were grown to 50% confluency and then transfected with 1 μg of each Cas9 gRNA plasmid using Lipofectamine 2000 (Invitrogen 11668-019) according to the manufacturer’s protocol. Cells were sorted 1 day after transfection gated for mCherry and GFP (H3.3 is endogenously tagged with GFP in all cell lines used), and 15,000 transfection positive cells were plated onto a 15 cm tissue culture plate. Cells were grown until individual colonies were visible and single colonies were transferred to a 96-well for screening. Seventy percent of each cell colony was used for genotyping, while the other 30% was grown for maintenance. For genotyping, cells were lysed using DirectPCR lysis reagent (Viagen Biotech #302C) according to the manufacturer’s protocol and lysates were used in a screening PCR to identify edited cells. Identified clones were further validated using Sanger sequencing (not shown), qRT-PCR, and western blotting. A list of gRNAs and primers used for knock-out and screening are in Supplementary file 6.

Generation of the overexpressed TRF2-HA cell line

Request a detailed protocol

mESCs were grown to 50% confluency and transfected using 1 μg of the TRF2-HA construct together with 1 μg of the Super Piggybac transposase plasmid (Gift from Tjian Lab) using Lipofectamine 2000 (Invitrogen 11668-019) according to the manufacturer’s protocol. After 24 hr, 500 μg/mL of G418 (Fisher BP673-5) was added to select for successful random integration of the construct. Cell media containing antibiotics was refreshed daily until all negative control cells were dead.

Antibodies for western blot

Request a detailed protocol

Primary antibodies: α-TBP 1:3000 (Abcam ab51841), α-H3K27 1:7000 (Cell Signaling Technologies C36B11), α-HA 1:1000 (EpiCypher 13-2010), α-Tubulin 1:7000 (Abcam ab6046), α-RPC7 1:2000 (Santa Cruz Biotechnology sc-48365), α-TAF4 1:2000 (Santa Cruz Biotechnology sc-136093), α-H3.3 1:5000 (Abnova, H00003021-M01), α-TRF2 1:3000 (Gift from Dr. László Tora), and α-TFIIA-γ 1:3000 (Santa Cruz Biotechnology, sc-374483). Secondary antibodies: IRDye 800CW Goat anti-mouse (Licor 926-32210), IRDye 800CW Goat anti-rabbit (Licor 925-32211), IRDye 680RD Goat anti-mouse (Licor 926-68070), or IRDye 680RD Goat anti-rabbit (Licor 926-68701).

ChIP-qPCR

Request a detailed protocol

Control or IAA-treated cells were crosslinked in 1% formaldehyde for 5 min at room temperature and quenched with the addition of glycine to 0.125 M for 5 min at room temperature. Cells were harvested and washed twice in cold 1× PBS, centrifuged for 5 min at 1000 × g at 4°C, and lysed on ice for 5 min with lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl pH 8.0, 1× protease inhibitors, 0.2 mM PMSF, 1 mM benzamidine). To fragment DNA, whole cell lysates were sonicated for 10 min (30 s on/30 s off intervals, Diagenode Biorupter). Lysates were then pre-cleared using 50 µL of magnetic Protein G Dynabeads and incubated on an end-over-end rotator for 2 hr at 4°C. Samples were then placed on a magnetic rack and aliquoted (30% vol/vol for input, 35% vol/vol for α-TBP IP, 35% vol/vol for α-IgG IP). IP buffer (150 mM NaCl, 20 mM Tris-HCl pH 8, 1% Triton X-100, 1× protease inhibitor, 0.2 mM PMSF, 1 mM benzamidine) was added to the lysates to obtain a total volume of 1 mL per immunoprecipitation. Lysates were then either incubated with 5 μg of α-TBP (Abcam ab51841) or anti-Mouse IgG (Abcam ab46540) overnight at 4°C, and bound to 50 μL of magnetic Protein G Dynabeads while rotating for 2 hr at 4°C. Beads were washed thrice with wash buffer A (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris-HCl pH 8), twice with wash buffer B (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM NaCl, 20 mM Tris-HCl pH 8), and thrice with wash buffer C (2 mM EDTA, 20 mM Tris-HCl pH 8, 10% glycerol). Crosslinked DNA was eluted thrice from the beads with 150 μL of elution buffer (1% SDS, 100 mM NaHCO3) for 15 min at 65°C, with vortexing every 3 min for a total of 450 μL of eluate, followed by RNAseA treatment for 1 hr at 37°C. Twenty μg proteinase K and NaCl (final conc: 0.3 M) were added to each eluate and input, and crosslinks were reversed overnight at 65°C. Following proteinase K digestion, DNA was purified by phenol-chloroform isolation and resuspended in 50 μL of TE buffer. Two μL of ChIP DNA was used in each qPCR with Luna Universal qPCR Master Mix (M3003) according to the manufacturer’s instructions. Samples were run on the QuantStudio 3 Real-Time PCR System. Primers used in qPCR are listed in Supplementary file 7.

NET-seq

Request a detailed protocol

NET-seq was performed as previously described (Mayer and Churchman, 2016) with the following modifications. Approximately 10–15 million mESCs were used in each fractionation. Drosophila S2 cells were added to each fractionation (3% by cell count) as spike-in for downstream analyses. Cells were lysed with cytoplasmic lysis buffer on ice for 7 min and nuclei were lysed in nuclei lysis buffer on ice for 2 min. Barcoded DNA linkers were riboadenylated as previously described (Song et al., 2015), with the following modifications. Adenylation reactions were performed in a reaction mixture (10 µL) containing 0.8 μg of barcoded DNA linker, 1× T4 RNA ligase buffer (NEB, B0216L), 2 mM ATP, 35% PEG, and 300 U of T4 RNA Ligase 1 (NEB, M0204L), incubated at 37°C for 8 hr, followed by 15 min of heat inactivation at 65°C. Riboadenylated barcoded DNA linkers were stored at –80°C until use. NET-seq library preparation was performed as described previously (Mayer and Churchman, 2016). cDNA containing sequences of a subset of sequenced snRNAs, snoRNAs, and rRNAs (Supplementary file 5) were specifically depleted using biotinylated DNA oligos (Integrated DNA Technologies). Sequencing was performed at the UBC Biomedical Research Centre with 55 bp single-end reads.

Processing and alignment of NET-seq reads

Request a detailed protocol

NET-seq data was processed as described in Mayer et al., 2015. Reads were trimmed and aligned using STAR (v.2.7.3a). Reverse transcription mispriming events, PCR duplication events, and splicing intermediates were removed using custom Python scripts provided by the Churchman group (https://github.com/churchmanlab, Mayer et al., 2015). The bam files of biological replicates were merged using SAMtools and normalized by subsampling each sample by the number of aligned Drosophila reads across all samples using SAMtools (Li et al., 2009). BedGraph coverage files for gene plots were generated from normalized bam files with a custom Python script provided by the Churchman group (https://github.com/churchmanlab) and visualized by IGV. Bigwig coverage files for heatmaps and TSS plots were generated using deepTools with the following parameters: -bs 1 --Offset 1 (Ramírez et al., 2016). Heatmaps and TSS plots were generated using deepTools with the following parameters: -a 1000 -b 1000 -bs 10 followed by filterValues --max 50. Read counts across tRNAs were generated from normalized bam files using bedtools and analyzed with GraphPad Prism (Quinlan and Hall, 2010). Read counts used in edgeR analysis were generated from bam files using featureCounts (Liao et al., 2014). Reads were then imported into R Studio and analyzed using the DGE tool of the edgeR Bioconductor package (Robinson et al., 2010). First, DGEList was used to specify reads and gene names, followed by filtering (filterByExpr) and normalization by TMM (default). Once reads are normalized, the design matrix was built and dispersion was estimated to determine biological variation. The command glmFit was used to determine differentially expressed peaks/genes (DEG) and the command topTags was used to show the top genes. Scatter plots were generated using the plotMD function and significant genes were obtained along with the raw values for downstream analyses such as heatmap and table generation.

CUT&Tag

Request a detailed protocol

CUT&Tag was performed as previously described (Kaya-Okur et al., 2019), but with the following modifications. Cells were harvested at room temperature and 100,000 mESCs were used per sample. With the EpiCypher pAG-Tn5 (EpiCypher EP151117), cryopreserved Drosophila S2 cells were spiked in at 20% (20,000 S2 per 100,000 mESCs) and 3% for the Pol II in the C64 6 hr IAA sample (Figure 1—figure supplement 2C) to account for a spike-in control, during the cell pelleting step before washing. pAG-Tn5 from the Henikoff lab did not require a spike-in since it contained trace amounts of E. coli. Antibodies used include TBP (Abcam ab51841), RNA Pol II (Cell Signaling Technology D1G3K), RNA Pol III (Santa Cruz Biotechnology sc-21754), Rabbit anti-Mouse IgG (Abcam ab46540), α-H3K27me3 (Cell Signaling Technology C36B11), α-HA (EpiCypher 13-2010), α-TAF1 (Cell Signaling Technology D6J8B), α-TAF4 (Santa Cruz Biotechnology sc-136093), and α-TFIIA (Santa Cruz Biotechnology sc-25365). Secondary antibodies used include Guinea Pig anti-Rabbit IgG (Antibodies-Online ABIN101961) and Rabbit anti-Mouse IgG (Abcam ab46540). Secondary antibody incubation times were 45 min at room temperature. The Henikoff lab pA-Tn5 adapter complex or the commercial Epicypher pA-Tn5 was added at a final concentration of 1:250 or 1:20, respectively. Sequencing was performed at the UBC Biomedical Research Centre using the NextSeq500 75 cycles and the NextSeq2000 50 cycles.

