1. Chromosomes and Gene Expression
  2. Genetics and Genomics
Download icon

Role of the pre-initiation complex in Mediator recruitment and dynamics

  1. Elisabeth R Knoll
  2. Z Iris Zhu
  3. Debasish Sarkar
  4. David Landsman
  5. Randall H Morse  Is a corresponding author
  1. University at Albany, United States
  2. National Center for Biotechnology Information, National Library of Medicine, United States
  3. New York State Department of Health, United States
Research Article
  • Cited 0
  • Views 986
  • Annotations
Cite this article as: eLife 2018;7:e39633 doi: 10.7554/eLife.39633

Abstract

The Mediator complex stimulates the cooperative assembly of a pre-initiation complex (PIC) and recruitment of RNA Polymerase II (Pol II) for gene activation. The core Mediator complex is organized into head, middle, and tail modules, and in budding yeast (Saccharomyces cerevisiae), Mediator recruitment has generally been ascribed to sequence-specific activators engaging the tail module triad of Med2-Med3-Med15 at upstream activating sequences (UASs). We show that yeast lacking Med2-Med3-Med15 are viable and that Mediator and PolII are recruited to promoters genome-wide in these cells, albeit at reduced levels. To test whether Mediator might alternatively be recruited via interactions with the PIC, we examined Mediator association genome-wide after depleting PIC components. We found that depletion of Taf1, Rpb3, and TBP profoundly affected Mediator association at active gene promoters, with TBP being critical for transit of Mediator from UAS to promoter, while Pol II and Taf1 stabilize Mediator association at proximal promoters.

https://doi.org/10.7554/eLife.39633.001

Introduction

The Mediator complex plays a central, highly conserved role in eukaryotic transcription by stimulating the cooperative assembly of a pre-initiation complex (PIC) and recruitment of RNA Polymerase II (Pol II) for gene activation (Allen and Taatjes, 2015; Jeronimo and Robert, 2017; Kornberg, 2005; Malik and Roeder, 2010; Soutourina, 2018). In budding yeast (Saccharomyces cerevisiae), Mediator is composed of 25 subunits divided into four domains based on structural and functional criteria: the tail, middle, head, and kinase modules (Guglielmi et al., 2004; Plaschka et al., 2016; van de Peppel et al., 2005). Mediator recruitment has generally been ascribed to sequence-specific activators engaging the tail module triad of Med2-Med3-Med15 (Lee et al., 1999; Myers et al., 1999; Zhang et al., 2004; Ansari and Morse, 2013); consistent with this model, the tail module can under some circumstances be recruited independently of the remainder of Mediator (Ansari et al., 2009; He et al., 2008; Paul et al., 2015; Petrenko et al., 2016; Zhang et al., 2004; Anandhakumar et al., 2016). Normally, however, the tail is recruited as a complex with the head, middle, and kinase modules at upstream activating sequences (UASs) (Jeronimo and Robert, 2014; Wong et al., 2014; Grünberg et al., 2016; Kuras et al., 2003). Once Mediator is recruited, the kinase module dissociates from the complex and Mediator bridges the enhancer to connect with Pol II and general transcription factors (GTFs) at the proximal promoter of active genes (Jeronimo et al., 2016; Petrenko et al., 2016). Mediator association with the PIC at proximal promoters is quickly disrupted following phosphorylation of the Pol II C-terminal domain (CTD) by Kin28, marking the end of initiation (Jeronimo and Robert, 2014; Wong et al., 2014).

This orchestrated recruitment of Mediator is broadly required for transcription, evidenced by studies that showed a loss of Mediator function results in greatly decreased mRNA levels and concomitant diminished association of Pol II with essentially all Pol II transcribed genes (Ansari et al., 2009; Holstege et al., 1998; Jeronimo and Robert, 2014; Paul et al., 2015; Plaschka et al., 2015; Thompson and Young, 1995). In spite of the purported role of the tail module triad in Mediator recruitment by transcriptional activators, med3Δ med15Δ yeast are viable, show defects in mRNA levels for only 5–10% of active genes during normal growth, and exhibit continued Mediator and Pol II occupancy by ChIP-seq (Ansari et al., 2012a; Jeronimo et al., 2016; Paul et al., 2015; Petrenko et al., 2016). Additional studies report that the tail module is not required for Mediator’s interaction with the PIC (Jeronimo et al., 2016; Petrenko et al., 2016; Plaschka et al., 2015). That transcription is unaffected at many genes in med3Δ med15Δ yeast supports the notion that Mediator is recruited through an unknown mechanism that is independent of the tail module, possibly through components of the PIC which are known to engage Mediator subunits at proximal promoters (Ansari and Morse, 2013; Esnault et al., 2008; Soutourina et al., 2011; Eychenne et al., 2016; Larivière et al., 2006). Indirect recruitment of Mediator via interactions with the general transcription machinery has also been proposed on the basis of in vitro experiments (Johnson and Carey, 2003). However, because Med2 is still present in med3∆ med15∆ mutants, the possibility that recruitment occurs via this remaining tail module triad subunit cannot formally be excluded.

Several studies show that a loss of Mediator leads to a decrease in PIC components and diminished Pol II presence at active genes, highlighting the unilateral importance of Mediator in stimulating the assembly of a PIC (Eyboulet et al., 2015; He et al., 2008; Holstege et al., 1998; Paul et al., 2015; Thompson and Young, 1995; Kuras and Struhl, 1999; Li et al., 1999). Yet, how the PIC itself influences Mediator recruitment has received little attention (Eychenne et al., 2016; Robinson et al., 2016). The close spatial relationship of Mediator with the PIC at proximal promoters has been shown in single particle cryo-electron microscopy and cross-linking mass spectrometry studies of yeast in which Mediator is bound with the PIC (Nozawa et al., 2017; Plaschka et al., 2015; Robinson et al., 2016; Tsai et al., 2017). The TFIID-TBP complex and Pol II in particular occupy proximal promoters and have the potential to engage Mediator directly. In vivo, however, Mediator’s interaction with the PIC at proximal promoters is likely brief in light of recent evidence demonstrating that Mediator association at proximal promoters is transient, making interactions with the PIC difficult to assess in living cells (Fan and Struhl, 2009; Jeronimo and Robert, 2014; Wong et al., 2014).

To investigate PIC-dependent Mediator recruitment in vivo we have performed ChIP-seq against Mediator subunits in Saccharomyces cerevisiae after employing the anchor away method of nuclear depletion to export individual components of the PIC to the cytoplasm (Haruki et al., 2008). We confirm Mediator stabilization at proximal promoters in the kin28-AA strain. In the same background, we show that in med2Δ med3Δ med15Δ yeast, which completely lack the tail module triad, association of Mediator and Pol II is greatly reduced but nonetheless still detected at the majority of targets. We also investigate Mediator association by ChIP-seq after depletion of PIC components TBP, Rpb3, and Taf1, both alone and together with Kin28, to assess the contribution of the PIC to Mediator occupancy. Taken together, our findings provide new insights into mechanisms of Mediator recruitment during gene activation in yeast.

Results

The tail module triad is non-essential but is important for recruitment of Mediator and Pol II to all promoters

As outlined in the Introduction, recruitment of Mediator by a mechanism independent of the tail module triad has been proposed on the basis of experiments employing med3∆ med15∆ yeast mutants. However, the possibility that residual tail module function supported by Med2 is sufficient to allow Mediator and Pol II recruitment has not been formally excluded. Therefore, to assess more rigorously the role of the tail module triad in Mediator function, we sought to determine the effect of deletion of all three tail module triad subunits on yeast viability and Mediator recruitment.

Segregants from diploid yeast heterozygous for MED2, MED3, and MED15 (complete ORF deletions) included viable med2∆ med3∆ med15∆ mutants, which showed modest growth defects similar to those seen for med3∆ med15∆ yeast (Figure 1A). RNA-seq showed that two independent segregants exhibited genome-wide expression patterns more similar to each other than to wild type yeast (Figure 1B) (Fisher’s z-score = 37, p<10−10 for the null hypothesis that the correlations are not significantly different). Down-regulated genes were enriched for SAGA-dominated promoters (75 of 164 genes down-regulated by >2X in both segregants; p<10−34, hypergeometric test), while up-regulated genes showed marginal enrichment for SAGA-dominated promoters (35 of 442 genes, p=0.04), consistent with previous findings (Ansari et al., 2012a). RNA-seq also confirmed the triple deletion phenotype, with no reads present for sequences from the MED2, MED3, or MED15 ORF.

Figure 1 with 1 supplement see all
Mediator recruitment in med2∆ med3∆ med15∆ yeast.

(A) Spot dilutions (five-fold from left to right) of the indicated strains were made on YPD or CSM plates and allowed to grow at 30 ˚C for 3 days. The two med2∆ med3∆ med15∆ strains are independent segregants. (B) Comparison of transcript levels (reads per million) by RNA-seq for two independent med2∆ med3∆ med15∆ segregants (left panel), and between segregant one and the wild type parent strain (right panel). Only data for mRNA genes are plotted, after normalization to total mRNA. (C) Normalized Med17 occupancy in kin28-AA and med2∆ med3∆ med15∆ kin28-AA yeast after 1 hr treatment with rapamycin (designated ‘AAR’). Reads were mapped to all SAGA-dominated genes, all TFIID-dominated genes, and to the 498 genes exhibiting detectable Mediator ChIP signal at UAS regions in wild type yeast (Jeronimo et al., 2016). Genes were normalized for length and aligned by transcription start site (TSS) and transcription end site (TES), and are sorted according to average signal intensity. A compressed color scale was used for heat maps to better allow visualization of signal. (D) Browser scan showing normalized Med17 occupancy in wild type and med2∆ med3∆ med15∆ yeast over a short region of Chromosome VII; the upper end of the occupancy scale (reads) is indicated on the upper right for each graph. Note that the peak associated with LEU1 (TFIID-dominated) is essentially unchanged in med2∆ med3∆ med15∆ relative to wild type yeast, while the two peaks at the left (associated with TFIID-dominated transcripts) and that associated with PMA1 (SAGA-dominated) are all reduced in med2∆ med3∆ med15∆ yeast.

https://doi.org/10.7554/eLife.39633.002

We next examined the effect of loss of the tail module triad on Mediator recruitment. Mediator association with UAS elements is seen in ChIP experiments at only a fraction of active yeast genes, and its association with active promoters appears to be transient, making it difficult to assess Mediator association with gene regulatory regions in wild type yeast (Fan et al., 2006; Fan and Struhl, 2009; Jeronimo and Robert, 2014; Paul et al., 2015; Wong et al., 2014). Mediator ChIP signal at promoters is greatly enhanced under conditions in which phosphorylation of the Pol II CTD by the Kin28 subunit of TFIIH is prevented, apparently by inhibiting escape of Pol II from the promoter proximal region (Jeronimo and Robert, 2014; Wong et al., 2014). We therefore constructed a Kin28 anchor away (kin28-AA) strain harboring the med2∆ med3∆ med15∆ mutations to assess Mediator occupancy genome-wide. Treatment of this strain with rapamycin tethers Kin28 to the Rpl13A ribosomal subunit via FKBP12 and FRB tags on the respective proteins, and leads to Kin28 eviction from the nucleus during ribosomal protein processing (Haruki et al., 2008). The strain also includes the tor1-1 mutation, which prevents the normal stress response provoked by rapamycin (Haruki et al., 2008).

We used ChIP-seq to examine occupancy by the head module subunit Med17 (Srb4) (Bourbon et al., 2004) at SAGA-dependent and TFIID-dependent promoters (Huisinga and Pugh, 2004). Consistent with previous work, treatment of kin28-AA yeast harboring intact Mediator allowed detection of myc-tagged Med17 at the majority of promoters (Figure 1C–D) (Jeronimo and Robert, 2014; Wong et al., 2014), while parallel experiments using an untagged strain or an input sample gave rise to very low signal (Figure 1—figure supplement 1A). Mediator peaks observed after depletion of Kin28 corresponded well with those identified in a previous study (Figure 1—figure supplement 1B) (Jeronimo and Robert, 2014). In kin28-AA yeast lacking the Mediator tail module triad, Med17 ChIP signal was reduced much more than the reduction in Mediator occupancy (less than two-fold) reported for med3∆ med15∆ yeast (Jeronimo et al., 2016). Nonetheless, Med17 occupancy was still detectable at the same promoters, while an untagged control normalized to input yielded no signal (Figure 1C–D; Figure 1—figure supplement 1C). The reduction in Med17 occupancy was proportionally larger at SAGA-dominated genes than at TFIID-dominated genes, possibly reflecting the enrichment of SAGA-dominated genes among genes depending most strongly on the tail module triad for transcription, but association with both categories of gene promoters was generally reduced (Figure 1C–D) (Ansari et al., 2012a).

We also examined Med17 occupancy for a set of 498 genes for which Mediator association was detected at UAS regions in KIN28-WT yeast, which we refer to as ‘UAS genes’ (Jeronimo et al., 2016) (Figure 1C, right panel). Med17 occupancy is readily visible at promoters and UAS regions of these genes in kin28-AA yeast having an intact tail module triad (Figure 1C, upper right panel). Mediator head module subunits exhibit weaker ChIP signal at many UAS regions than do tail module subunits (Jeronimo et al., 2016; Petrenko et al., 2016); nonetheless, Mediator occupancy was evident at many UAS regions, as seen in the heat map and the shoulder on the upstream side of the line graph in Figure 1C, upper right panel. This occupancy appears to be reduced to near baseline levels in med2∆ med3∆ med15∆ yeast, as was reported previously for med3∆ med15∆ yeast (Jeronimo et al., 2016; Petrenko et al., 2016).

Despite a considerable decrease in Mediator occupancy at gene promoters as measured by ChIP in med2∆ med3∆ med15∆ yeast, only modest effects were observed on growth and transcript levels. Previous work has uncovered several examples in which decreased transcription caused by loss or impairment of general transcription factors, including Pol II, is compensated for by increased mRNA stability (Baptista et al., 2017; Sun et al., 2012; Warfield et al., 2017). Therefore, to investigate more directly the effect of decreased Mediator recruitment on transcription, we examined Pol II occupancy in wild type and med2∆ med3∆ med15∆ yeast by ChIP-seq. Wild type and med2∆ med3∆ med15∆ yeast whole cell extracts were spiked with extracts from S. pombe prior to immunoprecipitation with a monoclonal antibody to Rpb1, the largest subunit of Pol II, to allow global changes in Pol II occupancy to be measured. We observed an approximately four-fold decrease in Pol II occupancy in med2∆ med3∆ med15∆ yeast, while the normalized spike-in control showed equal IP efficiency in the two samples (Figure 2). Two features of this data are notable: first, both Mediator and Pol II occupancy are reduced at essentially all genes in med2∆ med3∆ med15∆ yeast; and second, substantial occupancy by Mediator and Pol II is observed even in the complete absence of the tail module triad. Thus, the tail module triad contributes broadly if not universally to recruitment of Mediator and Pol II, but at the same time is not absolutely required at the large majority of mRNA genes.

Decreased Pol II occupancy in med2∆ med3∆ med15∆ yeast.