CUT&Tag analysis

Request a detailed protocol

Reads were mapped on mm10 genome build using Bowtie2 with the following parameters: --no-unal --local --very-sensitive-local --no-discordant --no-mixed --contain --overlap --dovetail --phred33 –I 10 –X 9999. PCR duplicate reads were kept as these sites may represent real sites from adapter insertion from Tn5 as per recommendation from the Henikoff lab. A normalization factor was determined from E. coli (normalized to the control) or Drosophila melanogaster (normalized to the control) alignment from Bowtie2 mapping and used to scale IAA-treated samples to control during generation of bigwig files. For the ChiPseqSpikeInFree pipeline, Pol II fastq files were subsampled to the lowest reads for each CUT&Tag run: 8 million reads (replicate 1&2) for C64 samples; 2 million reads (replicate 1&2); and 4 million reads (replicate 3&4) for B8 samples. Downstream analyses, heatmaps, TSS plots, Gene plots, k-means clustering were performed using IGV, DeepTools, and BedTools suite (Ramírez et al., 2016; Quinlan and Hall, 2010). ComputeMatrix from deeptools was done using binsize 10. Replicates were merged using the BigWigMerge tool, which sums the reads after normalizing treatment samples to the control samples of each run using Drosophila or E. coli spike-in. ChIPseqSpikeInFree provided its own normalization factors when running the pipeline. Alternate reference alignment was obtained from the UCSC browser tracks.

DGE analysis using edgeR bioconductor

Request a detailed protocol

Scatter plots of differential gene expression (DGE) analysis were performed using the DGE tool from the edgeR bioconductor package. Starting subsampled bam files were used for all Pol II samples. Read counts were obtained using featureCounts command with paired end specificity for the entire gene and pseudogenes were removed (Liao et al., 2014). Reads were then imported into R Studio and analyzed using the DGE tool following the Bioconductor manual using default parameters (i.e. 5% FDR). First, DGEList was used to specify reads and gene names, followed by gene annotations (optional) then filtering and normalization by TMM (default option). Read counts below 20 were filtered using filterByExpr (min.count=20). Once reads were normalized, the design matrix was built and dispersion was estimated to determine biological variation. The command glmFit was used to determine DEG and the command topTags was used to show the top genes. Scatter plots were generated using the plotMD function and significant genes were obtained along with the raw values for downstream analyses such as heatmap and table generation.

Gene ontology

Request a detailed protocol

GO was done on gene sets identified to be significant from the DGE analysis. The top HS and RA upregulated genes were input into http://geneontology.org/, filtered by molecular function for the M. musculus genome to obtain GO terms.

Pearson correlation analysis

Request a detailed protocol

Pearson correlation analyses were performed using the multiBigwigSummary command from the DeepTools suite with the bins option and the bedfile option for the regions surrounding the TSS for TAF1, TAF4, and TFIIA (-/+1000 bp of the TSS). The Pearson correlation coefficients were plotted as a heatmap using the plotCorrelation command with the remove outliers and skip zeros options.

Scatter plot read counts analyses and filtering

Request a detailed protocol

Read counts for scatter plots were obtained from bam files using the bedtools multicov command and plotted to the promoters of genes (–250 bp to TSS) (Quinlan and Hall, 2010). Read counts were then normalized to the scaling factor, summed and plotted as is or log transformed before being plotted. Regression analysis was then performed to obtain the slope and R2 value. Control vs. IAA samples were filtered by taking out genes that did not have any signal in the control (0 reads in the control samples). Fold changes between Pol II and TAF1/TAF4/TFIIA were filtered by genes with moderate to high levels of Pol II in the control (Pol II control summed reads 50 and below were excluded). Fold changes between TFIIA and TAF4 were filtered by moderate to high levels of TFIIA and TAF4 in the control (TFIIA and TAF4 control summed reads 50 and below were excluded).

qRT-PCR

Request a detailed protocol

Cells were cultured until ~90% confluency on tissue culture-treated plates at 37°C in a 5% CO2. After indicated treatments, cells were washed with 1× PBS, trypsinized and pelleted by centrifuging at 600 × g. RNA was then extracted from the pellet via Trizol extraction and concentrations were measured via DeNovix Nanodrop. Spike-in RNA was then added at 20% the sample amount. One μg of samples were then DNase treated following the Promega DNase kit (M6101). The entire sample was then reverse transcribed using the New Englands BioLabs LunaScript RT SuperMix kit (E3010). Samples were then diluted to 10 ng/μl and 20 ng was used for qRT-PCR experiment using the New Englands BioLabs Luna Universal qPCR Master Mix kit (M3003). Samples were then processed on the QuantStudio 3 Real-Time PCR System.

Immunofluorescence

Request a detailed protocol

Cells were grown on coverslips (Azn Scientific #ES0117650) that were pre-washed in 70% ethanol and coated with gelatin in tissue culture-treated six-well plates. After indicated treatments, cells were washed with 1 mL of PBS and fixed in 4% paraformaldehyde (UBC Chemical Stores #OR683105) for 15 min. After fixation, cells were washed and permeabilized with 1 mL 0.025% Triton-X for 5 min, followed by two 10 min washes with PBS. Samples were blocked by nutating in PBS 5% BSA for 30 min. TBP (Abcam ab51841) was diluted at 1:100 in PBS, and added to cells in coverslips for 1 hr. Samples are washed twice with PBS (5 min for each wash). Coverslips were then incubated in Alexa Fluor 594 (A11005) 1:100 for 1 hr, and washed with PBS twice (5 min for each wash). Samples are then incubated with DAPI (300 nM in PBS) for 5 min and washed twice with PBS (5 min for each wash). Coverslips were assembled using Vectashield mounting medium (BioLynx #VECTH1000). Fluorescent images were collected using the Leica DMI6000B inverted fluorescence microscope. Quantification was done through Fiji (Schindelin et al., 2012).

CoIP

Request a detailed protocol

Cells were grown to 90% confluency on a 15 cm gelatinized plate. After indicated treatments, cells were washed with 1× PBS, trypsinized and pelleted by centrifuging at 600 × g. Cell pellets were then lysed in 1 mL of Lysis Buffer (200 mM NaCl, 25 mM HEPES, 1 mM MgCl2, 0.2 mM EDTA, 0.5% NP-40, 1× Roche complete inhibitor, 0.2 mM PMSF, 1 mM benzamidine). The whole cell lysates were passed through a 25 G needle five times and incubated on an end-over-end rotator for 30 min at 4°C. Samples were spun at max speed at 4°C for 5 min. Supernatant was transferred to a new tube and the pellet was then digested as much as possible using 2 U of MNase in 1× MNase digestion buffer by nutating at 37°C for 30 min. Digested pellet was centrifuged at max speed and the new supernatant was added to the previous supernatant. Lysates were then pre-cleared using 20 µL/sample of Protein G Dynabeads and incubated on an end-over-end rotator for 2 hr at 4°C. Samples were then placed on a magnetic rack and aliquoted (5% vol/vol for input, 45% vol/vol for α-TAF4 IP, 45% vol/vol for α-IgG IP). Pull-down was performed on the cell lysates by adding 3 µg of α-TAF4 (Santa Cruz Biotechnology sc-136093) or mouse α-IgG (Santa Cruz Biotechnology sc-2025) crosslinked to Protein G Dynabeads and incubating on an end-over-end rotator overnight at 4°C. The next day, beads were washed six times with 600 µL Lysis Buffer. Proteins were eluted from the beads with 50 µL of 0.1 M glycine (pH 2.5) for 30 min and neutralized with 50 µL 1 M Tris-HCl (pH 8.0). Elution was repeated two more times and all eluates were combined. Five percent vol/vol of the final eluate was taken to assess the IPs by western blot, while the rest of the co-IP eluate was flash frozen with LN2 for MS analysis.

Mass spectrometry

Request a detailed protocol

Eluates were purified and digested following the Single-pot, solid-phase-enhanced sample preparation (SP3) method (https://doi.org/10.1038/s41596-018-0082-x) as follows. Eluates were reduced using 2.55 µL of 1 M dithiothreitol (DTT, 10 mM final concentration) for 30 min at 37°C, followed by alkylation using 25.5 µL of 500 mM chloroacetamide (CAA, 50 mM final concentration) for 30 min at room temperature in the dark, and quenched with 14.2 µL of 1 M DTT (50 mM final concentration) for 5 min at room temperature. Sera-Mag Speed Beads were prepared by mixing in equal volume (10 µL hydrophilic and 10 µL hydrophobic per sample), rinsed with water twice, and resuspended in half the volume (total of 10 µL mixed beads per sample) in water. Prepared beads were added to the sample and protein binding was facilitated by adding 100% ethanol (EtOH) to a final concentration of 80%. Samples were incubated for 5 min in a thermomixer at room temperature. Beads were magnetically isolated from the supernatant and proceeded to wash three times with 1 mL 90% EtOH. Excess EtOH presented in beads was removed by brief centrifugation and aeration. Beads were then resuspended in 100 mM ammonium bicarbonate and digested with trypsin/LysC at 1:75 protease:protein (w/w) ratio in 37°C thermomixer overnight.

Peptide digests were magnetically isolated from beads and acidified to pH 3–4 with trifluoroacetic acid (TFA). Peptides were desalted using self-packed C18 StageTips with the following steps: tips were conditioned with methanol, elution buffer and equilibrate with wash buffer before loading acidified peptide samples (loaded for a total of two times), peptides bound on disks were washed with washing buffer twice before eluting in 60 µL of elution buffer. Washing buffer consisted of 0.1% TFA in water and the elution buffer consisted of 60% acetonitrile in 0.1% formic acid (FA). All the above steps were performed in 100 µL volume and spun at 1200 × g unless stated otherwise. Eluted peptides were dried in a Speed Vac.

Mass spectrometry analysis was performed on a Q Exactive HF Orbitrap mass spectrometer coupled to an Easy nLC-1200 Liquid Chromatography system (Thermo Scientific). Dried peptides were resolubilized in 0.1% FA in water spiked with iRT Standard (Biognosys) and peptide concentration was measured by nanodrop. One µg of each sample was loaded onto a 50 cm µPAC HPLC Column (Thermo Scientific) maintained at 45°C and separated by a 1 hr gradient with Buffer B ramping from 4% to 27%, at 300 nL/min gradient flow. Buffer A was 2% ACN in 0.1% FA in water and Buffer B was 95% ACN in 0.1% FA in water. MS acquisition was performed with a full-scan MS spectrum between 300 and 1650 m/z, resolution of 120,000, AGC target of 3e6, and maximum IT of 60 ms followed by DIA MS2 acquisition using 24-variable isolation windows at 30,000 resolution, AGC target of 3e6, auto maximum IT, and stepped NCE of 25.5, 27, 30.

Acquired DIA data was analyzed using directDIA analysis in Spectronaut (version 14.10.201222.47784). The Spectronaut Pulsar search settings were as follows: Enzyme and digest type was Specific Trypsin/P with two missed cleavages. Carbamidomethyl (C) was set as fixed modification; acetyl (N-term) and oxidation (M) were defined as variable modifications. Spectra were matched against the curated mouse proteome fasta from Uniprot (obtained 2021-03-01) with 17,063 entries. False discovery rate (FDR) of 0.01 was set for peptide spectra matching, peptide and protein group identification and the precursor PEP cutoff was set at 1E-10. Missing peptides identifications were imputed using the ‘Global Imputing’ strategy with a Qvalue percentile cutoff of 0.2. The ‘minor group quantity’ was calculated as the mean precursor quantity and the ‘major group quantity’ was calculated as the median of 1–6 peptide quantities. Data was normalized across runs using a Qvalue percentile cutoff of 0.8. All data has been deposited to ProteomeExchange (PXD034171) through MassIVE (MSV000089562).

Gel filtration

Request a detailed protocol

Nuclei was isolated from ~25 million cells per sample as previously described (Mayer and Churchman, 2016) with the following modifications. Nuclei was resuspended in glycerol-free nuclei resuspension buffer (20 mM Tris HCl pH 8, 75 mM NaCl, 1× protease inhibitors, 0.1 mM PMSF) and lysed with equal volume of urea-free nuclei lysis buffer (1% NP-40, 50 mM HEPES, 325 mM NaCl, 1× protease inhibitors, 0.1 mM PMSF, 2 mM MgCl2, 50 µg/mL benzonase) on an end-over-end rotator at 4°C for 30 min. Samples were spun at 21,000 × g at 4°C for 5 min and supernatant was transferred to a new tube for gel filtration analysis.

A Superose 6 Increase 10/300 GL column was pre-equilibrated with running buffer (10 mM Tris HCl pH 8, 0.5% NP-40, 25 mM HEPES, 200 mM NaCl, 0.1 mM PMSF) prior to use; 0.5 mg of nuclear lysates were loaded onto the Superose 6 column using the AKTA Purifier FPLC system (GE Healthcare) and run at 0.5 mL/min. Filtered proteins were detected by absorbance at 280 nm, and 47×500 µL fractions were collected for subsequent analysis by western blot.

Data availability

All sequencing data have been deposited in Gene Expression Omnibus (Accession number GSE172401). All mass spectrometry data have been deposited to ProteomeExchange (PXD034171) through MassIVE (MSV000089562). All other data are available in the manuscript or in the supplementary materials.

The following data sets were generated
    1. Kwan JZJ
    (2021) NCBI Gene Expression Omnibus
    ID GSE172401. RNA Polymerase II transcription independent of TBP.

References

    1. Kwon H
    2. Green MR
    (1994)
    The RNA polymerase I transcription factor, upstream binding factor, interacts directly with the TATA box-binding protein
    The Journal of Biological Chemistry 269:30140–30146.
    1. Reinberg D
    2. Horikoshi M
    3. Roeder RG
    (1987)
    Factors involved in specific transcription in mammalian RNA polymerase II. functional analysis of initiation factors IIA and IID and identification of a new factor operating at sequences downstream of the initiation site
    The Journal of Biological Chemistry 262:3322–3330.
    1. Sayre MH
    2. Tschochner H
    3. Kornberg RD
    (1992)
    Reconstitution of transcription with five purified initiation factors and RNA polymerase II from Saccharomyces cerevisiae
    The Journal of Biological Chemistry 267:23376–23382.

Decision letter

  1. Irwin Davidson
    Reviewing Editor; Institut de Génétique et de Biologie Moléculaire et Cellulaire, France
  2. Kevin Struhl
    Senior Editor; Harvard Medical School, United States
  3. Irwin Davidson
    Reviewer; Institut de Génétique et de Biologie Moléculaire et Cellulaire, France

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "RNA Polymerase II transcription independent of TBP" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Irwin Davidson as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Kevin Struhl as the Senior Editor.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Your paper has been discussed amongst the referees, the reviewing editor Irwin Davidson and senior editor Kevin Struhl. The general opinion was that this study is of interest to eLife, but requires a thorough revision before publication can be considered. The referee's comments are included below. Several issues require particular attention.

1. The single molecule dynamics experiments are a very indirect way to assess TAF/TFIID association with promoters and since the experiments were performed with ectopically expressed TAFs, the interpretation requires a considerable number of additional controls to show that the proteins are integrated quantitatively into TFIID and not interacting with chromatin individually or in partial TFIID subcomplexes. We suggest that Figure 6 be removed altogether and that the data in Supplementary Figure 9 be moved into the main text. However, there are concerns with the reproducibility of the TAF1 experiments and it is contradictory to claim that a single TBP-free TFIID binds to promoters and at the same time present data showing that TAF1 binding is comparable to the control, whereas that of TAF4 is strongly reduced. All questions in the referee's comments on quantification of replicates require particular attention.

2. Given the important implications of the study, we request that the Cut&Tag data be consolidated by a TBP ChIP-seq with and without auxin treatment to fully ensure that TBP has been depleted from promoters. Similarly, it is essential to assess promoter recruitment of TFIIA and TFIIB upon TBP depletion.

3. In addition, assessment of TFIID assembly in absence of TBP by gel filtration would also strengthen the study.

4. Lastly, the effect of TBP depletion on Pol I transcription should be addressed and the text revised to tone down the interpretation as it is virtually impossible to exclude the idea that residual TBP is sufficient to ensure PIC formation.

We look forward to a revised version where these issues and those in the referee's reports have been addressed.

Reviewer #1 (Recommendations for the authors):

1. Concerning the TBP-independent TFIID complex. The authors show that in absence of TBP other TFIID subunits can coprecipitated with TAF4. The authors should perform gel filtration experiments to determine whether the residual complex is a single TFIID without TBP or whether it has dissociated into diverse TAF4-containing (or without TAF4 that are not taken into account by the MS experiments) subcomplexes of lower molecular mass. None of the experiments in the manuscript address the nature of the TFIID that is recruited to promoters in absence of TBP.

2. A second major issue is the TBP-independent recruitment of TFIIA and TFIIB to promoters. It is believed that both these proteins interact with TBP and that TBP-induced DNA bending aides in their promoter recruitment. How is their occupancy affected in the absence of TBP? Also, TFIIA shows a strong interaction with TRF2, what happens to TFIIA recruitment in the TBP-depleted TRF2 KO cells?