(A) ChIP-seq of Rpb1 in wild type (BY4741) and med2∆ med3∆ med15∆ (∆tail) yeast. Left panel: normalized occupancy at S. cerevisiae UAS genes, defined as in Figure 1. Right panel: normalized occupancy over 1150 s. pombe genes having expression levels > 1 (Shetty et al., 2017). (For unknown reasons, ChIP-seq against Rpb1 in S. pombe exhibits low signal at 5’ regions and stronger signal at and beyond 3’ regions of transcribed genes; see for example (Shetty et al., 2017; Lee et al., 2017)). (B) Browser scans showing normalized Rpb1 occupancy in wild type (BY4741) and med2∆ med3∆ med15∆ yeast.

https://doi.org/10.7554/eLife.39633.004

Mediator association at UAS regions is stabilized by loss of TBP or Pol II

Previous work had indicated, and results of Figure 1 demonstrate unequivocally, that Mediator can associate with many promoters independently of the tail module triad, albeit at lower occupancy (Jeronimo et al., 2016; Petrenko et al., 2016). Based on known interactions of Mediator head and middle subunits with the general transcription machinery (Plaschka et al., 2015; Plaschka et al., 2016; Ansari and Morse, 2013; Soutourina et al., 2011; Eychenne et al., 2016; Esnault et al., 2008; Nozawa et al., 2017; Tsai et al., 2017; Robinson et al., 2016), we had previously suggested that Mediator might be recruited to tail module-independent targets via interactions with PIC components (Ansari et al., 2012a; Ansari and Morse, 2012b). To examine this possibility, we monitored Mediator occupancy genome-wide before and after depletion of PIC components using the anchor away method (Haruki et al., 2008).

We chose three PIC components to subject to depletion: TATA-binding protein (TBP); Rpb3, an essential subunit of Pol II; and Taf1, a TFIID subunit that is not shared with the SAGA complex (Bhaumik, 2011). Depletion of any of these PIC components resulted in loss of viability, as expected (Figure 3—figure supplement 1A–B) (Haruki et al., 2008; Petrenko et al., 2017; Warfield et al., 2017). Pilot experiments showed that treatment of tbp-AA yeast with rapamycin resulted in depletion of TBP to near background levels, as measured by ChIP, within 30–60 min, consistent with prior work (Figure 3—figure supplement 2A) (Haruki et al., 2008; Wong et al., 2014; Tramantano et al., 2016; Grimaldi et al., 2014). Previous studies using anchor away of components of the transcription machinery have employed rapamycin treatment times varying from 30 min to 2 hr (Anandhakumar et al., 2016; Baptista et al., 2017; Haruki et al., 2008; Jeronimo et al., 2016; Petrenko et al., 2016; Petrenko et al., 2017; Tramantano et al., 2016; Warfield et al., 2017; Wong et al., 2014). Based on these studies and our results on TBP depletion, we chose to use one hr of rapamycin treatment as a time likely to allow a reasonable balance between attaining thorough depletion while also keeping indirect effects to a minimum.

We first examined Mediator binding before and after depletion of TBP, Rpb3, and Taf1 using myc-tagged Med15, from the tail module triad, and Med18, from the head module. As mentioned earlier, Mediator ChIP signal at promoters is generally low due to rapid turnover during transcription, and many active promoters do not exhibit detectable Mediator ChIP signal at UAS sites (Jeronimo and Robert, 2014; Wong et al., 2014). We therefore narrowed our examination of Mediator occupancy to the 498 ‘UAS genes’ that exhibit Mediator occupancy at UAS regions in wild type yeast (Jeronimo et al., 2016). In agreement with prior reports, we observed association of both Med15 and Med18 at UAS regions for most of these genes, while a shift to the promoter region was seen for both Mediator subunits in kin28-AA yeast after rapamycin treatment (Figure 3A).

Figure 3 with 4 supplements see all
Effect of depletion of TBP and Rpb3 on Mediator occupancy.

ChIP-seq was performed against (A) Med15 and (B) Med18 in tbp-AA, rpb3-AA, and taf1-AA yeast with (‘AAR’) and without (‘AA’) rapamycin treatment, and in kin28-AA yeast after rapamycin treatment. Heat maps and line graphs (normalized reads per million, RPM) depict results combined from two independent experiments (except for Med18 in taf1-AA yeast) averaged over 498 genes exhibiting detectable Mediator ChIP signal at UAS regions in wild type yeast (Jeronimo et al., 2016). (C) Browser scans showing occupancy of Med15 at two chromosomal regions in tbp-AA, rpb3-AA, and taf1-AA yeast with and without rapamycin treatment. YBL044W, between COR7 and ECM13, is an uncharacterized ORF and likely contains a UAS, as has been observed at many other uncharacterized or dubious ORFs (Paul et al., 2015). (D) Med2 and Med14 occupancy measured by ChIP-seq with and without depletion of TBP or Rpb3, at UAS genes.

https://doi.org/10.7554/eLife.39633.005

We had anticipated that depletion of PIC components might, by reducing interactions with Mediator at promoter regions, reduce Mediator occupancy at both promoter and UAS regions. However, Med15 and Med18 occupancy at UAS regions did not decrease upon depletion of TBP, Rpb3, or Taf1 (Figure 3A–B). For Med15, high occupancy was observed at UAS regions upon depletion of all three PIC components; occupancy was increased relative to that seen before depletion for TBP and Rpb3, but was constitutively high in the taf1-AA strain. This behavior was also evident at individual promoters, as Med15 occupancy increased at UAS sites upon depletion of TBP or Rpb3 but was constitutively high in taf1-AA yeast (Figure 3C). Consistent with these results, comparison of Med15 ChIP signal at UAS genes in FRB-tagged anchor away strains with an untagged control indicated that no effect of the epitope tag was seen in rpb3-AA yeast in the absence of rapamycin, while a modest increase (about 1.5 to 2-fold) was observed in tbp-AA yeast. A greater increase in signal was observed in taf1-AA yeast (Figure 3—figure supplement 2B); possibly the FRB tag in the latter strain interferes in some way with PIC assembly or function. For Med18, the ChIP signal was unchanged by depletion of TBP, Rpb3, or Taf1 (Figure 3B). ChIP-seq confirmed that all three PIC components were depleted, although Taf1 depletion was somewhat less efficient (Figure 3—figure supplements 3A–C and and 4A). Depletion of TBP essentially eliminated Pol II association, as expected, while depletion of Pol II resulted in a moderate reduction in TBP association, consistent with recently published work (Figure 3—figure supplement 3D) (Joo et al., 2017). ChIP-seq before and after depletion of TBP and Rpb3 against another tail module triad subunit, Med2, and against the scaffold subunit Med14 yielded results most similar to those seen with Med15, with both subunits showing increased occupancy after depletion of TBP or Rpb3 (Figure 3D).

The increased ChIP signal for Med15, Med2, and Med14 observed at UAS regions upon depletion of TBP or Rpb3 suggested that PIC component depletion might result in Mediator peaks being observed at UAS regions where they are normally difficult to detect. To test this idea, we examined Med15 ChIP-seq signal over all genes having transcription level >32 (Nagalakshmi et al., 2008), after filtering out the 498 genes identified as exhibiting Mediator UAS peaks, leaving about 1400 genes (Figure 4A). Indeed, this analysis revealed several hundred genes that showed increased Mediator occupancy in their UAS regions upon TBP depletion where negligible occupancy was observed by ourselves or others under normal growth conditions (Jeronimo et al., 2016). Gene ontology (GO) analysis of the 200 genes from this group showing the greatest increase in Med15 occupancy revealed a strong enrichment for ribosomal protein (RP) genes, and a direct examination of RP genes showed increased Mediator occupancy at UAS regions upon depletion of TBP (Figure 4B and C). Promoters of some non-RP, ‘non-UAS’ gene promoters also showed increased Mediator occupancy upon depletion of TBP; an example is shown in Figure 4C. Taken together, the results of Figures 34 indicate that Mediator association with UAS regions does not decrease upon depletion of PIC components, and that association apparently increases upon depletion of TBP or Pol II.

Effect of depletion of TBP on Mediator occupancy at ‘non-UAS’ genes.

(A) Med15 occupancy measured by ChIP-seq at the 2000 most highly transcribed genes, after removal of the 498 genes exhibiting detectable Mediator ChIP signal at UAS regions in wild type yeast (Jeronimo et al., 2016), in tbp-AA yeast with and without rapamycin treatment. Occupancy after depletion of Kin28 (bottom) is shown for reference. (B) Med15 occupancy measured by ChIP-seq at ribosomal protein (RP) genes and at the DLD1-PAR32 promoter in tbp-AA yeast with and without rapamycin treatment.

https://doi.org/10.7554/eLife.39633.010

A possible explanation for increased occupancy of Mediator subunits at UAS regions upon depletion of TBP or Rpb3 is that interruption of the transcription cycle by PIC disruption prevents the transit of Mediator from UAS regions to promoters and off the DNA. At genes for which Mediator association with UAS regions is observed, a single Mediator complex bridges the UAS and the promoter, with tail module subunits contacting the UAS and head and middle subunits engaging the promoter region through their contacts with Pol II and GTFs (Jeronimo et al., 2016; Petrenko et al., 2016). Interruption of the latter contacts could strand Mediator at the UAS. It is not clear why an increase in ChIP signal is observed for Med15, but not for Med18. One possibility is that the tail module, which can associate with UAS regions independently of the rest of Mediator under some conditions (Ansari et al., 2009; He et al., 2008; Paul et al., 2015; Zhang et al., 2004), may bind to UAS regions independently of the head module at least some of the time (Anandhakumar et al., 2016). Arguing against this explanation is the observation that Med14, which contacts subunits from Mediator head, middle, and tail modules, shows increased occupancy upon depletion of TBP or Rpb3, thus behaving similarly to Med15 and Med2 subunits from the tail module triad. It is also possible that Mediator may adopt an altered configuration under these conditions, such that immunoprecipitation of Med18 is less efficient upon depletion of TBP or Rpb3. In this regard, it is notable that under conditions of Kin28 depletion, ChIP signal for Med15 is observed at UAS regions of some genes while subunits from the head module do not give detectable signal, while ChIP signals are observed for both Med15 and head module subunits at proximal promoter regions (Figure 3—figure supplement 4B) (Jeronimo et al., 2016).

Although the notion of Mediator transit from UAS to promoter depending on PIC integrity is consistent with recent models of Mediator function at UAS regions (Jeronimo et al., 2016; Petrenko et al., 2016), the results so far do not help clarify how Mediator is recruited to promoters at the many genes for which Mediator association is not readily observed at upstream regions, or how recruitment occurs in the absence of the Mediator tail module triad. We therefore next modified our approach to allow PIC component depletion under conditions in which Mediator association with gene promoters was stabilized.

Mediator association with promoters depends on PIC components

We next asked how PIC component depletion affected Mediator association under conditions in which Mediator binding to proximal gene promoters was stabilized. For this purpose, we constructed double anchor away strains in which Kin28 and either TBP, Taf1, or Rpb3 were tagged with FRB to allow nuclear depletion upon treatment with rapamycin. We then separately introduced Myc-tagged Med15 and Med18 into each double anchor away strain. Depletion of Kin28 alone allows detection by ChIP of Mediator occupancy at promoters (Jeronimo and Robert, 2014; Wong et al., 2014); comparison with the double depletion strains after rapamycin treatment then allows determination of the effect of depleting TBP, Taf1, and Rpb3 on Mediator occupancy at gene promoters. As expected, these strains were all sensitive to rapamycin (Figure 3—figure supplement 1). We note that the kin28-tbp-AA strain harboring the Med15-myc epitope was slow growing; the med15-myc allele has been previously been found to behave as a hypomorph in conjuction with a med3∆ mutation (Ansari et al., 2012a; Zhang et al., 2004), and it evidently exhibits a negative growth phenotype in the context of the kin28-tbp-AA strain as well.

We first examined the effect of depleting PIC components on Mediator occupancy at all genes, separated into SAGA-dominated and TFIID- dominated genes (Huisinga and Pugh, 2004). Our rationale was predicated on findings that genes that depend on the tail module triad for transcription are enriched for SAGA- dominated genes, relative to TFIID- dominated genes (Ansari et al., 2012a), and that loss of the tail module caused a more pronounced decrease in Mediator occupancy at promoters of SAGA-dominated genes than at TFIID-dominated genes (Figure 1). Based on these observations, we hypothesized that TFIID-dominated genes might show greater dependence than SAGA-dominated genes on PIC components for Mediator occupancy.

Depletion of either Taf1 or Rpb3 resulted in decreased occupancy of Med15 and Med18 in comparison to depletion of Kin28 alone at almost all TFIID-dominated genes (Figure 5A–D). Decreased Med15 and Med18 occupancy was also observed at SAGA-dominated genes and UAS genes upon depletion of Rpb3, along with a partial shift towards the UAS region, while depletion of Taf1 also resulted in decreased Med18 occupancy but little change in Med15 occupancy, other than a small upstream shift, at SAGA-dominated and UAS genes. A more pronounced decrease in Med15 or Med18 occupancy at TFIID-dominated genes compared to SAGA-dominated genes after depletion of Rpb3 or Taf1 could also be observed at genes having similar levels of transcription (Figure 5—figure supplement 1), and examples can also be found of individual TFIID-dominated genes exhibiting greater loss of Mediator ChIP signal than at SAGA-dominated genes (Figure 5—figure supplement 2). We conclude that under conditions of Kin28 depletion, Mediator occupancy at promoters of TFIID-dominated genes, and to lesser extent SAGA-regulated genes, depends on both Taf1 and Pol II.

Figure 5 with 4 supplements see all
Effect of depletion of PIC components together with Kin28 on Mediator occupancy.

(A) Heat maps showing occupancy of Med15 after depletion of Kin28 alone, or in combination with Taf1, TBP, or Rpb3, at SAGA-dominated and TFIID-dominated genes. (B) Same as (A), for Med18 occupancy. (C) Line graphs (normalized reads per million) depicting Med15 occupancy after depletion as in (A) and (B), aligned at the TSS and shown for SAGA-dominated (top), TFIID-dominated (middle), and UAS (bottom) genes. (D) Same as (C), but for Med18. Combined data for two replicate experiments was used for kin28-tbp-AA yeast; all other plots derive from single experiments that were consistent with replicate experiments. We note that variable results were observed for the effect on Med15 occupancy in kin28-taf1-AA yeast at SAGA-dominated genes, with peak intensity being reduced in some experiments and nearly unchanged in others, while reduced occupancy was consistently observed at TFIID genes.

https://doi.org/10.7554/eLife.39633.011

TBP is required for transit of Mediator from UAS to promoter

The effect of depleting TBP together with Kin28 was markedly different from that of depleting Taf1 or Rpb3 and Kin28 (Figure 5A–D). At SAGA-dominated genes, occupancy of Med15 from the Mediator tail module did not decrease and shifted upstream in kin28-tbp-AA yeast compared to kin28-AA yeast (Figure 5A & C). This shift is also evident at UAS genes (Figure 5C), and yields a profile nearly identical to that seen upon depletion of TBP alone (Figure 5—figure supplement 3), suggesting that Mediator transit from UAS to promoter was inhibited by depletion of TBP. At TFIID-dominated genes, depletion of TBP resulted in an upstream shift of smaller magnitude, along with a reduction in the Med15 ChIP signal (Figure 5A & C). The smaller shift seen for TFIID-dominated genes than for SAGA-dominated genes is consistent with the average distance between UAS and TSS being larger for TATA-containing than for TATA-less promoters (Erb and van Nimwegen, 2011; Kristiansson et al., 2009), supporting the notion that depletion of TBP causes Mediator to be stranded at the UAS.