3. The authors argue that TFIID is recruited to promoters in absence of TBP and they show data for TAF1 and TAF4. Can the authors test whether Pol II transcription in general or at specific promoters is more sensitive to loss (siRNA knockdown) of certain TAFs in absence of TBP that in its presence.

Reviewer #2 (Recommendations for the authors):

(lines 30-32) The conformational changes induced by TBP within TFIID trigger

the binding of TFIIA and TFIIB (3), followed by TFIIF and Pol II, which further stabilizes the partial PIC (3). These statements are incorrect and the citation inappropriate.

This is reference (3):Tolić-Nørrelykke, S. F., Rasmussen, M. B., Pavone, F. S., Berg-Sørensen, K., and Oddershede, L. B. (2006) Stepwise Bending of DNA by a Single TATA-Box Binding Protein.830, Biophysical Journal. 90, 3694-3703

The authors may be referring to Roeder, R. G. 1996: The role of general initiation factors in transcription by RNA polymerase II. Trends Biochem. Sci. 21:327-335., which is what the (3) authors are mentioning in their introduction. But even in that case, the first statement is incorrect!

Also reference (4) does not seem at all the most appropriate. Roeder 1996 would be better.

Reference (5) is also totally inappropriate!!

In general, I would ask the authors to double check their citations!!

Concerning TFTC , also known as STAGA, now they are considered to both correspond to the SAGA complex.

(line 267 and beyond) The visualization of overexpressed halo-tagged TAFs does not seem the most appropriate way to look at the dynamics of the TAF-containing complexes. The proper experiment would have been to tag the chromosomal copies.

(lines 277-81) By bound TAF7 molecules, do they mean bound to TFIID or bound to DNA, or both or a mixture? Same for TAF4. If they refer to DNA bound, do they know if they are bound to DNA as part of TFIID or by themselves? With the overexpression, this is a little more difficult to deconvolute and to interpret cleanly.

For the residency time on DNA, how can the authors distinguish if the complex is bound to core promoter DNA or tethered to the +1 nucleosome via reader domains? And, again, is this occurring as individual subunits or in the context of the complex or a mixture? While the two different values for the two subunits make it less likely that both of them are binding as part of a full TFIID, the different possibilities should be kept in mind when interpreting the results.

On that note, TAF4 has activator binding domains, so part of its dynamics can be affected by that, and this has to also be seen in the context of the overexpression

That fewer TAF7 binds to chromatin when TBP is depleted, but those bound stay longer, again needs to be put in the context of the overexpression and of the partial depletion of TBP

If the authors are assuming that TAF7 and TAF4 somehow act differently because they are present in different TFIID complexes, should the immuno-purification via TAF4 show substochiometric amounts of some of the other components? Can they check that?

Can the authors add a statement about how the CUT&Tag experiments and the single particle experiments relate to each other quantitatively?

(316) "based on the crystal structure of the TFIID complex" – should be "cryo-EM structure". On that note, there are two copies of TAF4, and based on structural studies, each of them can be close or far from TBP depending on whether TBP has or not bound the core promoter. By the way, TAF1 is even closer (directly interacting with TBP through its N-terminus) in free TFIID, but further that at least one copy of TAF4, once TBP engages the core promoter.

(327-8) "In contrast to TAF4, the largest subunit maintains its binding to

promoters in the absence of TBP depletion." This does not make sense. Do they mean "in the case of TBP depletion"?

Finally, I am not an expert on the genomic analyses included, so I cannot judge the quality of the data or its interpretation beyond what the authors propose. So, I look forward to discussing the paper with the rest of the reviewers.

Reviewer #3 (Recommendations for the authors):

– The authors should show and quantify the depletion of TBP in all cellular compartments, cytoplasm, nuclear soluble and the chromatin associated fractions after 6 and 16 hours of auxin treatment. In the Result section and in the legend of Figure 1A the authors should indicate which type of extract they are analysing on the western blot.

– As the authors state that after the used auxin treatment about 10-20% TBP is remaining, the numerous statements giving the impression that TBP has no role in Pol II transcription should be tuned down throughout the whole manuscript. What if the remaining 10000-20000 TBP molecules can still work in Pol II transcription because they are very dynamic?

– How many TBP molecules work in Pol I, Pol II and Pol III GTF complexes? If there are many more TFIIIB complexes (containing TBP) than TFIID or SL1 complexes, it is logical that the first effect one will see after TBP depletion is the decrease of Pol III transcription. Can the authors quantify the number of TBP-containing TFIIIB complexes versus TFIID and SL1 complexes?

– Along the same lines, the authors should also test the effect of TBP depletion on Pol I transcription.

– How dynamic is TBP in comparison to Pol I, Pol II and Pol III TAFs and/or Brf1? What are the corresponding DNA residency times of these factors?

– From Figure S2A it is evident that the most highly depleted genes for TBP binding are the most highly transcribed (high Pol II and NET-seq signals). Could TBP be more dynamic there? The same correlation is visible with TAFs (Figure 7D-F). Could TAFs favor TBP dynamics at promoters (making it more "depletable", maybe through TAF1 TAND)? Authors should show, side-by-side, the CUT&Tag occupancy heatmaps for TBP, TAF1, TAF4, Pol II and NET-seq with genes ordered by TBP occupancy (+/- IAA).

– Figure Supp. 5G (not cited in the text) shows that under TBP and TBPL1 (TRF2) co-depletion Pol II occupancy increases about 5 fold (in replicate 2). How do the authors explain this increase? Could this be a normalization artefact? How is it possible that authors conclude from these conflicting replicate experiments that Pol II occupancy is not affected globally (lines 242-243)?

– Experiments described in Figure 6 are lacking several key controls before they can be interpreted. First the authors would need to demonstrate that ALL the "overexpressed" Halo-tagged TAF4 or Halo-TAF7 incorporate into endogenous TFIID complexes. Otherwise the presented Spot-On measurements are meaningless, because the authors may simply measure the behavior of free TAFs. What is the sense of deriving/showing the % of free/complexed/bound states in an overexpressed system (Figure 6)? No comparison can be made among different subunits (TAF7 vs TAF4) as attempted by the authors in the discussion (line 403, "our SPT results indeed show that TAF7 is much less DNA-bound..")

– To normalize their CUT&Tag data, the authors used a spike-free normalization method. This method has only been validated for very abundant histone mark ChIPs, not for Pol II. It is not clear why the authors think that spike-free normalization method is needed for Pol II CUT&Tag normalization. In Supp. Fig1F, the authors show two Pol II CUT&Tag replicates with E. coli spike-in and the replicates are quite different. Why do the authors think that the normalized data showing Pol II decrease in IAA condition is not correct?

– How do authors interpret the widespread CUT&Tag signals they obtain on gene bodies for TAFs (especially TAF1) and partially also for TBP. Do these correspond to non-specific, transient binding events? If yes one would expect that a portion of the signal localized around promoter regions would also be of the same nature.

– Heat-shock & RA induction might not be the best systems to test for the "reinitiation" hypothesis since TBP might already be promoter bound prior to activation. Indeed, there is a TBP peak even in non-treated condition in the tracks shown in Figure 3 and S3.

– The RA treatment experiment has to be repeated similarly to the HS experiment, first 6h auxin treatment to degrade TBP and then RA treatment. At the beginning of the RA treatment TBP is still present in the cells making the interpretation of these experiments impossible.

– In several endogenous TAF IPs TBP has been found to be sub-stoichiometric, could this mean that TBP is not a stable component of mammalian TFIID complexes? TBP dynamics are also regulated by NC2 and BTAF1 complexes (PMID: 20627952), that could influence the interpretation of the obtained result.

– The discussion sentence "The IP-MS data showing that the TFIID complex forms upon TBP depletion is surprising" is misleading, since at structural level TBP depletion is not expected to deeply affect any TFIID subunit or complex integrity.

– "The altered dynamics of TAF4 upon TBP depletion may be due to its close proximity to TBP, based on the crystal structure of the TFIID complex". There is no crystal structure of TFIID. TAF4 is instead involved in the interaction with TFIIA and, quite indirectly, with DNA-bound TBP.

https://doi.org/10.7554/eLife.83810.sa1

Author response

Essential revisions:

1. The single molecule dynamics experiments are a very indirect way to assess TAF/TFIID association with promoters and since the experiments were performed with ectopically expressed TAFs, the interpretation requires a considerable number of additional controls to show that the proteins are integrated quantitatively into TFIID and not interacting with chromatin individually or in partial TFIID subcomplexes. We suggest that Figure 6 be removed altogether and that the data in Supplementary Figure 9 be moved into the main text. However, there are concerns with the reproducibility of the TAF1 experiments and it is contradictory to claim that a single TBP-free TFIID binds to promoters and at the same time present data showing that TAF1 binding is comparable to the control, whereas that of TAF4 is strongly reduced. All questions in the referee's comments on quantification of replicates require particular attention.

We agree that the over-expression Halo-TAF SPT analyses may be too preliminary at this time. We have removed the original Figure 6 in the revised manuscript.