Med18, from the head module, as well as Med2 from the tail module triad and Med17 from the head module all showed a shift upstream at SAGA-dominated genes upon TBP depletion, but the signals for these subunits were reduced to varying degrees upon depletion of Kin28 and TBP compared to those observed upon depletion of Kin28 alone (Figure 5 and Figure 5—figure supplement 4). At TFIID-dominated genes, depletion of TBP and Kin28 resulted in a nearly complete loss of ChIP signal for Med18, Med2, and Med17 (Figure 5B and D and Figure 5—figure supplement 4). Examination of individual gene loci corroborates these observations, with Mediator occupancy shifting upstream at SAGA-dominated genes and often disappearing at TFIID-dominated genes upon TBP depletion (Figure 6).

TBP depletion shifts Mediator occupancy from promoter to UAS sites.

Browser scans showing normalized Med15 and Med18 occupancy in kin28-AA and kin28-tbp-AA yeast treated with rapamycin, as indicated. (A) Upstream shift of Mediator at a SAGA-dominated gene and loss of occupancy at a TFIID-dominated gene upon depletion of TBP. Arrows depict direction of transcription. (B) Loss of Mediator peaks at promoters and gain of Mediator peaks at upstream regions at both TFIID and SAGA-dominated genes. YBL044W is an uncharacterized ORF and likely contains a UAS, as has been observed at many other uncharacterized or dubious ORFs (Paul et al., 2015). (C) At some loci, new Mediator peaks are visible after dual depletion of Kin28 and TBP (boxed regions).

https://doi.org/10.7554/eLife.39633.016

Depletion of Rpb3 together with Kin28 strongly decreased occupancy of both Med15 and Med18, especially at TFIID-dominated genes (Figure 5). In addition, a shift in Med15 occupancy to more upstream regions (together with an overall decrease in occupancy) was seen at SAGA-dominated genes upon depletion of Rpb3 and Kin28 (Figure 5A & C). This molecular phenotype is also apparent upon analysis of the 498 genes at which Mediator occupancy is observed in wild type yeast (Figure 5C) (Jeronimo et al., 2016).

The effects seen in kin28-rpb3-AA yeast may be due to a direct effect on Mediator transit to and association with promoters, or may reflect the impact of loss of Pol II on the cooperative assembly of the PIC. Depletion of Pol II results in partial loss of TBP (Figure 3—figure supplement 3D) (Ansari et al., 2014; Joo et al., 2017; Sharma et al., 2003). TBP-containing preinitiation complexes that are present, but which lack Pol II, may facilitate transit of Mediator from the UAS to the promoter, but association may be destabilized by the absence of normal contacts between Mediator and Pol II (Soutourina et al., 2011) or other PIC components that are less stably bound in the absence of Pol II. The combination of these effects could produce the effects seen upon depletion of Kin28 and Rpb3. This differential dependence on Rpb3 and TBP for Mediator transit may pertain only under the special conditions of Kin28 depletion, as increased Mediator ChIP signal was observed at UAS regions when either TBP or Rpb3 was depleted when Kin28 activity was unperturbed (Figure 3).

The cooperative nature of PIC assembly must be taken into account when interpreting results of depletion of individual PIC components. Depletion of TBP abolishes recruitment of Pol II (Figure 3—figure supplement 3D), and this is also the case when Kin28 is simultaneously depleted (Figure 7A). These results, in conjunction with the upstream shift in Mediator occupancy upon depletion of Kin28 and TBP compared to depletion of Kin28 alone (Figure 5), allow us to infer that TBP is necessary for Mediator transit to promoter regions under conditions of Kin28 depletion, but not whether it is sufficient, as Pol II may contribute. In addition, since Taf1 association is only slightly decreased upon depletion of TBP by itself or simultaneously with Kin28 (Figure 7B), we infer that Taf1 is not sufficient for Mediator transit from UAS to promoter.

Normalized PIC occupancy at UAS genes after PIC component depletion.

(A) Rbp1 occupancy after depletion of Kin28 (black), Kin28 and TBP (red), and Kin28 and Taf1 (green). (B) Taf1 occupancy before (black) and after (red) depletion of TBP (top graph) or TBP and Kin28 (bottom graph). (C) Taf1 occupancy before (black) and after (red) depletion of Rpb3. The prominent peaks observed near the TES appear to represent artifactual ChIP signal sometimes seen at 3’ regions of highly transcribed genes; similar signals are observed in ChIP against GFP expressed in yeast (Paul et al., 2015).

https://doi.org/10.7554/eLife.39633.017

With regard to the role of Taf1 and Pol II in stable occupancy of Mediator at gene promoters (Figure 5), depletion of Rpb3 results in a modest (less than two-fold) reduction in association of both TBP and Taf1 (Figure 3—figure supplement 3D and Figure 7C). In contrast, depletion of Taf1 has been shown to reduce Pol II occupancy on average about 4-fold, with little difference between SAGA-dominated and TFIID-dominated genes (Warfield et al., 2017). However, Taf1 (and presumably TFIID) by itself appears not to be sufficient to retain Mediator at gene promoters after Kin28 depletion, as Taf1 is retained when Rpb3 is depleted, while Mediator occupancy is essentially abolished. Evidently the partial loss of Pol II accompanying depletion of Taf1 is not sufficient to cause appreciable loss of TBP, as no upstream shift in Mediator occupancy was observed upon depletion of Taf1 and Kin28. The alternative possibility that Taf1 is required for recruitment of Mediator to UAS regions is ruled out by our observation that Mediator occupancy of UAS regions was not reduced upon depletion of Taf1 alone (Figure 3). Finally, it appears that the level of Pol II still associated with gene promoters after Taf1 depletion is sufficient to allow Mediator to stably associate, at least at SAGA-dominated genes (compare effects on Med15 occupancy in kin28-rpb3-AA and in kin28-taf1-AA yeast (Figure 5A & C).

In summary, our results indicate that TBP, which is critical for PIC formation, is required for transit of Mediator from UAS to gene promoters, while partially formed PICs present after depletion of Taf1 or Rpb3 allow this transit but lead to unstable Mediator-PIC association at promoters under conditions that prevent the normal rapid turnover of Mediator that accompanies release of Pol II.

Discussion

The major findings reported in this work are that 1) the Med2-Med3-Med15 tail module triad plays a major role in recruitment of Mediator and Pol II, but is dispensable for viability; 2) Mediator association with active gene promoters is observed even in the complete absence of the tail module triad; 3) PIC components, in particular Taf1 and Pol II (Rpb3) contribute to Mediator association with gene promoters, especially at TFIID-dominated genes, and 4) TBP plays a critical role in the transit of Mediator from UAS to promoter sites.

Previous studies reported altered transcript levels for only 5–10% of all genes in mutants lacking one or two tail module triad subunits, suggesting that this moiety might serve as a requisite target for transcriptional activators only for a limited fraction of the genome (Ansari et al., 2012a; van de Peppel et al., 2005). Pol II occupancy was only moderately reduced in med3∆ med15∆ yeast, particularly at TFIID-dominated genes, and Mediator occupancy at gene promoters under conditions of Kin28 inactivation was reduced by less than two-fold in this same mutant, further suggesting a limited role for the tail module triad in recruitment of Mediator and Pol II (Jeronimo et al., 2016; Paul et al., 2015). In the present work, however, we find a more substantial reduction of occupancy by Mediator and Pol II in med2∆ med3∆ med15∆ yeast, indicating a widespread role for the tail module triad in recruitment of the Mediator complex and functional redundancy in the tail module subunits, reminiscent of previous observations in comparing med3∆ med15∆ yeast to single deletion mutants (Ansari et al., 2012a). This redundancy could reflect ‘fuzzy’ interactions between activators and subunits of the tail module triad, such that loss of any individual tail module triad subunit would result in loss but not elimination of recruitment and consequent activation (Tuttle et al., 2018).

Recruitment of Mediator via the tail module triad is likely to occur via activators bound to UAS sites, as in the canonical model for gene activation (Figure 8A). Consistent with this model is the finding that Mediator is detected upstream of active genes by ChEC-seq, at distances consistent with occupancy of UAS sites, in both the presence and absence of functional Kin28 (Grünberg et al., 2016; Grünberg and Zentner, 2017). This argument is further supported by our findings that Mediator occupancy is observed by ChIP at upstream regions of genes when both TBP and Kin28 are depleted (Figure 5). Altogether, then, these various observations support a model in which Mediator is principally recruited first to UAS sites by activators contacting subunits of the tail module triad before transiting to the promoter and facilitating PIC formation.

Pathways of Mediator recruitment.

(A) (i) Mediator recruitment via interactions between activator proteins bound to UASs and the Med2-Med3-Med15 triad. Contacts between Mediator and components of the general transcription machinery facilitate transit of Mediator to the proximal promoter region, and stabilize association of both Mediator and GTFs. (ii) TBP plays a critical role in this transition, and its depletion results in Mediator being stranded at the UAS. (B) (i) Mediator recruitment via interactions between the middle/head modules and components of the PIC. In this pathway, the PIC is first recruited via interactions between UAS-bound activators and PIC components, such as Tafs within TFIID. (ii) Depletion of PIC components destabilizes association of Mediator with the proximal promoter, particularly at TATA-less, TFIID-dominated promoters.

https://doi.org/10.7554/eLife.39633.018

Although the dominant mechanism of Mediator recruitment is likely to occur via tail-module dependent interactions with UAS-bound activators, our findings that Mediator and Pol II are recruited to gene promoters even in the complete absence of the tail module triad indicate the existence of an alternative, tail-module-independent mechanism for Mediator recruitment (Figure 8B). In this second pathway, PIC components could be recruited before or together with Mediator, presumably via activator-mediated interactions, and Mediator association would be facilitated by contacts with PIC components and in turn stabilize binding of those components (Figure 8B). In both pathways, following formation of the Mediator-PIC complex, TFIIH is recruited and its Kin28 subunit phosphorylates the carboxyl-terminal domain of the Rpb1 subunit of Pol II; this allows escape of Pol II into the elongation complex, and results in rapid loss of Mediator association (Jeronimo and Robert, 2014; Wong et al., 2014). Direct recruitment of Mediator to promoter regions via interactions with PIC components may be favored at TFIID-dominated genes relative to SAGA-dominated genes; this would be consistent with the greater effect on Mediator occupancy seen at the former compared to the latter class upon depletion of PIC components (Figure 5).

Our results indicate that PIC components are important for transit of Mediator from UAS to promoter, and for association with promoters following that transit. PIC components are not necessary for recruitment of Mediator to UAS sites, as their depletion does not lead to reduction in Mediator association (Figure 3). Depletion of Taf1 results in greatly diminished Mediator occupancy at TFIID-dominated promoters, and little effect at SAGA-dominated promoters, whereas depletion of TBP causes a shift in Mediator occupancy to UAS regions at both promoter categories, along with diminished occupancy at TFIID-dominated genes (Figure 5). Thus, Taf1, and presumably TFIID, is important for stabilizing Mediator occupancy at promoters of TFIID-dominated genes following its transit from the initial site of recruitment, whereas TBP is critical for the transit from UAS to promoter of both SAGA- and TFIID-dominated genes (Figure 8, right). Depletion of Rpb3 together with Kin28 causes a decrease in Mediator occupancy at promoters, but also results in a new, upstream peak of Mediator occupancy that is much lower in intensity than that seen upon depletion of TBP and Kin28 (Figure 5C–D). We interpret this as a combined effect of Pol II stabilizing Mediator at promoter regions, together with Pol II depletion causing a partial decrease in TBP occupancy, which causes Mediator occupancy to shift upstream in a fraction of the cells being sampled (Figure 3—figure supplement 3D) (Joo et al., 2017).

Related to our findings that PIC components facilitate Mediator association with active gene promoters, in vitro results have indicated cooperative interactions among Mediator, TFIID, and Pol II (Johnson and Carey, 2003; Johnson et al., 2002; Baek et al., 2002; Takahashi et al., 2011). In vivo evidence for such cooperative interactions, however, has been sparse. An ‘activator bypass’ experiment showed that direct recruitment of TFIIB, via tethering to the DNA-binding domain of an activator, resulted in recruitment of a functional PIC and the Mediator complex in yeast, pointing to ‘backwards’ recruitment of Mediator via PIC interactions (Lacombe et al., 2013). More recently, depletion of Taf1 was shown to result in a widespread reduction in association of head module subunit Med8 with active genes when mapped by ChEC-seq (Grünberg et al., 2016; Grünberg and Zentner, 2017). Neither of these in vivo studies, however, examined PIC-Mediator cooperativity at promoter regions, and the distinct effects of depleting different PIC components on Mediator association (Figure 5) could not have been predicted on the basis of previous work.

Several interactions between Mediator and TFIID components, and Mediator and Pol II, have been documented that could contribute to Mediator association with promoters. Structural studies have shown the head module subcomplexes Med8-Med18-Med20 and Med11-Med17-Med22 bind with TBP (Cai et al., 2010; Larivière et al., 2006). Disruption of the Med8-Med18-Med20 subcomplex in med18Δ or med20Δ yeast does not cause loss of viability, suggesting that multiple Mediator-TBP interactions are possible at the proximal promoter for initiation (Ranish et al., 1999). In addition, interaction between the tail subunits Med15 or Med16 with TFIID subunits Taf14 or Taf1 have been observed in yeast two-hybrid and mass spectrometry experiments (Gavin et al., 2006; Gavin et al., 2002; Lim et al., 2007). Interactions between Mediator and Pol II have also been demonstrated in both structural and genetic studies (Davis et al., 2002; Esnault et al., 2008; Plaschka et al., 2015; Soutourina et al., 2011; Cai et al., 2010; Robinson et al., 2015).

Interactions between Mediator and Pol II or TFIID appear more important for stabilizing Mediator occupancy at TFIID-dominated genes than for SAGA-dominated genes, based on the stronger effect on Mediator occupancy in med2∆ med3∆ med15∆ yeast at the latter class than the former (Figure 1), and the stronger effect seen at TFIID-dominated genes upon depletion of Taf1 or Rpb3 (Figure 5). The distinction of these two pathways in terms of dependence on PIC components is also suggested by recent work showing that disruption of a Mediator-TFIIB connection via med10-ts yeast differently reduces association of Mediator, TFIIB, and GTFs (Eychenne et al., 2016). TFIID-dominated genes exhibit a greater reduction in Mediator rather than GTFs, while SAGA-dominated genes exhibit a greater reduction in GTFs rather than Mediator (Eychenne et al., 2016). Furthermore, two recent studies report that Mediator in general, and not just tail module subunits, contributes more strongly to Pol II recruitment at SAGA-dominated than TFIID-dominated genes (Bruzzone et al., 2018; Petrenko et al., 2017).