We also agree that the replicates for TAF1 CUT&Tag are not as concordant as other experiments. To address this, we have performed two more replicates of TAF1 CUT&Tag in control and TBP-depleted conditions, for a total of 4 replicates. Plots for each replicate are now shown in new Figure 6—figure supplement 1A-B. All four replicates are consistent with each other, confirming that TAF1 binding is largely unchanged when TBP is depleted.

Based on the TAF4 IP-MS and TAF1 and TAF4 CUT&Tag, we conclude that the complex can form even upon TBP depletion, and that some subunits can bind to promoters after TBP depletion. We wish to clarify that we do not conclude that “a single TBP-free TFIID binds to promoters”. Indeed, in the original manuscript, we speculated in the Discussion section that TFIID may not behave as a single monolithic complex. Rather, it could behave as a dynamic hub such that subunits would come and leave the ‘complex’ with different dynamics. This speculative model of a dynamic complex would be consistent with Halo-TAFs having different residence times as measured by SPT. Also, it would also be consistent with TAF1 and TAF4 having different CUT&Tag profiles after TBP depletion if they have different residence times on DNA.

In the revised version, we have removed references to the TAF dynamics since we have removed the SPT data. Instead, we include references to previous studies showing that TAFs form subcomplexes and may assemble into TFIID in a modular fashion. We also include new gel filtration data showing that TAF4 does not form other complexes when TBP is depleted, and new analysis of the TAF4 IP-MS data showing no change in the relative stoichiometries of the TAF subunits when TBP is depleted (see response to consensus comment 3). These new data and analyses are consistent with our conclusion that the full TFIID complex forms when TBP is depleted. However, biochemical detection of the full complex does not necessarily provide clear information on how the subunits bind to DNA. The behavior of individual subunits would be a combination of the dynamics of its association with the full complex, and the rate of dissociation from the DNA. Techniques such as CUT&Tag would be heavily influenced by how long TFs bind to DNA (residence time), which is inversely proportional to the rate of dissociation. Therefore, we argue that the IP-MS and CUT&Tag data do not provide contradictory information. Rather, they provide information on different aspects of TAF behaviors.

We also note that new studies using live cell imaging are beginning to show that biochemically isolated complexes can have very dynamic behaviors in live cells. These new studies are leading to a revised concept of a ‘complex’ in cellular context, including the CTCF/Cohesin complex in mESCs (Hansen et al. 2019 eLife; Gabriele et al. 2022 Science). With this new framework in mind, it is possible for a biochemical complex to form (TFIID as measured by IP-MS) while also having distinct subunit dynamics that manifests as different binding profiles measured by CUT&Tag.

2. Given the important implications of the study, we request that the Cut&Tag data be consolidated by a TBP ChIP-seq with and without auxin treatment to fully ensure that TBP has been depleted from promoters. Similarly, it is essential to assess promoter recruitment of TFIIA and TFIIB upon TBP depletion.

We agree, and we have performed ChIP-qPCR on select promoters to assess TBP occupancy after depletion. We observed decreases in ChIP levels down to IgG background levels, and this data is now included in the revised Figure 1—figure supplement 1D and is referenced in the main text. Importantly, the ChIP-qPCR results are consistent with the CUT&Tag data, which shows over 90% depletion at promoter regions genome-wide. Taken together, we have provided five orthogonal experiments to validate the TBP depletion system – Western, IF, ChIP-qPCR, CUT&Tag, and IP-MS.

We also agree that assessing TFIIA and TFIIB promoter occupancy after TBP depletion would be informative. Towards addressing this comment, we have performed spike-in normalized CUT&Tag using anti-TFIIA and TFIIB. The TFIIA CUT&Tag was successful, but the TFIIB antibodies did not produce signal above IgG background (we tried both Santa Cruz Biotechnology α-TFIIB (C-18) sc-225 and Cell Signaling Technology α-TFIIB 2F6A3H4). We aim to examine TFIIB through other means and other PIC members in the future. For now, we have included the results for TFIIA CUT&Tag in control and TBP-depleted conditions in revised Figure 6, with replicate information now shown in new Figure 6—figure supplement 2. A summary of the findings are now included in the main manuscript. Briefly, we found that TFIIA binding to promoters is decreased by ~50-60% relative to control conditions when TBP is depleted. We also performed a reanalysis of the TAF4 IP-MS data, which shows that the interaction between TFIIA and TAF4 is decreased when TBP is depleted (Figure 6—figure supplement 2A), even though we observed no change in TFIIA protein levels (Figure 6—figure supplement 2A). Consistent with the lack of effect on Pol II occupancy and activity, however, the genes that experience loss of TFIIA binding show no change in Pol II binding (Figure 6—figure supplement 3A,C). Therefore, we conclude that TBP depletion affects distinct members of the PIC, but that the system remains sufficiently robust to allow for Pol II transcription.

3. In addition, assessment of TFIID assembly in absence of TBP by gel filtration would also strengthen the study.

We agree that determining the nature of TFIID without TBP would be informative. As we discussed in our response to consensus comment #1, we have performed a gel filtration experiment on nuclear lysates in control and TBP depleted cells. We detected no changes in TAF4 elution patterns between control and TBP depleted conditions, and this data is now presented in revised Figure 5—figure supplement 1. We chose to analyze the nuclear fraction to specifically focus on the complex that could form on promoters, which means that we are blind to the existence of subcomplexes found in the cytoplasm. However, by analyzing the IP-MS data, we observe that the relative stoichiometries of TAF subunits do not change when TBP is depleted. Therefore, regardless of the nature of any subcomplexes forming in the cytoplasm, we show that their interactions do not change when TBP is depleted. We have now included this analysis in Figure 5—figure supplement 1.

4. Lastly, the effect of TBP depletion on Pol I transcription should be addressed and the text revised to tone down the interpretation as it is virtually impossible to exclude the idea that residual TBP is sufficient to ensure PIC formation.

We look forward to a revised version where these issues and those in the referee's reports have been addressed.

To determine the effects of TBP depletion on Pol I transcription, we examined two orthogonal data sets: (1) NET-seq data; and (2) previously published chromatin-associated RNA-seq on control and TBP-depleted mESCs (PMID: 29939130). For both data sets, we quantified the signal for all rRNA genes in the mm10 genome in control and TBP-depleted conditions, and plotted the data as a scatter plot (see Author response image 1). In both data sets, the data for all rRNA genes fall mostly on the diagonal. These results suggest that, like Pol II, Pol I transcription may not be affected by acute TBP depletion. However, a more stringent examination is needed, much like we have done for Pol II, but this is beyond the scope of this manuscript and we look forward to pursuing this new direction in future studies.

Author response image 1

We agree with the reviewers that the interpretation of our results need to be restricted. Throughout the manuscript, we have revised the text to adhere strictly to the conclusion that acute TBP depletion does not affect Pol II transcription. Furthermore, in the Discussion section, we have provided a thorough discussion on the inherent limitations of the mAID system and what it could mean for our study. We also added that although TBP depletion does not affect Pol II transcription, our system does not provide information on what TBP is doing when bound to promoters. Specifically, we state: “However, these results do not provide information on the functional role of TBP when recruited to Pol II promoters. One possibility is that Pol II transcription has evolved to have multiple redundancies such that removal of one factor (TBP) may not fully disrupt the whole system. Another possibility is that the transient nature of the embryonic cell state, with its unique chromatin landscape, enables Pol II transcription without TBP.” (Discussion section).

Reviewer #1 (Recommendations for the authors):

1. Concerning the TBP-independent TFIID complex. The authors show that in absence of TBP other TFIID subunits can coprecipitated with TAF4. The authors should perform gel filtration experiments to determine whether the residual complex is a single TFIID without TBP or whether it has dissociated into diverse TAF4-containing (or without TAF4 that are not taken into account by the MS experiments) subcomplexes of lower molecular mass. None of the experiments in the manuscript address the nature of the TFIID that is recruited to promoters in absence of TBP.

Please see our response to consensus comment #1 and 3 above.

2. A second major issue is the TBP-independent recruitment of TFIIA and TFIIB to promoters. It is believed that both these proteins interact with TBP and that TBP-induced DNA bending aides in their promoter recruitment. How is their occupancy affected in the absence of TBP? Also, TFIIA shows a strong interaction with TRF2, what happens to TFIIA recruitment in the TBP-depleted TRF2 KO cells?

As we discussed above, we have performed TFIIA and TFIIB CUT&Tag in control and TBP-depleted mESCs. Unfortunately, the TFIIB antibody was not successful in the assay, but the TFIIA CUT&Tag shows strong signal at the TSS of all active genes (revised Figure 6) in control conditions. TBP-depletion led to a decrease in TFIIA binding by 50-60%, as measured by the genome-wide average profile. Consistent with the TAF4 results, the decrease in TFIIA binding when TBP is depleted suggests that: (1) Pol II is robust enough that a decrease in binding of some of the GTFs does not disrupt its activity; and (2) although Pol II transcription is unaffected, the PIC may undergo major changes in dynamics and/or DNA-binding kinetics. These findings are now summarized in the revised manuscript. Furthermore, a reanalysis of the TAF4 IP-MS data shows that the interaction between TAF4(TFIID) with TFIIA is significantly reduced when TBP is depleted. This data is now presented in the new Figure 6—figure supplement 2A.