What is the basis for the differential dependence of SAGA-dominated and TFIID-dominated genes on Mediator? The average distance between UAS and TSS is greater for TATA-containing than TATA-less genes (Kristiansson et al., 2009; Erb and van Nimwegen, 2011). UASs show diminished ability to activate transcription from a reporter gene as the distance between UAS and TATA box increases (Dobi and Winston, 2007), and Mediator mutations were found to affect the dependence of UAS activity on distance (Reavey et al., 2015). Thus, the greater UAS-TSS distance of TATA-containing gene promoters may cause a greater dependency on Mediator for activation.

In summary, the work here provides new insights into the role of the Mediator tail module triad and PIC components in Mediator recruitment and dynamics. Future work will be aimed at a fuller understanding of the mechanism of tail module-independent recruitment of Mediator, and of the differential dependence of SAGA-dominated and TFIID-dominated genes on PIC components for stable Mediator association at gene promoters.

Materials and methods

Key resources table
Reagent
type (species)
or resource
DesignationSource
or reference
IdentifiersAdditional
information
Strain, strain background (S. cerevisiae)yFR1321PMID: 24704787Francois Robert (University of Montreal)
Strain, strain background (S. cerevisiae)KHW127PMID: 24746699Kevin Struhl (Harvard Medical School)
Strain, strain background (S. cerevisiae)LS01PMID: 15254252Alan Hinnebusch (National Institutes of Health)
Strain, strain background (S. cerevisiae)LS10PMID: 15254252Alan Hinnebusch (National Institutes of Health)
Strain, strain background (S. pombe)FWP510PMID: 24100010Fred Winston (Harvard Medical School)
Antibody8WG16 (Pol II) (mouse monoclonal)CovanceCat# 664906ChIP, 2 μg (1:150)
AntibodyRpb3 (mouse monoclonal)NeocloneCat# W0012ChIP, 1 μg (1:300)
Antibody9E10 (Myc) (mouse monoclonal)RocheCat# 11667141001ChIP, 2 μg (1:150)
Antibody9E10 (Myc) (mouse monoclonal)SigmaCat# 11667149001ChIP, 5 μg (1:300)
Antibodyanti-protein A (rabbit polyclonal)SigmaCat# P3775ChIP, 2.5 μg (1:250)
AntibodyTBP (affinity-purified rabbit polyclonal)Tony Weil, Vanderbilt UniversityNAChIP, 2.5 μg (1:150)
Antibody58C9 (TBP) (mouse monoclonal)AbcamCat# ab61411ChIP, 5 μg (1:300)
AntibodyTaf1 antisera (rabbit polyclonal)Joseph Reese and Song Tan, Pennsylvania State UniversityNAChIP, 2 μL (1:250)
Recombinant DNA reagentpFA6a-FRB-kanMX6EuroscarfP30578
Recombinant DNA reagentpFA6a-2-FKB12-His3MX6EuroscarfP30583
Recombinant DNA reagentpSC11Burke et al., 2000
Commercial assay or kitMagnetic Bead Isolation KitNew EnglandBiolabsCat# S1550S
Commercial assay or kitNEB Next Ultra RNA Library Prep KitNew England BiolabsCat# E7530S
Commercial assay or kitNEBNext Ultra II Library Prep KitNew England BiolabsCat# E7645S
Commercial assay or kitNEBNext Multiplex Oligos for IlluminaNew England BiolabsCat# E7710
Commercial assay or kitNEXTflex barcodesBIOO ScientificCat# 514122
Chemical compound, drugRapamycinLC LaboratoriesCat# R-5000
Software, algorithmbwaPMID: 10571001RRID: SCR_010910
Software, algorithmSICERPMID: 19505939RRID: SCR_010843
Software, algorithmBioConductorPMID: 15461798RRID:SCR_006442
Software, algorithmIntegrative Genomics ViewerPMID: 21221095RRID: SCR_011793
Software, algorithmtophat2PMID: 23618408RRID: SCR_013035 (Tophat)

Yeast strains

Anchor away (AA) strains used in this study were constructed as described previously and contain the tor1-1 mutation and deletion of FPR1 (Haruki et al., 2008). To allow the AA technique, Taf1, TBP, and Rpb3 were chromosomally tagged with the FRB domain by lithium acetate transformation of PCR products amplified from pFA6a-FRB-kanMX6 into the AA parent strain (yFR1321, generous gift of F. Robert, Institut de Recherches Cliniques de Montréal) (Hill et al., 1991; Longtine et al., 1998; Jeronimo and Robert, 2014). Mediator subunits Med2-TAP, Med14-TAP, Med15-myc, Med18-myc, and Med17-myc were tagged by transformation of PCR products amplified from prior strains, as described (Ansari et al., 2014; Paul et al., 2015).

To generate double kin28-taf1-AA and kin28-rpb3-AA strains containing either Med15-myc or Med18-myc, the kin28-AA strain (KHW127, a generous gift of Kevin Struhl, Harvard Medical School) (Wong et al., 2014) was first switched from MATα to MATa using the pSC11 plasmid (Burke et al., 2000), and then mated with the taf1-AA and rpb3-AA strains, followed by sporulation and isolation of desired products. We were not able to obtain tetrads from kin28-AA tbp-AA diploids, and so this strain was constructed by introducing the FRB epitope tag directly into the kin28-AA parent strain using standard methods (Longtine et al., 1998). Similarly, med2∆ med3∆ med15∆ yeast were generated by crossing strains LS01H (med2∆) and LS10 (med3∆ med15∆) harboring a plasmid bearing the URA3 gene, followed by sporulation, tetrad dissection, and identification of med2∆ med3∆ med15∆ haploids. A kin28-AA strain was constructed from one such segregant by integration of the tor1-1 mutation, deletion of FPR1, and chromosomally tagging RPL13A with 2XFKBP and KIN28 with FRB (Haruki et al., 2008).

Strain genotypes are provided in Table S1, and oligonucleotides used in strain construction are listed in Table S2. Loss of viability of anchor away strains in the presence of rapamycin is documented in Figure 3—figure supplement 1. For spot dilution assays, yeast cells were grown to an OD600 1.0 and serial dilutions were spotted on YPD medium and incubated at 30°C for 2–3 days, as indicated in the figure legends.

RNA-seq

RNA was prepared from exponentially growing yeast cultures by the hot phenol method, using two independently derived med2∆ med3∆ med15∆ isolates and two biological replicate samples for wild type yeast (BY4741) (Schmitt et al., 1990). PolyA +RNA was prepared using a Magnetic Bead Isolation Kit (New England Biolabs), yielding from 1.3% to 9% recovery. Library preparation was performed using the NEB Next Ultra RNA Library Prep kit. Sequencing was performed on the Illumina NextSeq platform at the Wadsworth Center, New York State Department of Health.

ChIP-seq

Whole cell extracts (WCE) of AA strains were prepared from 50 mL of culture grown in yeast peptone dextrose (YPD) at 30°C to doubling phase at an OD600 of 0.8. Rapamycin (LC Laboratories, Woburn, MA) was then added one hour prior to crosslinking to a final concentration of 1 µg/mL from a 1 mg/mL stock, stored in ethanol at −20 ˚C for not more than one month. (Concentration of rapamycin stock solutions was determined using A267 = 42 and A277 = 54 for a 1 mg/ml solution.) Immunoprecipitations were performed using the following antibodies: 5.0 µg Pol II unphosphorylated CTD (8WG16, Biolegend, San Diego, CA), 1 μg Rpb3 (Neoclone), 2–5 µg c-Myc epitope (9E10; Roche, Sigma), 2.5 µg protein A (Sigma), 2.5–5 µg TBP (58C9, Abcam, or 5 µL serum, generous gift from A. Weil, Vanderbilt University), and 2.0 µL Taf1 (serum, generous gift from J. Reese and Song Tan, Penn State University). For spike-in experiments, whole cell extract was prepared from S. pombe strain FWP510 (DeGennaro et al., 2013) grown in Yeast Extract Supplemental (YES) medium and added to S. cerevisiae whole cell extract in a ratio of 1:4 to 1:5 prior to antibody incubation. At least two biological replicates (from independent cultures, on different days) were performed for all ChIP-seq experiments except as follows: Single ChIP-seq experiments were performed for tbp-AA yeast against Rpb1, with and without rapamycin; for ChIP against Med2-TAP and Med14-TAP in tbp-AA and rpb3-AA yeast; for Med17-myc in kin28-tbp-AA yeast; and for experiments targeting Rpb1 and Rpb3 (separately) in rpb3-AA yeast, with and without rapamycin. See Table S3 for details.

Library preparation for Illumina paired-end sequencing was performed with the NEBNext Ultra II library preparation kit (New England Biolabs) according to manufacturer’s protocol and barcoded using NEXTflex barcodes (BIOO Scientific, Austin, TX) or NEBNext Multiplex Oligos for Illumina. In some experiments, a size selection step was performed on barcoded libraries by isolating fragment sizes between 200–500 bp on a 2% E-Gel EX agarose gel apparatus (ThermoFisher Scientific). Sequencing was performed on the Illumina NextSeq 500 platform at the Wadsworth Center, New York State Department of Health.

ChIP-Seq and RNA-seq analysis

Unfiltered sequencing reads were aligned to the S. cerevisiae reference genome (Saccer3) using bwa (Seoighe and Wolfe, 1999). Up to one mismatch was allowed for each aligned read. Reads mapping to multiple sites were retained to allow evaluation of associations with non-unique sequences (Seoighe and Wolfe, 1999) and duplicate reads were retained. Binding peaks were identified using SICER (Zang et al., 2009) with the following parameters: effective genome size 0.97 (97% of the yeast genome is mappable,), window size 50 bp, and gap size 50 bp. Calculation of coverage, comparisons between different data sets, and identification of overlapping binding regions were preceded by library size normalization, and were performed with the ‘chipseq’ and ‘GenomicRanges’ packages in BioConductor (Gentleman et al., 2004). Control subtraction was carried out in the following way: coverage (exp)/N1 − coverage (control)/N2, in which ‘exp’ is the data set (in. bam format) to be examined, N1 is the library size of the experimental data (‘exp’), and N2 is the library size of the control. For the spike-in experiments of Figure 2, normalization was performed as described (Orlando et al., 2014). Occupancy profiles were generated using the Integrative Genomics Viewer (Robinson et al., 2011). For metagene analysis, a list of 498 genes exhibiting Mediator ChIP signal upstream of promoters (‘UAS genes’) was downloaded from Jeronimo et al. (2016); SAGA-dominated and TFIID-dominated genes were selected according to Huisinga and Pugh (2004); and genes were sorted according to transcript levels using data from Nagalakshmi et al. (2008).

RNA-seq reads were aligned to SacCer3 using tophat2 (Kim et al., 2013). Non-mRNA species were removed after mapping, and the total number of reads mapping to mRNA species were then calculated and mRNA coverage depth re-normalized. For metagene analysis, including heat maps, reads were normalized against input from kin28-AA yeast (KHW127) unless otherwise noted.

RNA-seq and ChIP-seq reads have been deposited in the NCBI Short Read Archive under project number PRJNA413080.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
    Methods in Yeast Genetics: A Cold Spring Harbor Laboratory Course Manual
    1. D Burke
    2. D Dawson
    3. T Stearns
    (2000)
    New York: Cold Spring Harbor Press.
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70
  71. 71
  72. 72
  73. 73
  74. 74
  75. 75
  76. 76
  77. 77
  78. 78
  79. 79
  80. 80
  81. 81
  82. 82
  83. 83
  84. 84
  85. 85

Decision letter

  1. Alan G Hinnebusch
    Reviewing Editor; Eunice Kennedy Shriver National Institute of Child Health and Human Development, United States
  2. Kevin Struhl
    Senior Editor; Harvard Medical School, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Role of the pre-initiation complex in Mediator recruitment and dynamics" for consideration by eLife. Your article has been reviewed by three peer reviewers and the evaluation has been overseen by a Reviewing Editor, Alan Hinnebusch, and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will is not acceptable for publication in eLife.

The reviewers found intriguing the data obtained from anchor-away of TBP in the kin28-AA background, which seemed to reverse the normal transition of Mediator from UAS to promoter, whereas anchor-away of Rpb3 did not have this effect and resulted instead in loss of Mediator from the promoter. As you have concluded, these results suggest that the transition from UAS to promoter specifically requires TBP, such that Mediator appears stranded at the UAS in TBP's absence; whereas on depletion of Pol II, the TBP-mediated transfer presumably still occurs but Mediator is subsequently lost from the promoter and Mediator occupancy at both UAS and promoter is very low. Your results further suggested that Taf1 might be required along with Pol II to stabilize Mediator at the promoter; although it was felt that additional experiments are required to rule out the possibility that Taf1 is required instead for initial recruitment of Mediator to the UAS. There was also an appreciation of new findings regarding Mediator recruitment that are germane to the functional distinction between TFIID- versus SAGA-dependent genes.

Although the role of the PIC in stabilizing Mediator at promoters is strongly predicted by previous studies, uncovering distinct functions for TBP versus Pol II and Rpb3 in the UAS to promoter transition would represent novel findings. The difficulty is that other interpretations of the existing data are possible, and it was felt that significant additional experiments would be required to confirm the existing findings. One critically important question is whether the different outcomes elicted by depletion of TBP, Rpb3, or Taf1 merely reflect different quantitative degrees of PIC reduction rather than specific effects of depleting particular PIC components. It is important, therefore, to analyze the consequences of depleting each factor on the occupancies of the others, and ideally a fourth PIC component like TFIIB, to determine whether a coherent model can still be proposed in which TBP and Pol II, and Taf1 make distinct contributions to the UAS to promoter transition by Mediator.

There are also substantial differences in the data regarding the behavior of Med15 and Med8 on depletion of TBP versus Rpb3/Taf1, and there is a concern that the Med8 results are not accurately representing the behavior of Mediator head. Thus, additional head subunit(s) would have to be analyzed to confirm the Med8 findings and firmly establish whether or not different Mediator modules are affected differently by the three depletions. Finally, there were also several instances in which results were not presented comprehensively and, thus, doubts emerged about how representative the data being presented might be of the entire data set collected.

In view of the considerable amount of new experimental work that seems necessary, the extended time-frame this would likely entail, and the largely uncertain consequences of the new results for your final model, it was decided that the manuscript should be rejected rather than returned to you for revisions. The following list summarizes the most important criticisms, some of which were already mentioned above.

1) Repeat the Taf1 anchor away experiment in a WT KIN28 strain to determine whether Med15 occupancy at UASs is increased (as in Tbp3 depletion) or decreased, because if the latter was observed, this would implicate Taf1 in the initial recruitment of Mediator to promoters in addition to, or instead of, the transition of Mediator from UAS to PIC.

2) To confirm the differences seen between the Tail Med15 and Head Med18 occupancies on depletion of TBP in the kin28-AA strain in Figure 4C, examine the genome-wide behavior of additional Head subunits.

3) Compare the effects of depleting TBP, Rpb3, or Taf1 on PIC assembly to rule out the trivial possibility that the different outcomes on depletion of each factor represent different extents of reducing PIC assembly rather than indicating distinct functions in Mediator transfer or retention at the promoter. This would also address whether depletion of one factor by anchor away is simply more efficient than another.