Since TRF2 is not functionally redundant to TBP, we believe that what happens to TFIIA when TBP is depleted in the TRF2 KO cells is outside the scope of this study. However, we are excited to pursue other TRF2-specific mechanisms in future studies.

3. The authors argue that TFIID is recruited to promoters in absence of TBP and they show data for TAF1 and TAF4. Can the authors test whether Pol II transcription in general or at specific promoters is more sensitive to loss (siRNA knockdown) of certain TAFs in absence of TBP that in its presence.

We would love to test combinatorial knockdowns of TAFs with TBP. However, as there are 13 subunits, the scale is currently beyond what we can accomplish in a revision for this manuscript. However, we plan to systematically test all 13 subunits in future studies.

Reviewer #2 (Recommendations for the authors):

(lines 30-32) The conformational changes induced by TBP within TFIID trigger

the binding of TFIIA and TFIIB (3), followed by TFIIF and Pol II, which further stabilizes the partial PIC (3). These statements are incorrect and the citation inappropriate.

This is reference (3):Tolić-Nørrelykke, S. F., Rasmussen, M. B., Pavone, F. S., Berg-Sørensen, K., and Oddershede, L. B. (2006) Stepwise Bending of DNA by a Single TATA-Box Binding Protein.830, Biophysical Journal. 90, 3694-3703

The authors may be referring to Roeder, R. G. 1996: The role of general initiation factors in transcription by RNA polymerase II. Trends Biochem. Sci. 21:327-335., which is what the (3) authors are mentioning in their introduction. But even in that case, the first statement is incorrect!

Also reference (4) does not seem at all the most appropriate. Roeder 1996 would be better.

Reference (5) is also totally inappropriate!!

In general, I would ask the authors to double check their citations!!

We agree with the reviewer and apologize for the oversight. We have checked and revised the references throughout the manuscript.

Concerning TFTC , also known as STAGA, now they are considered to both correspond to the SAGA complex.

We agree and we have modified the text to reflect this change: “TFTC was subsequently shown to correspond to the SAGA complex…” (pg. 3).

(line 267 and beyond) The visualization of overexpressed halo-tagged TAFs does not seem the most appropriate way to look at the dynamics of the TAF-containing complexes. The proper experiment would have been to tag the chromosomal copies.

We agree that our over-expression system is limited in capacity, and that endogenous tagging is the more rigorous approach. Since our imaging experiments are somewhat preliminary, we have removed these experiments from the manuscript. Because of this, comments 2-8 are no longer applicable. We hope to use these comments to better design imaging experiments for future studies.

(lines 277-81) By bound TAF7 molecules, do they mean bound to TFIID or bound to DNA, or both or a mixture? Same for TAF4. If they refer to DNA bound, do they know if they are bound to DNA as part of TFIID or by themselves? With the overexpression, this is a little more difficult to deconvolute and to interpret cleanly.

This comment is no longer applicable.

For the residency time on DNA, how can the authors distinguish if the complex is bound to core promoter DNA or tethered to the +1 nucleosome via reader domains? And, again, is this occurring as individual subunits or in the context of the complex or a mixture? While the two different values for the two subunits make it less likely that both of them are binding as part of a full TFIID, the different possibilities should be kept in mind when interpreting the results.

On that note, TAF4 has activator binding domains, so part of its dynamics can be affected by that, and this has to also be seen in the context of the overexpression

This comment is no longer applicable.

That fewer TAF7 binds to chromatin when TBP is depleted, but those bound stay longer, again needs to be put in the context of the overexpression and of the partial depletion of TBP

If the authors are assuming that TAF7 and TAF4 somehow act differently because they are present in different TFIID complexes, should the immuno-purification via TAF4 show substochiometric amounts of some of the other components? Can they check that?

This comment is no longer applicable.

Can the authors add a statement about how the CUT&Tag experiments and the single particle experiments relate to each other quantitatively?

This comment is no longer applicable.

(316) "based on the crystal structure of the TFIID complex" – should be "cryo-EM structure". On that note, there are two copies of TAF4, and based on structural studies, each of them can be close or far from TBP depending on whether TBP has or not bound the core promoter. By the way, TAF1 is even closer (directly interacting with TBP through its N-terminus) in free TFIID, but further that at least one copy of TAF4, once TBP engages the core promoter.

We agree and we have revised the Discussion section regarding TFIID structure. This new paragraph is now in page 14.

(327-8) "In contrast to TAF4, the largest subunit maintains its binding to

promoters in the absence of TBP depletion." This does not make sense. Do they mean "in the case of TBP depletion"?

We agree and have modified the text to reflect this change “Therefore, TAF1 binding is unaffected by TBP-depletion" (pg. 11)

Reviewer #3 (Recommendations for the authors):

– The authors should show and quantify the depletion of TBP in all cellular compartments, cytoplasm, nuclear soluble and the chromatin associated fractions after 6 and 16 hours of auxin treatment. In the Result section and in the legend of Figure 1A the authors should indicate which type of extract they are analysing on the western blot.

We agree and have performed cell fractionation experiments. We assessed mAID-TBP levels +/- IAA through Western blot analysis for each fraction. We show that TBP depletion is evident in both the cytoplasmic and chromatin bound fractions, which is consistent with the CUT&Tag data that primarily detects chromatin bound proteins. We have now included this cell fractionation experiment as a new panel in Figure1—figure supplement 1, and have referenced it in the main Results section. We have also clarified throughout the manuscript when the Western blots are performed on whole cell lysates.

– As the authors state that after the used auxin treatment about 10-20% TBP is remaining, the numerous statements giving the impression that TBP has no role in Pol II transcription should be tuned down throughout the whole manuscript. What if the remaining 10000-20000 TBP molecules can still work in Pol II transcription because they are very dynamic?

Previous biochemical studies, especially in vitro transcription experiments, have shown numerous evidence that TBP is necessary and sufficient for Pol II transcription. We agree that our system can only test if TBP is required/necessary for Pol II transcription in mESCs. We have revised the manuscript to specifically adhere to the conclusion that acute TBP depletion does not affect Pol II transcription, which we believe is justified by several orthogonal lines of evidence in our study. We agree that we cannot say that TBP has no role/function in Pol II transcription. Towards clarifying this in our manuscript, we also add in our Discussion section the following: “However, these results do not provide information on the functional role of TBP when recruited to Pol II promoters. One possibility is that Pol II transcription has evolved to have multiple redundancies such that removal of one factor (TBP) may not fully disrupt the whole system. Another possibility is that the transient nature of the embryonic cell state, with its unique chromatin landscape, enables Pol II transcription without TBP. Regardless, at the level of TBP depletion that we observe, we conclude that in mESCs Pol II transcription can occur independently of TBP” (pg 14-15).

Although it is possible, we currently have no evidence to suggest that the decreased levels of TBP after depletion would alter TBP-DNA binding dynamics. SPT analysis for endogenously tagged Halo-TBP has been published (PMID: 29939130) shows that TBP is one of the more stable-binding TFs imaged thus far, with average residence times on the order of minutes (~60-90 seconds). In contrast, most other TFs have residence times of seconds or less. Indeed, as this reviewer noted below, TBP dynamics is regulated by NC2 and BTAF1 complexes (Mot1 in yeast). These enzymes have been shown to be required to remove TBP from DNA, which is consistent with the long residence times of TBP. It is unclear how these enzymes would act under severely decreased TBP molecules. However, the Martianov 2002 paper (PMID: 12411709) has shown that TBP knockout embryos were still capable of generating newly transcribed RNA. Our molecular/genomics analyses using TBP depletion are consistent with this previous study, which we cited in our manuscript, and significantly advance the field by considering three potential mechanisms for TBP-independent Pol II transcription.

– How many TBP molecules work in Pol I, Pol II and Pol III GTF complexes? If there are many more TFIIIB complexes (containing TBP) than TFIID or SL1 complexes, it is logical that the first effect one will see after TBP depletion is the decrease of Pol III transcription. Can the authors quantify the number of TBP-containing TFIIIB complexes versus TFIID and SL1 complexes?

Absolute quantification of the number of molecules for TBP (and CTCF in the cited research article) was performed using rigorous Fluorescence Correlation Spectroscopy of endogenously tagged factors. To our knowledge, no similar study has been performed for Pol I, Pol II, or Pol III GTFs. Furthermore, this FCS approach quantifies molecules and cannot distinguish between different complex-bound molecules. Unfortunately, with the current limitations of the technology, we are unable to address this comment.

– Along the same lines, the authors should also test the effect of TBP depletion on Pol I transcription.

Please see our response to consensus comment #4 above.

– How dynamic is TBP in comparison to Pol I, Pol II and Pol III TAFs and/or Brf1? What are the corresponding DNA residency times of these factors?