4) Results for all four Mediator subunits, for all groups of genes, must be shown for a comprehensive analysis of the effects of depleting TBP vs. Rpb3 vs Taf1.

5) Provide a biological replicate for the results on Med2 in Figure 1A, as well as an input or untagged control.

6) Present time-courses of anchor away to justify the rapamycin treatments chosen for each protein.

Reviewer #1:

Summary of findings:

This study probes the role of PIC assembly, particularly the contributions of TBP, Taf1, and Pol II itself in the transition of Mediator from UAS to promoters in the course of the initiation of transcription in budding yeast. They show that a triple deletion of Mediator tail subunits is viable and allowed appreciable transcription with significant down-regulation of SAGA-dependent genes. In a kin28-AA strain, in which Mediator occupancy at promoters is enhanced, the triple tail deletion reduced Mediator Head occupancies at promoters and essentially abolished it at UAS elements. This finding largely confirms results/conclusions made previously from analysis of a double deletion of Med3/Med15 that the tail module strongly contributes to Mediator recruitment to the majority of genes, including those in which Mediator occupancy can be observed only on depletion of Kin28 and attendant shift of Mediator from UAS to promoter owing to loss of the CDK module, and its stabilization at the promoter owing to loss of Kin28 kinase activity.

They then show that depletion of TBP or Rpb3 actually increases WT Mediator tail subunit occupancy at UAS regions, while not affecting head subunit occupancy, for genes that show strong UAS occupancy of Mediator in WT cells, and also allows detection of Mediator at genes, e.g. RPGs, where it is not normally detected. This provides support for the model proposed by Struhl and Robert that, at genes for which Mediator association with UAS regions is observed, a single Mediator complex bridges the UAS and the promoter, with tail module subunits contacting the UAS and head and middle subunits engaging the promoter region through contacts with Pol II and GTFs; and suggests that interruption of the latter contacts by impairing PIC assembly (by depleting TBP or Rpb1) appears to strand Mediator at the UAS. These results also show that Mediator is actually present at genes (RPGs) where it was previously undetectable, as its transit from UAS to promoter is being blocked by impairing PIC assembly.

They go on to show that depleting TBP in cells where Kin28 is also depleted suppresses the effect of depleting Kin28 alone in shifting Mediator tail occupancy from UAS to PIC, providing evidence that this transition, which has been attributed to loss of the CDK module with attendant enhancement of Mediator interaction with the PIC at promoters, strongly depends on the presence of TBP in the PIC. However, depletion of Rpb3 in the kin28-AA background has a rather different effect-there is some shift from promoter to UAS, but the amount of tail signal (as well as head signal) overall is strongly reduced. This could be interpreted to indicate that Pol II is required for recruitment of Mediator per se, even to UAS elements; however, this possibility is disfavored by the fact that depletion of Rpb3 alone increases UAS occupancy of tail and scaffold subunits (Figure 2). Thus, they interpret the decrease in Mediator occupancy on Rbp3 depletion, seen only in the presence of Kin28 depletion, to result from a failure to retain Mediator at promoters following the shift from UAS to promoters that is evoked by depletion of the CDK module, thus implicating Pol II itself in stabilizing the PIC at the promoter. The shift of some mediator occupancy from promoter to UAS observed on depleting Rpb1 together with Kin28 is being interpreted (I believe) as an indirect result of reduced TBP occupancy on depletion of Rpb3 and attendant reduction in the TBP-dependent shift from UAS to promoter at certain genes; however the reduced TBP occupancy on depletion of Rpb3 should probably be confirmed experimentally if not already demonstrated previously.

Depleting Taf1 together with Kin28 has yet another consequence, of greatly reducing Mediator occupancy at the promoters, without evoking any shift back to the UAS. Thus, it seems that Taf1 could also be required for stabilizing Mediator at promoters and its depletion would have no effect on the shift from UAS to promoter and only on Mediator retention at promoters following the shift evoked by depletion of the CDK module. However, it seems possible that these results could be explained differently if Taf1 contributes to initial recruitment, which I think they are actually suggesting in panel B of the model in Figure 6. I believe that they should be asked to deplete Taf1 alone to determine whether Mediator occupancy at UASs is increased (as in Rpb3 depletion) or decreased, and if the latter is observed, this would implicate Taf1 in the initial recruitment of Mediator to promoters.

General critique:

While some of their results can be viewed as being largely confirmatory of previous findings from the Struhl and Robert labs, other findings increase our understanding of how Mediator is recruited to genes, and the transition of Mediator from the UAS to the promoter. Examination of the triple Mediator tail subunit deletion, in which the tail is completely eliminated provides stronger evidence that obtained previously with just two subunits deleted, that the tail module is important for Mediator recruitment at most active genes; but also reveal a tail-independent pathway that allows appreciable Mediator recruitment and transcription to continue, which they are able to argue involves Mediator recruitment by Pol II. Their findings that TBP is required for the transition of Mediator from UAS to promoter, even when the CDK module is depleted, whereas Pol II is important for retention of Mediator at promoters, and might contribute only indirectly to enhancing the UAS-promoter transition by enhancing TBP occupancy at the promoter, are significant. The role of Taf1 is clearly distinct from that of TBP, but it remains unclear whether Taf1 is enhancing Mediator recruitment to the UAS (like TBP) or stabilizing Mediator at the promoter after the UAS-promoter transition (like Pol II).

To resolve this last issue, the authors should conduct the additional experiment of depleting Taf1 alone to clarify its role in Mediator recruitment vs stabilization at promoters.

Reviewer #2:

The manuscript by Knoll and colleagues describes the effect of PIC components depletion on Mediator occupancy in budding yeast. They show that Mediator association with core promoters depends on the integrity of the PIC. Although this had never been formally demonstrated, this finding was expected based on the bulk of data available on Mediator-PolII and Mediator-GTFs interactions in vitro, including recent structural data. They also show that tampering with the PIC leads to increased detection of Mediator (although not all subunits tested) at UAS. This last finding was less expected and the fact that not all subunits behave the same is intriguing. The experiments are well executed and controlled. As is stands, however, the manuscript only represents a small increment of knowledge relative to current literature. Further exploring the effect of PIC depletion on Mediator-UAS interactions could potentially increase the impact of the work.

Specific comments:

First section entitled "Mediator is recruited to gene promoters in the absence of the tail module triad" is simply confirming data from Struhl and Robert labs. The only difference here is the use of a triple deletion of the triad (instead of a double deletion that leads to the loss of the third component). The data is very nice but constitutes a very small increment over published data.

The second section entitled "Mediator association at UAS regions is stabilized by loss of TBP or Pol II" shows that Mediator (Med2, Med15 and Med14) detection at UAS is increased upon disruption of the PIC by Rpb3 or TBP anchor-away. This observation is interesting and the authors interpret this data in the context of the "transition model" proposed by Struhl and Robert labs and suggest that interfering with the PIC leads to the stabilisation of Mediator at UAS. Unexpectedly, however, no change is observed with the Head subunit Med18. If the whole Mediator was stabilised at UAS as proposed by the authors, then all subunits should be affected. The Med18 data may suggest more complex interpretations. For instance, it may be that after its transient interactions with the PIC, Head/Middle dissociates, leaving the Tail stably associated with the UAS. In subsequent rounds, the Head/Middle would re-attach on the UAS-bound Tail and so on. This section of the manuscript has probably the most potential for novelty, but more experiments are needed. I would also suggest reorganizing Figure 2 (and Figure 4). The data for all subunits tested should be represented the same way. Why are Med18, Med2 and Med14 not fully represented as for Med15? Also, why are Med2 and Med14 mapped on SAGA and TFIID genes while the other subunits are mapped on "UAS genes".

The third and fourth sections "Mediator association with promoters depends on PIC components" and "Depletion of TBP, and to lesser extent Pol II, stabilizes Mediator occupancy at UAS regions" examine the effect of TBP, TAF1 and Rpb3 depletion on Mediator occupancy at promoters and UAS in kin28AA cells. I found the description of this data somewhat confusing and some of the conclusions not to be supported by the data. First, both sections really describe the same data so I do not see why separating them. It would be simpler to go through the data systematically and (as suggested for Figure 2) represent all datasets the same way (Figure 4). Currently, the third section describes TAF1 and Rpb3 and says nothing about TBP. It is not clear why. Second, it seems to me that all three depletions lead to the same phenotype, although with different severity. The TBP depletion is described as having fundamentally different effects on Mediator compared to Rpb3 and TAF1 depletions but one could argue that TBP depletion simply has a more pronounced effect on Mediator as a consequence of a more profound effect on the PIC. The authors should show the effect of their depletions on a few PIC components. One would expect TBP depletion to abolish the PIC altogether, whereas TAF1 and Rpb3 depletions may leave partial PIC behind. Alternatively, the differences in magnitude may be due to variable efficiencies (or timing) of depletion for TBP, Rpb3 and TAF1, a possibility not acknowledged. On a related note, the authors should show the depletion of their AA strains over time (easily done by ChIP-qPCR) in order to justify the time point used and relativize the effect observed for each of them. Third, I do not see what data allows the authors to conclude about timing of events as claimed in their discussion: "[…] TBP being most important for transit of Mediator from UAS to promoter and Rpb3 and Taf1 being more important for stabilizing occupancy at the promoter following transit".

Reviewer #3:

This paper describes a very nice set of genome-wide experiments that demarcate how and where Mediator is recruited to promoters. In one pathway the UAS/Activator complex recruits Mediator via the tail module to the UAS region, and then ultimately it sets down over the PIC. In the second pathway, which is UAS-independent and tail-independent, Mediator is directly recruited to the PIC, which most likely involves TFIID. The data and manuscript were presented clearly, and in general the conclusions were supported by the data. Overall the manuscript represents an important contribution to our understanding of Mediator, and will be valued by the scientific community.

Specific comments

1) It seems that single replicates were performed for some "main figure" experiments, including Figure 1A (Med2 TAP). Please confirm Figure 1A with a second biological replicate for Med2, since this is a main figure and drives the rationale for the rest of the manuscript. Also, is there an untagged/input control for Figure 1A? This would be helpful to get a sense of the recruitment of the tailless mutant over background.

2) There needs to be a correlation coefficient for Figure 1C to get a quantitative measure of difference in the gene expression in mutant relative to WT. "dispensable for most gene expression" would be more convincing with showing the numerical value for the graph.

3) Figure 2B, why isn't occupancy of Med 18 changed with AAR?

4) Anchor away was performed for 1 hr. Please provide an explanation for why this time frame, as opposed to less time like 20-30 min. Obviously, there is a trade-off between completeness of the sequestering and the potential for indirect effects to creep in. Was this condition chosen for a reason or was it arbitrary? Is there GFP localization experiment to verify the anchor away?

5) The results in Figure 4 are striking, and show that when Kin28 is depleted Mediator occupancy at TFIID promoters is highly dependent on Pol II and Taf1. I'm a bit puzzled as to why there needs to be a dependence on Kin28. Or is that a product of the experimental design, in that Taf1 or pol II AAR was not conducted in a WT KIN28 strain? Why are opposite effects seen for Med 15 and Med 18 in tbp+kin28 AAR?

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "Role of the pre-initiation complex in Mediator recruitment and dynamics" for consideration by eLife. Your article has been reviewed by three peer reviewers and the evaluation has been overseen by a Reviewing Editor, Alan Hinnebusch, and Kevin Struhl as the Senior Editor. The reviewers have opted to remain anonymous.

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

While two of the reviewers were satisfied with the additional work and revisions you made to the previous (rejected) version of the manuscript, and considered the results now to be worthy of publication after some minor revisions, reviewer #2 has raised serious concerns about the apparent lack of reproducibility between certain replicates, and a lack of clarity about whether results from two replicates have been used generally to draw key conclusions throughout the work. The recommended approach would be to draw conclusions after combining results from at least two highly correlated replicate datasets. Hence, it is possible that you would be obliged to conduct additional replicate experiments and re-evaluate your findings and conclusions after combining results from only highly correlated datasets. Reviewer #2 has also raised concerns that the AA tag on Taf1 is affecting its function in the absence of Rapamycin, and this complication has to be taken into account in the interpretation of results obtained in the presence of Rapamycin for this mutant, as the tag is setting a baseline of constitutive impairment against which the effect of the drug must be compared. Reviewer #3 has also raised some minor concerns that would also need to be addressed.

Reviewer #1:

The authors have done a considerable amount of additional work to address the major criticisms of the previous version. In particular, they carried out the experiment I had requested of depleting Taf1 in a strain with WT Kin28, and can now rule out a role for Taf1 in Mediator recruitment.

The work is significant for several reasons. Their findings indicating that TBP is required for transit of Mediator from UAS to promoter, whereas Taf1 of TFIID is dispensable for that function but appears to be as important as Pol II itself in retaining Mediator at promoters following the transfer from UAS are both important findings that could not have been predicted from previous work. It is also interesting that the importance of Taf1 and Pol II in retaining Mediator at promoters differs significantly in magnitude between TFIID and SAGA-dominated promoters. The additional work done to create the triple tail subunit deletion and quantify the effects of a completely tail-less mediator on recruitment of Mediator and Pol II occupancy is also valuable, as the effects are more severe than observed for the double tail subunit deletion. They also show unequivocally that the tail is not essential for Mediator recruitment, and that other interactions of Mediator with Pol II, TFIID, or TBP suffice for less efficient recruitment in the absence of the tail. On the other hand, the TBP anchor away experiments show that PIC formation is dispensable for Mediator recruitment in the presence of an intact Mediator tail, as Pol II recruitment is nearly abolished on TBP AA, with a shift in Mediator occupancy to UAS regions indicating that Mediator is being recruited but is stranded at the UAS in the absence of TBP's function in Mediator transit. These latter experiments also reveal that Mediator is recruited to the UASs of many genes where it cannot normally be detected.

I have no additional criticisms.

Reviewer #2:

The manuscript by Knoll et al. describes how yeast Mediator occupancy at UAS and promoters is affected by PIC components. Using combinations of anchor-away experiments coupled to ChIP-seq, they propose that TBP is required for the transit of Mediator from UAS to promoters while Pol II and Taf1 are required for stabilising Mediator at promoters. They also provide data showing that the Tail triad is not required for the association of Mediator with promoters.

This work is building on prior work from the Struhl and Robert lab and several aspects are largely confirmatory. The most novel aspect concerns how different PIC components contribute to the association of Mediator with promoters. Unfortunately, the data appears to be of variable quality and the conclusions are based on interpreting cherry-picked replicates while ignoring others that would have led to very different conclusions.

My understanding is that some experiments were done as only one replicate while others were done in duplicates. The main problem is that the analysis appears to have been done on one of the replicates (which is shown in the main Figures) while the second replicate is (at least in some occasions and it is not clear whether this is systematic) shown as a figure supplement and not used to build the story. Normally, experiments are done at least in duplicate and some correlation metrics are used to insure they are alike. Then replicates are combined and analysed further. Here, not only the correlation between replicates has not been assessed but one is used and the other relayed to a sup figure. Yet, at least in some cases, the replicates are so different that considering the alternate would have led to a completely different story. Here is a striking example: Figure 3A shows that Med15 occupancy is massively increased upon anchor-away of TBP. As the authors emphasised, this is radically different from Med18, the occupancy of which is not affected by the anchoring of TBP (as shown in Figure 3B). Yet, the second replicate of these experiments, shown in Figure 3—figure supplement 3, shows no effect of TBP anchoring on Med15 (rather that a massive increase) and a decrease for Med18 (rather than no change). Similar inconsistencies are also present for the rpb3 and taf1 anchoring on the same sets of Figures. The massive increase of Med15 occupancy upon TBP anchoring being the founding data for the story, I am finding it difficult to consider the rest of the story. In addition to Figure 3, the main results of this paper are those in Figure 5. In this case, I was not able to figure out whether the data presented in Figure 5 was for 1 rep out 1, 1 rep our 2 (in which case rep2 is nowhere to be found) or a combination of 2 reps. Given the variability between reps observed in Figure 3, and not knowing about the replicate status of Figure 5, I am not confident in interpreting data from Figure 5.

Another major concern with the manuscript has to do with the Taf1 anchor-away data. As acknowledged by the authors, the taf1-FRB strain appears to show a phenotype even in the absence of rapamycin. This means that the tagging affected the function of the protein. Because rapamycin did not lead to any addition effect, I do not think that the data can be interpreted as a consequence of conditional depletion. Unless the authors find a way to conditionally affect the function of Taf1 (perhaps auxin degron or tagging FRB in N-terminal would work better), I believe the Taf1 data has to be removed.

In sum, I recommend the authors go back to the basics and make sure their experiments are reproducible. For that, they need to show that their replicates are highly correlated. At current state, I suspect that most of the effects observed are batch effects rather than biological effects. For instance, simply inspecting the red tracks in Figure 3—figure supplement 5A clearly shows that rep2 did not work nearly as well as rep1. Not seeing the labels, one could not even tell how to pair them. Also, looking at the blue tracks from the same Figures suggests that the samples cluster by batches rather than by conditions.

Reviewer #3:

In this revised manuscript, the authors show that mediator recruitment to upstream activated regions (UAS) via transcription activators may be dependent or independent of the tail module. Importantly their triple KO when compared to WT has reduced occupancy of mediator complex (via Med17 ChIP) and PolII (via Rbp1 ChIP) at most genes, but occupancy of both is present in the absence of the Med2, Med3, and Med15. This reinforces the role of the Tail in binding to UAS and that the UAS enhances transcription, but importantly highlighting that even without Med2, there is minor mediator occupancy at UAS under conditions of basal transcription. This point is not merely confirmatory but key to understanding the overall role of mediator in basal and stress-induced transcription not only in yeast, but even in mammals akin to the role of enhancers enhancing transcription through the tail when needed. They show increased occupancy of Med15 (tail) with depletion of Tbp or Rpb3, but this is not noted for Med18 (head) or Med14 (backbone). This may suggest that PIC disruption may lead to Mediator not moving from the UAS region to the promoters (being stalled). Further, they show that Med15 and Med18 occupancy is decreased more at TFIID-dominated genes than SAGA-dominated genes upon Taf1 and Rpb3 depletion in the context of Kin28 anchor away. TBP depletion causes an upstream shift of mediator occupancy at SAGA-dominated genes (corresponding to UAS) and usually a loss of mediator occupancy at TFIID-dominated genes. Rpb3 loss leads to decreased occupancy, particularly at TFIID-dominated genes and an upstream shift of Med15 occupancy at SAGA-dominated genes with concomitant Kin28 depletion.

This paper clearly shows key mediator occupancy differences at SAGA- and TFIID-dominated genes in connection with PIC stabilization that is in line with many other papers. Importantly, they connect the mediator tail module to SAGA-dominated genes via their triple Tail KO. They demonstrate that the movement of the mediator from UAS to promoter is dependent on TBP and that the mediator is stabilized at promotors especially in TFIID-dominated genes by Taf1. While the model of dependent and independent recruitment of mediator by its tail module is plausible, the differentiation of mediator dynamics at SAGA- vs. TFIID-dominated genes is a very exciting aspect of this paper and nicely highlights mediator movement after initial recruitment in the context of PIC components.

Multiple aspects of this paper connect other previously isolated papers such as SAGA & TFIID-dominated genes, UAS and TSS distance in relationship to mediator, and PIC-component interactions with mediator modules; in so doing, they put minor aspects of key mechanisms at the UAS-promoter looping into context of each other to present a clear and more encompassing model. We greatly appreciate them starting to perturb the assumed complex relationship between PIC components and mediator and think their data will enable others to further explore these relationships. Moreover, the authors have sufficiently addressed our concerns from the previous review. We find this paper not to be merely confirmatory but addressing key issues of mediator biology.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Role of the pre-initiation complex in Mediator recruitment and dynamics" for further consideration at eLife. Your revised article has been favorably evaluated by Kevin Struhl (Senior Editor), a Reviewing Editor, and one reviewer.

The manuscript has been improved with the addition of new datasets, but there are several remaining issues raised by reviewer #2 that need to be addressed before acceptance. Please make appropriate revisions of text in response to all of the criticisms/comments in the review below and provide a point-by-point explanation of your responses along with the revised manuscript.

Reviewer #2:

The revised manuscript is improved, notably by the addition of datasets completing some that were done only once and by the replacement of non-consistent replicates. However, there are a few issues that are still of concern and should be addressed.

First, some of the anchor-away strains behave strangely. While this is unlikely to drive erroneous conclusions, it should at least be mentioned. For instance, Kin28-TBP-AA Med15-Myc is very sick according to Figure 3—figure supplement 1. More worrisome, tagging various PIC components with FRB greatly affects Mediator ChIP in the absence of rapamycin. This is evident by looking at several figures, notably in Figure 3 (pay attention to the y-axis for all the no rapa traces (black) or look at the intensities of the no rapa heatmaps). This is also directly shown in Fig 3—figure supplement 2B. The authors use this figure supplement to claim that Taf1-AA leads to increased Mediator ChIP signal in absence of rapamycin but they fail to acknowledge that this is also the case for TBP-AA (the average signal is 2 to 3 fold up compared to non-tag). It is not clear whether these differences are batch differences (due to technical variation between experiments performed on different days) or biological differences (caused by partial loss of function of tagged proteins).

Second, the Med18 ChIP data displayed in Figure 3 and showing that Med18 does not change in TBP-AAR or Rpb3-AAR does not fit the author's model very well. In subsection “Mediator association at UAS regions is stabilized by loss of TBP or Pol II” the authors say "The increase in ChIP signal observed for Med15, but not for Med18, would be consistent with the most immediate contacts with UAS-bound activators occurring through tail module subunits (such as Med15), while ChIP signal for Med18 at UAS regions would depend on protein-protein cross-links, thus lowering the efficiency of ChIP for this subunit". While I agree that this explains the lower ChIP efficiency of Med18 at UAS, it does not explain why, unlike other subunits tested, Med18 signal does not increase at UAS in TBP-AAR and Rpb3-AAR. The observed Med18 ChIP signal is clearly over the UAS so if Mediator is stuck there, this Med18 signal should increase together with other subunits. The fact that it does not (assuming the data is real), suggests a more complicated model. This result should at least be described more critically.

Third, the conclusion that TBP is required for the transition while Rpb3 and Taf1 are stabilising the PIC-associated Mediator is in my opinion a hazardous extrapolation. As explained by the authors in subsection “TBP is required for transit of Mediator from UAS to promoter”, depletion of PIC components has impact on other PIC components. This makes the attribution of specific roles to specific PIC components in Mediator dynamics difficult. Hence, with these limitations in mind, it seems like an over interpretation to say "TBP being critical for transit of Mediator from UAS to promoter, while Pol II and Taf1 stabilize Mediator association at proximal promoters". A more conservative interpretation would be that the PIC is necessary to allow Mediator transiting to the promoter (as shown when PIC is severely compromised; TBP-AA) and that partially formed PICs (without trying to attribute specific roles to subunits or proteins), while allowing the transit, lead to unstable Mediator-PIC association.

These last two points had been raised in previous review rounds but are still not clearly dealt with in the most recent version of the manuscript.

Finally, it is not clear why the authors were expecting a decrease of Mediator at UAS upon depletion of TBP, Rpb3, or Taf1. Isn't an increase what the "dynamic model" proposed by the Struhl and Robert labs predicted?

https://doi.org/10.7554/eLife.39633.028

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

[…] In view of the considerable amount of new experimental work that seems necessary, the extended time-frame this would likely entail, and the largely uncertain consequences of the new results for your final model, it was decided that the manuscript should be rejected rather than returned to you for revisions. The following list summarizes the most important criticisms, some of which were already mentioned above.

1) Repeat the Taf1 anchor away experiment in a WT KIN28 strain to determine whether Med15 occupancy at UASs is increased (as in Tbp3 depletion) or decreased, because if the latter was observed, this would implicate Taf1 in the initial recruitment of Mediator to promoters in addition to, or instead of, the transition of Mediator from UAS to PIC.

We have included Taf1 anchor away experiments as requested; these are included in the revised Figure 3 (formerly Figure 2). These experiments clearly show that there is no decrease in Mediator occupancy at UAS regions upon depletion of Taf1.

2) To confirm the differences seen between the Tail Med15 and Head Med18 occupancies on depletion of TBP in the kin28-AA strain in Figure 4C, examine the genome-wide behavior of additional Head subunits.

We have included new experiments (Figure 5—figure supplement 4) examining the occupancy of two additional Mediator subunits, Med17 from the head module and Med2 from the tail module, after simultaneous depletion of Kin28 and TBP. The results corroborate those obtained with Med15 and Med18.

3) Compare the effects of depleting TBP, Rpb3, or Taf1 on PIC assembly to rule out the trivial possibility that the different outcomes on depletion of each factor represent different extents of reducing PIC assembly rather than indicating distinct functions in Mediator transfer or retention at the promoter. This would also address whether depletion of one factor by anchor away is simply more efficient than another.

We have included data on the effect of depleting PIC components on occupancy of other PIC subunits, and also cited relevant literature on this point (we had cited a bioRxiv paper on this previously, which has now been published in Genes & Development). Specifically, we document depletion of TBP, Rpb3, and Taf1 (which is less complete) in the corresponding anchor away strains in Figure 3—figure supplement 4 and 5. We also show the effects on occupancy of Pol II following depletion of TBP (Figure 3—figure supplement 4D), TBP and Kin28 (Figure 7A), and Taf1 and Kin28 (Figure 7A); on occupancy of Taf1 after depletion of TBP or TBP and Kin28 (Figure 7B); on occupancy of Taf1 after depletion of Rpb3 (Figure 7C); and on occupancy of TBP after depletion of Rpb3, and occupancy of Rpb1 after depletion of TBP (Figure 3—figure supplement 4). The implications of these results on interpreting Mediator occupancy changes after PIC component depletion are considered towards the end of the Results section.

4) Results for all four Mediator subunits, for all groups of genes, must be shown for a comprehensive analysis of the effects of depleting TBP vs. Rpb3 vs Taf1.

The results for the effects on occupancy of four Mediator subunits following depletion of TBP or Rpb3 are shown in Figure 3 (formerly Figure 2) in the revised manuscript, and results for two subunits are shown for Taf1 (all that we have done), as little change was seen for Taf1, and the major point for that PIC component is that recruitment of Mediator is not decreased upon its depletion. These results are now presented in the same way (at “UAS genes”) in all cases.

5) Provide a biological replicate for the results on Med2 in Figure 1A, as well as an input or untagged control.

We conducted a biological replicate for the results in Figure 1A of the original submission, and discovered that the signal we observed for Med2 occupancy was not significantly different to that seen with a control, untagged strain. We therefore have removed this part of Figure 1 from the manuscript.

6) Present time-courses of anchor away to justify the rapamycin treatments chosen for each protein.

We have presented a time course for anchor away of TBP, and also provide justification of our choice of one hour depletion by comparison with other published work using this method to deplete components of the transcription machinery.

In addition, we have included new data showing that Pol II occupancy is reduced genome-wide to about 25% of wild type levels in the med2∆ med3∆ med15∆ mutant, demonstrating that the tail module plays an important role in Pol II occupancy at essentially all genes.

Reviewer #1:

[…] While some of their results can be viewed as being largely confirmatory of previous findings from the Struhl and Robert labs, other findings increase our understanding of how Mediator is recruited to genes, and the transition of Mediator from the UAS to the promoter. Examination of the triple Mediator tail subunit deletion, in which the tail is completely eliminated provides stronger evidence that obtained previously with just two subunits deleted, that the tail module is important for Mediator recruitment at most active genes; but also reveal a tail-independent pathway that allows appreciable Mediator recruitment and transcription to continue, which they are able to argue involves Mediator recruitment by Pol II. Their findings that TBP is required for the transition of Mediator from UAS to promoter, even when the CDK module is depleted, whereas Pol II is important for retention of Mediator at promoters, and might contribute only indirectly to enhancing the UAS-promoter transition by enhancing TBP occupancy at the promoter, are signficant. The role of Taf1 is clearly distinct from that of TBP, but it remains unclear whether Taf1 is enhancing Mediator recruitment to the UAS (like TBP) or stabilizing Mediator at the promoter after the UAS-promoter transition (like Pol II).

To resolve this last issue, the authors should conduct the additional experiment of depleting Taf1 alone to clarify its role in Mediator recruitment vs stabilization at promoters.

We thank the reviewer for a thoughtful and thorough critique, and are pleased that s/he finds our results significant. The reviewer states that the experiments using the med2∆ med3∆ med15∆ are “largely confirmatory” of previous work from the Struhl and Robert labs, but also states that this result “provides stronger evidence than obtained previously”. We agree with both of these comments, mostly; although the effect of the triple deletion is not unexpected, the results establish definitively that the tail module triad is non-essential, contributes considerably to Mediator recruitment at most active genes, and demonstrates that alternative modes of Mediator recruitment and gene activation, independent of the tail module triad, exist. Prior to this work, the possibility existed that the tail module triad was essential, and that residual function of Med2 was sufficient to allow viability in the double mutant; this is now conclusively ruled out. We have also added a new experiment showing that Pol II occupancy is reduced to ~25% of wild type levels in the med2∆ med3∆ med15∆ mutant across the genome.

With regard to the last point, that it is unclear whether Taf1 is important for Mediator recruitment to the UAS or for stabilizing Mediator at the promoter after transit from the UAS, we have included in the revised manuscript experiments that provide evidence that Mediator continues to occupy UAS regions after Taf1 depletion, in the presence of active Kin28. These data support the idea that Taf1 is important for stabilization of Mediator at promoters after transit from the UAS.

Reviewer #2:

[…] First section entitled "Mediator is recruited to gene promoters in the absence of the tail module triad" is simply confirming data from Struhl and Robert labs. The only difference here is the use of a triple deletion of the triad (instead of a double deletion that leads to the loss of the third component). The data is very nice but constitutes a very small increment over published data.