As mentioned above, SPT analysis for endogenously tagged Halo-TBP has been published (PMID: 29939130), revealing long residence times compared to most other TFs. TAF4 and TAF7 have much shorter residence times compared to TBP, as measured by over-expression constructs. For a more rigorous approach, we would need to endogenously tag all factors the reviewer requested and perform the SPT analysis. This is currently not feasible in the time frame, but we are excited to pursue this line of research in the future.

– From Figure S2A it is evident that the most highly depleted genes for TBP binding are the most highly transcribed (high Pol II and NET-seq signals). Could TBP be more dynamic there? The same correlation is visible with TAFs (Figure 7D-F). Could TAFs favor TBP dynamics at promoters (making it more "depletable", maybe through TAF1 TAND)? Authors should show, side-by-side, the CUT&Tag occupancy heatmaps for TBP, TAF1, TAF4, Pol II and NET-seq with genes ordered by TBP occupancy (+/- IAA).

Genomics-based analyses such as our CUT&Tag and NET-seq data do not provide information on TBP dynamics. The higher genomics signal is only indicative of binding at the population level, which is correlated to the level of transcription. This observation is generally believed to be one of two reasons: (1) genes that are highly expressed have these factors bound for more time instances compared to less transcribed genes (bursting mechanism); or (2) at a population level, more cells are transcribing these genes while other cells are not (cell-to-cell variability). Therefore, it’s expected that highly expressed genes would have high Pol II signal, and high GTFs (TBP and TAFs) signal because their occupancy are highly correlated to transcription level.

It is also not surprising that the largest TBP depletion is most evident on promoters with high TBP occupancy, but the relative level of depletion is the same for promoters with high vs low occupancy, about 80-90% depletion. If highly expressed genes are “more depletable”, then the scatter plot in Figure 1E (TBP Ctl vs IAA) would look more exponential than a linear correlation. Instead, the r2 value for the linear regression analysis is very high, which means that highly expressed genes are experiencing similar levels of TBP depletion as middle/lowly expressed genes.

Furthermore, as we discussed in previous comments, the previously published residence time of TBP is a description of the average molecule and is independent of locus-specificity. That is, this data lacks information on whether TBP binding dynamics would change depending on which promoter it binds to. It is possible that highly expressed genes would also have more binding of NC2/BTAF1 but this data would still be correlational and would not provide direct evidence of differential TBP binding dynamics.

In the revised Figure 6—figure supplement 3D, we now include the side-by-side heatmaps for TBP, TAF1, TAF4, TFIIA, Pol II, and NET-seq as ordered by TBP occupancy.

– Figure Supp. 5G (not cited in the text) shows that under TBP and TBPL1 (TRF2) co-depletion Pol II occupancy increases about 5 fold (in replicate 2). How do the authors explain this increase? Could this be a normalization artefact? How is it possible that authors conclude from these conflicting replicate experiments that Pol II occupancy is not affected globally (lines 242-243)?

We have performed two additional replicates of Pol II CUT&Tag in TBP-depletion + TRF2 KO cells (new Figure 4—figure supplement 1I-K). For three of the four replicates, we find that Pol II signal is very similar between the TRF2 knockout and the double TBP-depletion + TRF2 KO. Therefore, we believe that the increase seen in replicate 2 (Figure 4—figure supplement 1J) is likely an artifact of normalization. We have included a citation to (now new) Figure 4—figure supplement 1J in the manuscript.

– Experiments described in Figure 6 are lacking several key controls before they can be interpreted. First the authors would need to demonstrate that ALL the "overexpressed" Halo-tagged TAF4 or Halo-TAF7 incorporate into endogenous TFIID complexes. Otherwise the presented Spot-On measurements are meaningless, because the authors may simply measure the behavior of free TAFs. What is the sense of deriving/showing the % of free/complexed/bound states in an overexpressed system (Figure 6)? No comparison can be made among different subunits (TAF7 vs TAF4) as attempted by the authors in the discussion (line 403, "our SPT results indeed show that TAF7 is much less DNA-bound..")

As discussed above with reviewer 2’s comments, we agree that our over-expression system is limited in capacity, and that endogenous tagging is the more rigorous approach. Since our imaging experiments are somewhat preliminary, we have removed these experiments from the manuscript.

– To normalize their CUT&Tag data, the authors used a spike-free normalization method. This method has only been validated for very abundant histone mark ChIPs, not for Pol II. It is not clear why the authors think that spike-free normalization method is needed for Pol II CUT&Tag normalization. In Supp. Fig1F, the authors show two Pol II CUT&Tag replicates with E. coli spike-in and the replicates are quite different. Why do the authors think that the normalized data showing Pol II decrease in IAA condition is not correct?

We agree that the E. coli spike-in normalized replicates are showing increased variability. We had aimed to use the ChIPseqSpikeInFree normalization method as an orthogonal system for validation. However, to gain more confidence in the results, we have performed two additional replicates of the Pol II CUT&Tag +/- IAA using S2 (Drosophila) cells as spike in control. We now include the heatmaps for these replicates normalized to S2 reads in the new Figure 1—figure supplement 2A and are referenced in the main text. Importantly, the S2-normalized data is consistent with the ChIPseqSpikeInFree normalization, and with our conclusions that Pol II transcription is largely unaffected by acute TBP depletion.

– How do authors interpret the widespread CUT&Tag signals they obtain on gene bodies for TAFs (especially TAF1) and partially also for TBP. Do these correspond to non-specific, transient binding events? If yes one would expect that a portion of the signal localized around promoter regions would also be of the same nature.

CUT&Tag uses pA-Tn5, which has an intrinsic preference for accessible DNA when integrating the adapters. Although the CUT&Tag protocol is designed to minimize this intrinsic preference, some of the background signals on our IgG CUT&Tag controls are located on accessible DNA (See new Figure 1—figure supplement 1F). The highest expressed genes have been shown to have lower nucleosome occupancy than mid-level expressed genes (PMID: 22085965 and others), and heat shock genes have been shown to be massively depleted of nucleosomes upon activation (PMID: 18614012). We speculate that the CUT&Tag signal that we observe within gene bodies could be a product of the pA-Tn5 integration spreading away from the TSS when the surrounding nucleosomes are disrupted.

Non-specific transient binding events could contribute to the CUT&Tag signal, but presumably, other genomics-based methods like ChIP-seq would also have a percentage of the signal corresponding to non-specific binding. Our IgG CUT&Tag controls would represent a measure of such non-specific/background binding levels. We have now included the heatmaps for the IgG controls in the new Figure 1—figure supplement 1F. With this control, it is evident how strong the specific binding is for our GTFs to the TSS.

– Heat-shock & RA induction might not be the best systems to test for the "reinitiation" hypothesis since TBP might already be promoter bound prior to activation. Indeed, there is a TBP peak even in non-treated condition in the tracks shown in Figure 3 and S3.

We agree that the HS response may not be ideal to test for reinitiation because it is well known to be regulated at the paused Pol II stage. We have added this statement in the text to reflect this change:

“The heat shock response is a classical gene induction system regulated at the level of paused Pol II elongation (ref), and HS genes may retain a scaffold of PIC components following promoter pausing release. Therefore, we also examined the role of TBP in an orthogonal gene activation system through retinoic acid (RA)-mediated differentiation of C64 mESCs, which leads to the silencing of pluripotency markers and activation of ectoderm-specific genes” (pg 8).

It is unclear in the literature whether the RA-induced genes are regulated through Pol II pausing, but at least some of the genes only have background level TBP CUT&Tag signal at the promoter region, including Cdx1 (Figure 3—figure supplement 1D). We have also examined the change in Pol II CUT&Tag levels relative to TBP CUT&Tag levels in control conditions (see scatter plots in Author response image 2).

Author response image 2

For each gene, we calculated the total number of normalized Pol II or TBP reads +/- 1 Kb of the TSS. The left scatter plot shows all genes, and the right one shows all genes with increased Pol II CUT&Tag during RA and with TBP levels under 200 read counts. We highlighted several genes in red whose expression increased upon RA treatment but had very low TBP levels under control conditions.

– The RA treatment experiment has to be repeated similarly to the HS experiment, first 6h auxin treatment to degrade TBP and then RA treatment. At the beginning of the RA treatment TBP is still present in the cells making the interpretation of these experiments impossible.

We had originally planned to do this treatment (6 hours of IAA, then overnight IAA+RA). However, we found that 16-18 hours of IAA treatment is as much as we can do without significant cell death. At 24 hours of TBP depletion, we observe over 50% cell death. So we are limited with our IAA treatment timing to avoid secondary effects due to cell death. Unfortunately, RA treatment does not result in gene expression changes prior to overnight (16-18 hours) of treatment. Most published studies perform 24-48 hours before assessing changes in gene expression. We had done a preliminary test by performing RT-qPCR on two early RA-induced genes and observed no changes after 6 hours of RA treatment, but induction occurred after 16 hours. We believe we are at the cusp of the initial RA effects at 16 hours, and likely this is why we observe only about ~100 differentially expressed genes. Although IAA treatment started concurrently with RA treatment, we believe that the TBP degradation is near complete before the early RA genes are induced. We acknowledge that our compromise of simultaneous IAA+RA 16 hr treatment is not ideal, but we are also limited by the restrictions of the system.