We are pleased that the reviewer considers our data to be solid. We disagree that the first section, regarding the knockout of the complete tail module triad, is “simply confirming data from Struhl and Robert labs.” Our finding that Mediator occupancy at UAS regions in med2∆ med3∆ med15∆ yeast is nearly completely eliminated can fairly be regarded as confirmatory, as a similar result was reported for the double mutant (med3∆ med15∆) and the clear expectation would be that loss of Mediator occupancy at UAS regions would be seen in the triple mutant also. However, it could not be predicted with confidence that the triple mutant would be viable, nor what the effect would be on Mediator occupancy at promoters. Our results provide the first definitive evidence that the Mediator tail module triad is not essential, and that Mediator can be recruited to gene promoters in its absence. We also show in the revised manuscript that association of both Mediator and Pol II is substantially decreased in the triple mutant, considerably more than the less than two-‐fold effect reported previously (Jeronimo et al., 2016), showing that the tail module triad is broadly important for recruitment of these components.

We agree with the reviewer that based on the in vitro results of Carey and co-workers, it seemed likely that Mediator association with gene promoters would depend on PIC integrity. But in vitro results do not always accurately predict in vivo interactions, and this prediction had not previously been tested experimentally.

The second section entitled "Mediator association at UAS regions is stabilized by loss of TBP or Pol II" shows that Mediator (Med2, Med15 and Med14) detection at UAS is increased upon disruption of the PIC by Rpb3 or TBP anchor-away. This observation is interesting and the authors interpret this data in the context of the "transition model" proposed by Struhl and Robert labs and suggest that interfering with the PIC leads to the stabilisation of Mediator at UAS. Unexpectedly, however, no change is observed with the Head subunit Med18. If the whole Mediator was stabilised at UAS as proposed by the authors, then all subunits should be affected. The Med18 data may suggest more complex interpretations. For instance, it may be that after its transient interactions with the PIC, Head/Middle dissociates, leaving the Tail stably associated with the UAS. In subsequent rounds, the Head/Middle would re-attach on the UAS-bound Tail and so on. This section of the manuscript has probably the most potential for novelty, but more experiments are needed. I would also suggest reorganizing Figure 2 (and Figure 4). The data for all subunits tested should be represented the same way. Why are Med18, Med2 and Med14 not fully represented as for Med15? Also, why are Med2 and Med14 mapped on SAGA and TFIID genes while the other subunits are mapped on "UAS genes".

We have remade Figure 2 (Figure 3 in the revised manuscript) so that it shows heat maps and line graphs at “UAS genes” for all four Mediator subunits. Because depletion of Taf1 showed little effect on Med15 and Med18, we did not examine the effect of Taf1 depletion on Med2 and Med14. As to the possibility that the tail module might interact with UASs in the absence of the rest of Mediator, we do mention this in the text but do not favor this possibility. First, the increased occupancy seen for Med14 upon depletion of Rpb3 or TBP (Figure 3D) is consistent with results for Med15 and Med2, and suggests that Mediator as a whole shows increased occupancy under these conditions. We believe this interpretation is most consistent with results from the Struhl and Robert labs, and in particular with the model shown in Figure 7 of Jeronimo et al., 2016. We show in Figure 3—figure supplement 5B that, as in the Jeronimo et al. paper, we observe ChIP signal at some genes after Kin28 depletion at both UAS and promoter regions for Med15, but only at promoter regions for Med17 and Med18. When considered together with the re-ChIP experiments in Petrenko et al., 2016, Figure 1, these results seem most consistent with Med18 exhibiting weaker ChIP signal at UAS regions than does Med15.

The third and fourth sections "Mediator association with promoters depends on PIC components" and "Depletion of TBP, and to lesser extent Pol II, stabilizes Mediator occupancy at UAS regions" examine the effect of TBP, TAF1 and Rpb3 depletion on Mediator occupancy at promoters and UAS in kin28AA cells. I found the description of this data somewhat confusing and some of the conclusions not to be supported by the data. First, both sections really describe the same data so I do not see why separating them. It would be simpler to go through the data systematically and (as suggested for Figure 2) represent all datasets the same way (Figure 4). Currently, the third section describes TAF1 and Rpb3 and says nothing about TBP. It is not clear why. Second, it seems to me that all three depletions lead to the same phenotype, although with different severity. The TBP depletion is described as having fundamentally different effects on Mediator compared to Rpb3 and TAF1 depletions but one could argue that TBP depletion simply has a more pronounced effect on Mediator as a consequence of a more profound effect on the PIC. The authors should show the effect of their depletions on a few PIC components. One would expect TBP depletion to abolish the PIC altogether, whereas TAF1 and Rpb3 depletions may leave partial PIC behind. Alternatively, the differences in magnitude may be due to variable efficiencies (or timing) of depletion for TBP, Rpb3 and TAF1, a possibility not acknowledged. On a related note, the authors should show the depletion of their AA strains over time (easily done by ChIP-qPCR) in order to justify the time point used and relativize the effect observed for each of them. Third, I do not see what data allows the authors to conclude about timing of events as claimed in their discussion: "[…] TBP being most important for transit of Mediator from UAS to promoter and Rpb3 and Taf1 being more important for stabilizing occupancy at the promoter following transit".

We agree that depletion of TBP is likely to have a more profound effect on the PIC than depletion of Rpb3 or Taf1, but we are puzzled by the statement that all three depletions lead to the same phenotype, but with different severity. The data in Figures 4C and D (Figure 5 in the revised manuscript) show that both Med15 and Med18 peaks shift upstream upon depletion of TBP at SAGA-dominated genes and genes having upstream Mediator occupancy in wild type yeast (“UAS genes”); Med15 can be seen to shift upstream at TFIID-dominated genes as well. Depletion of Rpb3 or Taf1 leads to a partial upstream shift or no upstream shift, respectively, but a strong decrease in signal. If TBP depletion simply had a more pronounced effect than depletion of Taf1, for instance, it should result in a stronger decrease and no upstream shift, or Taf1 should give rise to a partial upstream shift. These results indicate that TBP is necessary for Mediator occupancy to shift from UAS to promoter, but cannot determine whether it is sufficient, as Rpb3 occupancy is abolished by loss of TBP (Figure 3—figure supplement 4D and Figure 7A). In contrast, Mediator occupancy does not shift upstream upon depletion of Taf1, and only partially upon depletion of Rpb3, suggesting that these PIC components are not necessary for Mediator to transit from UAS to promoter region. Hence their behavior is qualitatively different. I believe the other two reviewers agree with this point.

We have included new data and citations to show the effect of individual PIC components on each other both with and without Kin28 depletion (Figure 3—figure supplement 4 and Figure 7). As the reviewer suggests, depletion of TBP eliminates Pol II association, while depletion of Rpb3 only causes partial loss of TBP, consistent also with recently published work (Joo et al., 2017). Perhaps surprisingly, Taf1 occupancy was only modestly reduced upon depletion of TBP or Rpb3 (Figure 7). We have added text to the Results that considers the results seen on Mediator occupancy in light of cooperative assembly of the PIC.

With regard to our rationale for using a one hour treatment with rapamycin, we chose this as a reasonable balance that would allow near complete depletion but not allow much time for indirect effects, as reviewer #3 suggests. Our ChIP data for anchoring away TBP indicated that depletion was essentially complete after 30 minutes, and we have now included this experiment as supplemental data. Other studies using anchor away to deplete components of the transcription machinery have used similar times of rapamycin treatment (30 min for Taf proteins (Warfield et al., 2017), 1 hr for Spt7 (Baptista et al., 2017) and for Mediator subunits, TBP, and Rpb1 (Wong and Struhl, 2014; Petrenko, 2016, 2017); 90 min for Mediator subunits (Jeronimo et al. 2016); 120 min for Mediator subunits (Anandhakumar et al., 2016)). Interestingly, the latter study found only about 80% depletion of Mediator subunits after 1 hr of rapamycin treatment, whereas the Struhl laboratory found depletion to background levels in 1 hr (Wong and Struhl, 2014), although the two studies used different criteria to monitor depletion. It is possible that the GFP tags also employed in the former study affected kinetics of depletion, but this has not been rigorously investigated.

Finally, with regard to organization, we prefer to keep the two sections mentioned separate. The first reports the evidence that Rpb3 and Taf1 are required for stable association of Mediator at promoters, while the second reports on the shift to the UAS region seen upon depletion of TBP. These distinct outcomes are best emphasized, in our opinion, by being discussed in separate sections.

Reviewer #3:

[…] 1) It seems that single replicates were performed for some "main figure" experiments, including Figure 1A (Med2 TAP). Please confirm Figure 1A with a second biological replicate for Med2, since this is a main figure and drives the rationale for the rest of the manuscript. Also, is there an untagged/input control for Figure 1A? This would be helpful to get a sense of the recruitment of the tailless mutant over background.

We repeated the experiment of Figure 1A and included an untagged control. The signal seen in this experiment for Med2 at UAS regions was small (as in the original figure), but importantly not different from that seen with the untagged control. We have therefore not included this figure in the revised manuscript. However, we do not (and did not) consider this to be essential as a rationale for examining the complete deletion of the tail module triad. Prior to the work reported here, it remained possible that Med2 could contribute to Mediator recruitment to UAS regions and/or promoters in med3∆ med15∆ yeast; we have now formally excluded this possibility, while also defining rigorously the extent to which the tail module triad contributes to Mediator and Pol II recruitment genome-wide.

With regard to other experiments, in all cases where exact replicate experiments were not performed, we have performed corroborating experiments by doing ChIP against multiple Mediator subunits under the same conditions (e.g. Figure 3), or have relied on published results that corroborate our experiments (e.g. decreased occupancy of TBP upon depletion of Pol II). We have also added new experiments in which occupancy of two additional Mediator subunits, Med2 and Med17, is examined after depletion of Kin28 and TBP. The results are consistent with those observed for Med15 and Med18, with the tail module subunit, Med2, exhibiting greater signal at the UAS following depletion, while the head module subunit, Med17, shows strongly decreased signal relative to depletion of Kin28 alone, but very little upstream signal.

2) There needs to be a correlation coefficient for Figure 1C to get a quantitative measure of difference in the gene expression in mutant relative to WT. "dispensable for most gene expression" would be more convincing with showing the numerical value for the graph.

We have added correlation coefficients to the graphs in question and calculated Fisher’s z-score for the difference between correlation coefficients of the two graphs, yielding a p-value < 10-10. However, we have revised our conclusion regarding the role of the tail module in gene expression based on the new data showing that Pol II association is substantially (about 4‐fold) reduced in the tail module deletion mutant: evidently the tail module is broadly important for Pol II recruitment, but the effect of this reduced recruitment is likely offset by altered RNA decay, as has been observed for other perturbations in PIC components.

3) Figure 2B, why isn't occupancy of Med 18 changed with AAR?

We do not have a definitive answer for this question, but speculate that it may reflect Med18 being less proximate to UAS regions and correspondingly exhibiting a weaker ChIP signal there, as proposed by Jeronimo et al., 2016; see also our response to reviewer #2. Alternatively, Mediator may exist in a configuration that decreases accessibility to Med18 under the conditions of TBP depletion. Reviewer #2 suggests it could reflect loss of the Mediator head module (or perhaps only of specific subunits), but we do not favor this explanation at present as there is little evidence to support it.

4) Anchor away was performed for 1 hr. Please provide an explanation for why this time frame, as opposed to less time like 20-30 min. Obviously, there is a trade-off between completeness of the sequestering and the potential for indirect effects to creep in. Was this condition chosen for a reason or was it arbitrary? Is there GFP localization experiment to verify the anchor away?

We chose 1 hr as a reasonable balance that would allow near complete depletion but not allow much time for indirect effects, as the reviewer suggests. Our ChIP data for anchoring away TBP indicated that depletion was essentially complete after 30 minutes, and we have now included this experiment as supplemental data. Other studies using anchor away to deplete components of the transcription machinery have used similar times of rapamycin treatment (30 min for Taf proteins (Warfield et al., 2017), 1 hr for Spt7 (Baptista et al., 2017) and for Mediator subunits, TBP, and Rpb1 (Wong and Struhl, 2014;Petrenko 2016, 2017); 90 min for Mediator subunits (Jeronimo et al., 2016); 120 min for Mediator subunits (Anandhakumar et al., 2016)). Interestingly, the latter study found only about 80% depletion of Mediator subunits after 1 hr of rapamycin treatment, whereas the Struhl laboratory found depletion to background levels in 1 hr (Wong and Struhl, 2014). It is possible that the GFP tags also employed in the former study affected kinetics of depletion, but this has not been rigorously investigated.

5) The results in Figure 4 are striking, and show that when Kin28 is depleted Mediator occupancy at TFIID promoters is highly dependent on Pol II and Taf1. I'm a bit puzzled as to why there needs to be a dependence on Kin28. Or is that a product of the experimental design, in that Taf1 or pol II AAR was not conducted in a WT KIN28 strain? Why are opposite effects seen for Med 15 and Med 18 in tbp+kin28 AAR?

Depletion of kin28 in the experiment of Figure 4 was used to allow detection of Mediator at promoters of active genes, as shown by the Robert and Struhl labs (Jeronimo and Robert, 2014; Wong and Struhl, 2014). We have edited the text to clarify this point. We did include an experiment in which Pol II was depleted (“pol II AAR”), shown in Figure 2 (now Figure 3), and have added data for Taf1 depletion in the presence of wild type KIN28 to Figure 3.

I believe that what the reviewer is referring to as “opposite effects seen for Med15 and Med18 in tpb + kin28 AAR” is the apparent increased signal at SAGA-dominated genes for Med15 and decrease for Med18. We believe that this reflects Mediator being stranded at the UAS region after TBP depletion; because the contact with Med15 is much closer, ChIP efficiency is high and the signal is strong, whereas the Med18 subunit is more proximate to the promoter in kin28-‐AAR conditions than to the UAS in kin28-tbp-AAR conditions, resulting in decreased ChIP efficiency and lower signal. This interpretation is consistent with the model proposed by the Struhl and Robert labs, and specifically with the detailed structural considerations discussed in Jeronimo et al., 2016. We have added text to clarify this interpretation. See also Figure 3—figure supplement 5B, and our response to reviewer #2.

[Editors' note: the author responses to the re-review follow.]

Reviewer #2:

The manuscript by Knoll et al. describes how yeast Mediator occupancy at UAS and promoters is affected by PIC components. Using combinations of anchor-away experiments coupled to ChIP-seq, they propose that TBP is required for the transit of Mediator from UAS to promoters while Pol II and Taf1 are required for stabilising Mediator at promoters. They also provide data showing that the Tail triad is not required for the association of Mediator with promoters.

This work is building on prior work from the Struhl and Robert lab and several aspects are largely confirmatory. The most novel aspect concerns how different PIC components contribute to the association of Mediator with promoters. Unfortunately, the data appears to be of variable quality and the conclusions are based on interpreting cherry-picked replicates while ignoring others that would have led to very different conclusions.