– In several endogenous TAF IPs TBP has been found to be sub-stoichiometric, could this mean that TBP is not a stable component of mammalian TFIID complexes? TBP dynamics are also regulated by NC2 and BTAF1 complexes (PMID: 20627952), that could influence the interpretation of the obtained result.

In all previously published biochemical purification of TFIID that we are aware of, TBP is a stable component of the complex in human cells (for example PMID: 9882510) and in yeast cells (for example PMID: 2662184). In the study that the reviewer cites (“Chromatin interaction of TATA-binding protein is dynamically regulated in human cells” PMID: 20627952), the authors show that NC2 and BTAF1 regulate the DNA/chromatin binding dynamics of TBP, but not its complex formation with TFIID.

Regarding the reviewer’s comment “… that could influence the interpretation of the obtained result”, unfortunately we are unclear which interpretation of which result the reviewer is referring to. Is the reviewer referring to the IP-MS data (Figure 5), and our interpretation is that TFIID forms even when TBP is depleted? If so, we do not believe that our interpretation would be affected by the DNA/chromatin binding dynamics of TBP as regulated by NC2/BTAF1. Given that previous literature showing TBP is a stable component of TFIID through biochemical purification, and more recently that TBP forms stable ternary complex with TAF11 and TAF13 (PMID: 29111974), our initial prediction was that TFIID would not form when TBP is depleted, but the IP-MS data shows otherwise.

– The discussion sentence "The IP-MS data showing that the TFIID complex forms upon TBP depletion is surprising" is misleading, since at structural level TBP depletion is not expected to deeply affect any TFIID subunit or complex integrity.

Based on the reanalyses and new experiments recommended by all three reviewers, we have revised our Discussion section regarding TFIID. We believe that the discussion is now stronger as a result of the revisions.

– "The altered dynamics of TAF4 upon TBP depletion may be due to its close proximity to TBP, based on the crystal structure of the TFIID complex". There is no crystal structure of TFIID. TAF4 is instead involved in the interaction with TFIIA and, quite indirectly, with DNA-bound TBP.

We agree and have corrected this oversight in the revised manuscript.

https://doi.org/10.7554/eLife.83810.sa2

Article and author information

Author details

  1. James ZJ Kwan

    Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Vancouver, Canada
    Contribution
    Conceptualization, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    Thomas F Nguyen
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4467-1749
  2. Thomas F Nguyen

    Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Vancouver, Canada
    Contribution
    Conceptualization, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Contributed equally with
    James ZJ Kwan
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-3269-0846
  3. Anuli C Uzozie

    1. Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, Canada
    2. Michael Cuccione Childhood Cancer Research Program, BC Children’s Hospital Research Institute, Vancouver, Canada
    Contribution
    Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
  4. Marek A Budzynski

    Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Vancouver, Canada
    Contribution
    Funding acquisition, Investigation, Visualization, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
  5. Jieying Cui

    Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Vancouver, Canada
    Contribution
    Investigation, Visualization, Writing - review and editing
    Competing interests
    No competing interests declared
  6. Joseph MC Lee

    Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Vancouver, Canada
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  7. Filip Van Petegem

    Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Vancouver, Canada
    Contribution
    Methodology
    Competing interests
    No competing interests declared
  8. Philipp F Lange

    1. Department of Pathology and Laboratory Medicine, University of British Columbia, Vancouver, Canada
    2. Michael Cuccione Childhood Cancer Research Program, BC Children’s Hospital Research Institute, Vancouver, Canada
    Contribution
    Funding acquisition, Investigation, Visualization, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1171-5864
  9. Sheila S Teves

    Department of Biochemistry and Molecular Biology, Life Sciences Institute, University of British Columbia, Vancouver, Canada
    Contribution
    Conceptualization, Supervision, Funding acquisition, Visualization, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    sheila.teves@ubc.ca
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1220-2414

Funding

Canadian Institutes of Health Research (PJT-162289)

  • Sheila S Teves

Natural Sciences and Engineering Research Council of Canada (RGPIN-2020-06106)

  • Sheila S Teves

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Dr. Steven Henikoff for generously providing the pA-Tn5 enzyme, and Dr. László Tora for generously providing the anti-TRF2 antibody. We thank Reid Warsaba, Jibin Sadasivan, and Dr. Eric Jan for providing S2 cells and for providing insight on qRT-PCR S2 spike-in normalization. We thank R Vander Werff and T Stach (BRC-seq, UBC) for Illumina sequencing and S Flibotte (LSI Bioinformatics facility, UBC) for implementation of NET-seq analyses. This work was supported by Life Sciences Institute Cores (LSI Imaging, ubcFLOW, and QPCR Core), supported by the UBC GREx Biological Resilience Initiative. For insightful comments on the manuscript, we thank Drs. Annie Ciernia and Ethan Greenblatt. MAB is a postdoctoral fellow supported by the Sigrid Jusélius Foundation. ACU is supported by the BC Children’s Hospital Foundation. PFL is a Canada Research Chair and Michael Smith Foundation for Health Research Scholar. SST is a Canada Research Chair and Michael Smith Foundation for Health Research Scholar. Funding: This work was supported by the Canadian Institutes for Health Research Project Grant award to SST (PJT-162289). The National Sciences and Engineering Research Council Discovery Grant award to SST (RGPIN-2020-06106).

Senior Editor

  1. Kevin Struhl, Harvard Medical School, United States

Reviewing Editor

  1. Irwin Davidson, Institut de Génétique et de Biologie Moléculaire et Cellulaire, France

Reviewer

  1. Irwin Davidson, Institut de Génétique et de Biologie Moléculaire et Cellulaire, France

Version history

  1. Preprint posted: March 29, 2021 (view preprint)
  2. Received: September 29, 2022
  3. Accepted: March 29, 2023
  4. Accepted Manuscript published: March 30, 2023 (version 1)
  5. Version of Record published: May 11, 2023 (version 2)

Copyright

© 2023, Kwan, Nguyen et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,069
    Page views
  • 355
    Downloads
  • 3
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. James ZJ Kwan
  2. Thomas F Nguyen
  3. Anuli C Uzozie
  4. Marek A Budzynski
  5. Jieying Cui
  6. Joseph MC Lee
  7. Filip Van Petegem
  8. Philipp F Lange
  9. Sheila S Teves
(2023)
RNA Polymerase II transcription independent of TBP in murine embryonic stem cells
eLife 12:e83810.
https://doi.org/10.7554/eLife.83810

Share this article

https://doi.org/10.7554/eLife.83810

Further reading

    1. Chromosomes and Gene Expression
    2. Structural Biology and Molecular Biophysics
    Matthew R Marunde, Harrison A Fuchs ... Catherine A Musselman
    Research Article Updated

    Histone post-translational modifications (PTMs) play a critical role in chromatin regulation. It has been proposed that these PTMs form localized ‘codes’ that are read by specialized regions (reader domains) in chromatin-associated proteins (CAPs) to regulate downstream function. Substantial effort has been made to define [CAP: histone PTM] specificities, and thus decipher the histone code and guide epigenetic therapies. However, this has largely been done using the reductive approach of isolated reader domains and histone peptides, which cannot account for any higher-order factors. Here, we show that the [BPTF PHD finger and bromodomain: histone PTM] interaction is dependent on nucleosome context. The tandem reader selectively associates with nucleosomal H3K4me3 and H3K14ac or H3K18ac, a combinatorial engagement that despite being in cis is not predicted by peptides. This in vitro specificity of the BPTF tandem reader for PTM-defined nucleosomes is recapitulated in a cellular context. We propose that regulatable histone tail accessibility and its impact on the binding potential of reader domains necessitates we refine the ‘histone code’ concept and interrogate it at the nucleosome level.

    1. Chromosomes and Gene Expression
    2. Evolutionary Biology
    Katherine Rickelton, Trisha M Zintel ... Courtney C Babbitt
    Research Article Updated

    Primate evolution has led to a remarkable diversity of behavioral specializations and pronounced brain size variation among species (Barton, 2012; DeCasien and Higham, 2019; Powell et al., 2017). Gene expression provides a promising opportunity for studying the molecular basis of brain evolution, but it has been explored in very few primate species to date (e.g. Khaitovich et al., 2005; Khrameeva et al., 2020; Ma et al., 2022; Somel et al., 2009). To understand the landscape of gene expression evolution across the primate lineage, we generated and analyzed RNA-seq data from four brain regions in an unprecedented eighteen species. Here, we show a remarkable level of variation in gene expression among hominid species, including humans and chimpanzees, despite their relatively recent divergence time from other primates. We found that individual genes display a wide range of expression dynamics across evolutionary time reflective of the diverse selection pressures acting on genes within primate brain tissue. Using our samples that represent a 190-fold difference in primate brain size, we identified genes with variation in expression most correlated with brain size. Our study extensively broadens the phylogenetic context of what is known about the molecular evolution of the brain across primates and identifies novel candidate genes for the study of genetic regulation of brain evolution.