We felt this reviewer was unduly harsh in accusing us of “interpreting cherry-picked replicates while ignoring others that would have led to very different conclusions”. This statement is based on the replicate data shown in Figure 3—figure supplement 3 in the critiqued manuscript, which included data (high signal for both Med15 and Med18 in the TBP anchor away strain in the absence of rapamycin) that was inconsistent with data in the main figure. In the text of the previously submitted version we argued that the preponderance of our results were most consistent with the interpretation we provided; it is not in our opinion accurate to state that we ignored the inconsistent results. Nonetheless the reviewer’s concern regarding reproducibility is valid and was a concern to us as well; that is why we chose to include replicate data for this experiment as a supplementary figure. We have now repeated this experiment—ChIP-seq of Med15 and of Med18 in the tbp-AA yeast strain in the absence and presence of rapamycin—and the results are consistent with those we had presented in Figure 3, including results for ChIP against Med2 and Med14 (see Author response image 1). We have revised Figure 3 by presenting data for Med15 and Med18 ChIP averaged from two independent biological replicate experiments (except for the Med18 ChIP in the taf1-AA strain, for which we have only one replicate).

Author response image 1
Replicates used for graphs and heat maps of Figure 3.

Replicate data was combined except for taf1-AA and taf1-AAR for Med15, as the second ChIP experiment failed.

https://doi.org/10.7554/eLife.39633.024

Reviewer #2 was also concerned about whether data in Figure 5 was based on replicate experiments. The figure as presented in the critiqued manuscript included combined data for two experiments that were consistent with two additional replicate experiments for both Med15 and Med18 ChIP in kin28-tbp-AA yeast (four replicates in total), and data for single experiments that were consistent with replicate experiments for Med15 and Med18 ChIP in kin28-AA yeast. The reason for combining data only for kin28-tbp-AA yeast is that we obtained fewer reads for those experiments, and combining data yielded better signal to noise. For Med18 in kin28-taf1-AA and kin28-rpb3-AA strains, we showed data from single experiments that were qualitatively consistent with replicate experiments. However, the extent to which Med18 signal was affected in the kin28-taf1-AA mutant differed quantitatively in these replicate experiments. We have now performed an additional replicate of this experiment. All three replicates are qualitatively consistent (see Author response image 2): depletion of Kin28 together with Taf1 decreases signal at TFIID-dominated genes, and does not substantially shift Mediator peaks at UAS genes as observed for kin28-tbp-AA yeast (see Author response image 2). The effect of depletion of Kin28 and Taf1 on Med18 occupancy at SAGA-dominated and UAS genes (which are mostly SAGA-dominated) was variable in our three replicate experiments with regard to peak magnitude relative to depletion of Kin28 alone, although it was always a less pronounced decrease than seen for TFIID-dominated genes. We have modified Figure 5 to use the replicate experiment that shows intermediate changes in signal intensity, and make a note of this variability in the figure legend of the revised manuscript. Finally, only single replicates had been performed for Med15 ChIP using kin28-rpb3-AA and kin28-taf1-AA yeast. We have repeated these experiments and obtained consistent results in kin28-rpb3-AA yeast; however, we have seen stronger signals for Med15 in kin28-taf1-AA yeast in the presence of rapamycin than in the experiment, which was used for Figure 5 in the critiqued version (see Author response image 2). Again, we see an upstream shift upon depletion only with TBP and not with Taf1 or Rpb3, and strong decrease in signal at TFIID-dominated genes (see revised Figure 5). We have replaced the data for Med15 ChIP in kin28-taf1-AA in Figure 5 with data from one of the new replicates (the “run40” replicate shown in Author response image 2), to reflect the stronger signal we see for kin28-taf1-AAR relative to kin28-AAR in the new replicates. We have not used combined data for Figure 5 (except as noted above), as the signal to noise and number of reads has varied among replicates, and the combined data is dominated by the “cleanest” data. Information on replicate experiments provided as Supplementary file 3 has been updated to reflect these additional replicate experiments.

Author response image 2
Replicate data for Figure 5.
https://doi.org/10.7554/eLife.39633.025

The reviewer also states that experiments are normally done in duplicate (at least) and some correlation metrics used to insure they are alike. We have found that use of correlation coefficients between ChIP-seq datasets is not informative, as variance in regions outside of the peaks that we focus on (the large majority of the genome) obscures the similarities or differences among the regions of interest in terms of quantitative metrics. Rather, we have used metagene analysis, as in Figures 3 and 5 and the Author response image 1 and 2 to assess similarity between replicate experiments.

Another major concern with the manuscript has to do with the Taf1 anchor-away data. As acknowledged by the authors, the taf1-FRB strain appears to show a phenotype even in the absence of rapamycin. This means that the tagging affected the function of the protein. Because rapamycin did not lead to any addition effect, I do not think that the data can be interpreted as a consequence of conditional depletion. Unless the authors find a way to conditionally affect the function of Taf1 (perhaps auxin degron or tagging FRB in N-terminal would work better), I believe the Taf1 data has to be removed.

The reviewer states that “addition of rapamycin did not lead to any additional effect”, but this is not correct: although adding rapamycin did not affect the ChIP signal for Med15 or Med18 (Figure 3), it did cause lethality (Figure 3—figure supplement 1) and reduced Taf1 occupancy (Figure 3—figure supplement 3 in the revised manuscript); furthermore, anchoring away Taf1 together with Kin28 suppresses the effect of anchoring away Kin28 at TFIID-dominated genes (Figure 5). We therefore believe that the data presented support the main conclusion of experiments on the taf1-AA strain, which is that Mediator association with UAS regions does not decrease upon depletion of Taf1, and therefore that Taf1 is not required for Mediator recruitment to UAS regions. Furthermore, we have observed in three independent experiments that depletion of Taf1 together with Kin28 results in preferential decrease in Mediator occupancy at TFIID-dominated genes. However, the effect of this double depletion on Med15 occupancy at SAGA-dominated genes has been variable in our experiments, as mentioned above, and so we have tempered our conclusions regarding this specific effect.

In sum, I recommend the authors go back to the basics and make sure their experiments are reproducible. For that, they need to show that their replicates are highly correlated. At current state, I suspect that most of the effects observed are batch effects rather than biological effects. For instance, simply inspecting the red tracks in Figure 3—figure supplement 5A clearly shows that rep2 did not work nearly as well as rep1. Not seeing the labels, one could not even tell how to pair them. Also, looking at the blue tracks from the same Figures suggests that the samples cluster by batches rather than by conditions.

We have revised this Figure (now Figure 3—figure supplement 4) so that tracks are normalized across samples for each strain (e.g., all four tbp-AA strain scans are normalized) rather than only between samples within a single experiment (e.g. tbp-AA-1 and tbp-AAR-1 samples were normalized relative to one another but separately from tbp-AA-2 and tbp-AAR-2 samples). This shows more clearly the reduced signal of TBP upon addition of rapamycin.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Reviewer #2:

The revised manuscript is improved, notably by the addition of datasets completing some that were done only once and by the replacement of non-consistent replicates. However, there are a few issues that are still of concern and should be addressed.

First, some of the anchor-away strains behave strangely. While this is unlikely to drive erroneous conclusions, it should at least be mentioned. For instance, Kin28-TBP-AA Med15-Myc is very sick according to Figure 3—figure supplement 1. More worrisome, tagging various PIC components with FRB greatly affects Mediator ChIP in the absence of rapamycin. This is evident by looking at several figures, notably in Figure 3 (pay attention to the y-axis for all the no rapa traces (black) or look at the intensities of the no rapa heatmaps). This is also directly shown in Figure 3—figure supplement 2B. The authors use this figure supplement to claim that Taf1-AA leads to increased Mediator ChIP signal in absence of rapamycin but they fail to acknowledge that this is also the case for TBP-AA (the average signal is 2 to 3 fold up compared to non-tag). It is not clear whether these differences are batch differences (due to technical variation between experiments performed on different days) or biological differences (caused by partial loss of function of tagged proteins).

We have edited the text to note the effect of the tbp-FRB allele on Mediator occupancy and the slow growth phenotype of the kin28-tbp-AA Med15-myc strain (Results section). The elevated Med15 occupancy observed in taf1-AA yeast in the absence of rapamycin was seen in two out of two replicate experiments, while lower levels of Med15 occupancy, similar to the control lacking any FRB tag, was seen in two out of two experiments with rpb3-AA yeast and two out of three experiments in tbp-AA yeast. One experiment showed elevated Med15 occupancy in tbp-AA yeast in the absence of rapamycin; that is the experiment that led to the request for an additional replicate, which showed low occupancy in the absence of rapamycin in tbp-AA yeast that increased upon rapamycin treatment. We have also edited Figure 3—figure supplement 2 by replacing the data for the “no FRB tag” control with data from a newer replicate (from October this year). The original data had higher levels of artifactual signal over ORF regions (as we and others often see), which made it more difficult to visually compare signal strength over the UAS regions.

Second, the Med18 ChIP data displayed in Figure 3 and showing that Med18 does not change in TBP-AAR or Rpb3-AAR does not fit the author's model very well. In subsection “Mediator association at UAS regions is stabilized by loss of TBP or Pol II” the authors say "The increase in ChIP signal observed for Med15, but not for Med18, would be consistent with the most immediate contacts with UAS-bound activators occurring through tail module subunits (such as Med15), while ChIP signal for Med18 at UAS regions would depend on protein-protein cross-links, thus lowering the efficiency of ChIP for this subunit". While I agree that this explains the lower ChIP efficiency of Med18 at UAS, it does not explain why, unlike other subunits tested, Med18 signal does not increase at UAS in TBP-AAR and Rpb3-AAR. The observed Med18 ChIP signal is clearly over the UAS so if Mediator is stuck there, this Med18 signal should increase together with other subunits. The fact that it does not (assuming the data is real), suggests a more complicated model. This result should at least be described more critically.

We agree with this criticism, and have modified the text accordingly. We did not make our point clearly, which is that under some circumstances, such as depletion of Kin28 (Figure 3—figure supplement 4B, and also Jeronimo et al., 2016), Med15 can be detected at UAS regions while Med18 gives rise to virtually no ChIP signal at those sites; in the same sample, signals from both Med15 and Med18 are easily discerned at the proximal promoter. This suggests that Mediator may adopt a configuration in which Med15 signal is robust and Med18 undetectable, or nearly so; such a configuration could result in increased Med15 signal with no increase in Med18 signal. The way we presented this before was poorly constructed at best, and we thank the reviewer for noticing this.

Third, the conclusion that TBP is required for the transition while Rpb3 and Taf1 are stabilising the PIC-associated Mediator is in my opinion a hazardous extrapolation. As explained by the authors in subsection “TBP is required for transit of Mediator from UAS to promoter”, depletion of PIC components has impact on other PIC components. This makes the attribution of specific roles to specific PIC components in Mediator dynamics difficult. Hence, with these limitations in mind, it seems like an over interpretation to say "TBP being critical for transit of Mediator from UAS to promoter, while Pol II and Taf1 stabilize Mediator association at proximal promoters". A more conservative interpretation would be that the PIC is necessary to allow Mediator transiting to the promoter (as shown when PIC is severely compromised; TBP-AA) and that partially formed PICs (without trying to attribute specific roles to subunits or proteins), while allowing the transit, lead to unstable Mediator-PIC association.

These last two points had been raised in previous review rounds but are still not clearly dealt with in the most recent version of the manuscript.

We believe that our statement that TBP is critical for transit of Mediator from UAS to promoter is logically accurate. It does not imply that other factors, in this case other PIC components that depend on TBP for stable association with promoters do not contribute. As we state in subsection “TBP is required for transit of Mediator from UAS to promoter”, we can infer that TBP is necessary for Mediator transit to promoter regions under conditions of Kin28 depletion, but not whether it is sufficient. In contrast, the evidence presented indicates that under conditions of Kin28 depletion, neither Taf1 nor Pol II are critical for this transit. However, the reviewer’s summary statement does provide a succinct summary of these points, and we have edited the text to borrow his/her phrasing (hopefully with permission).

Finally, it is not clear why the authors were expecting a decrease of Mediator at UAS upon depletion of TBP, Rpb3, or Taf1. Isn't an increase what the "dynamic model" proposed by the Struhl and Robert labs predicted?

We had initially anticipated that PIC depletion might reduce Mediator occupancy at both promoters and UAS regions. This could occur, for example, if Mediator contacted both UAS and promoter via DNA looping and reduced PIC occupancy destabilized Mediator occupancy at both regions. This was evidently not the right way to think about it, but it still seems to me not a totally far-fetched possibility in the absence of data. We have edited the text to clarify this point. Clearly what we anticipated was not as important in any event as what we observed.

https://doi.org/10.7554/eLife.39633.029

Article and author information

Author details

  1. Elisabeth R Knoll

    Department of Biomedical Sciences, School of Public Health, University at Albany, Albany, United States
    Contribution
    Conceptualization, Formal analysis, Validation, Investigation, Visualization, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1083-6472
  2. Z Iris Zhu

    Computational Biology Branch, National Center for Biotechnology Information, National Library of Medicine, Bethesda, United States
    Contribution
    Data curation, Formal analysis, Visualization, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Debasish Sarkar

    Wadsworth Center, New York State Department of Health, Albany, United States
    Contribution
    Validation, Investigation
    Competing interests
    No competing interests declared
  4. David Landsman

    Computational Biology Branch, National Center for Biotechnology Information, National Library of Medicine, Bethesda, United States
    Contribution
    Formal analysis, Supervision, Funding acquisition, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9819-6675
  5. Randall H Morse

    1. Department of Biomedical Sciences, School of Public Health, University at Albany, Albany, United States
    2. Wadsworth Center, New York State Department of Health, Albany, United States
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Writing—original draft, Writing—review and editing
    For correspondence
    randall.morse@health.ny.gov
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0000-8718

Funding

National Science Foundation (MCB1516839)

  • Elisabeth R Knoll
  • Debasish Sarkar
  • Randall H Morse

National Institutes of Health (Intramural program)

  • Z Iris Zhu
  • David Landsman

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

Acknowledgements

This work was supported by grant MCB1516839 to RHM, and was also funded in part by the NIH Intramural Research Program at the National Library of Medicine (ZIZ and DL). We are grateful to Todd Benziger for help with strain construction. We thank Kevin Struhl, Francois Robert, and Fred Winston for yeast strains and reagents; Todd Gray, Tony Weil, Song Tan, and Joe Reese for generous provision of antibodies; and Joe Wade, Fred Winston, and Ameet Shetty for helpful discussions. We gratefully acknowledge help from the Wadsworth Center Applied Genomics Technology and Tissue Culture and Media Cores.

Senior Editor

  1. Kevin Struhl, Harvard Medical School, United States

Reviewing Editor

  1. Alan G Hinnebusch, Eunice Kennedy Shriver National Institute of Child Health and Human Development, United States

Publication history

  1. Received: June 28, 2018
  2. Accepted: December 12, 2018
  3. Accepted Manuscript published: December 12, 2018 (version 1)
  4. Version of Record published: January 7, 2019 (version 2)

Copyright

© 2018, Knoll et al.

This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.

Metrics

  • 986
    Page views
  • 205
    Downloads
  • 0
    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)

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

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

Further reading

    1. Biochemistry and Chemical Biology
    2. Chromosomes and Gene Expression
    Jason Saba et al.
    Research Article Updated
    1. Chromosomes and Gene Expression
    2. Genetics and Genomics
    Ashley Tehranchi et al.
    Research Communication