Early observations demonstrated that SFP1, even though not essential for viability, it is one of the strongest regulators of cell size amongst a collection of over 4,000 viable deletion strains (Jorgensen et al., 2002). Altering SFP1 expression level, either by deletion or overexpression, leads to cell size reduction or increase, respectively ((Jorgensen et al., 2002; Lempiainen et al., 2009); Figure1). Given the primary structure of Sfp1 and its nuclear localization under optimal growth conditions (Fingerman et al., 2003; Marion et al., 2004), a logical explanation for the Sfp1-mediated cell size effect is that it is a transcription factor that regulates the expression of genes involved in growth. Indeed, DNA-microarray experiments (Jorgensen et al., 2002) demonstrated that genes required for ribosome production and protein translation are Sfp1 transcriptional targets. The first group includes both ribosomal protein genes themselves and the large suite of genes required for ribosome biogenesis, referred to as RiBi genes. ChIP experiments have detected Sfp1 at the promoters of some of these genes (i.e.
RPGs), further supporting its role as a positive regulator of growth-related gene expression (Jorgensen et al., 2002; Lempiainen et al., 2009; Marion et al., 2004). An interesting and still unresolved paradox is that Sfp1 ChIP signal is indistinguishable from the background when measured at the promoters of most of its major targets (Jorgensen et al., 2004), the RiBi genes. This suggests that Sfp1 might be a peculiar transcription factor that can affect gene expression without a direct DNA interaction and using different mechanisms for different gene classes. To further elucidate how Sfp1 influences gene expression and to possibly identify a molecular mechanism for its action, we performed a ChIP-sequencing (ChIP-seq) experiment aimed at characterizing Sfp1 chromatin binding on a genome-wide level. This was followed by a ChIP-seq analysis of RNAPII binding under normal or elevated SFP1 expression, or upon conditional nuclear depletion of Sfp1.
Finally, preliminary evidence suggested a role of Sfp1 in RNAPIII transcriptional control (Lempiainen et al., 2009). During this thesis work, we tried to further study this hypothesis and to establish if Sfp1 controls growth by coordinating RNAPII and RNAPIII transcriptional output.
Figure 1. Sfp1 strongly affects cell size
A. Deletion of SFP1 (YHL108) decreases colony size and causes size adaptation to different carbon sources to be less pronounced compared to WT. B. SFP1 overexpression (YST50) increases cell mean volume approximately after 2 generations, as shown by the growth curves.
Sfp1 regulates gene expression by controlling RNAPII recruitment
As previously described, deletion of SFP1 (sfp1Δ) causes a strong defect on the expression of RPGs and RiBis as well as other gene classes mainly involved in protein translation (Jorgensen et al., 2002; Jorgensen et al., 2004). On the contrary, RPG and RiBi genes transcripts strongly increased upon SFP1 overexpression. We decided to begin our analysis of Sfp1’s mechanism of action by determining whether the altered expression observed at the mRNA level for Sfp1 targets is indeed a consequence of decreased RNAPII binding at these genes. RNAPII recruitment was analyzed by ChIP on a subset of genes using a specific antibody for the large core subunit Rpb1. Two different strategies were developed: SFP1 overexpression was obtained by replacing the endogenous SFP1 promoter with the strong, inducible GAL1 promoter as described previously ((Jorgensen et al., 2002); Figure 2A). In a second approach, designed to reveal the immediate effect of Sfp1 removal from the nucleus, I took advantage of the “away” (AA) system (Haruki et al., 2008). Briefly, the anchor-away technique depletes the nucleus of a protein of interest (the target; Sfp1 in this study) by conditional tethering it to an abundant cytoplasmic protein (the anchor; Rpl13 in this study) that shuttles through the nucleus during ribosome assembly. This is obtained by C-terminal tagging the target protein with theFKBP12-rapamycin-binding (FRB) domain of human mTOR and the anchor protein with the human FK506 binding protein (FKBP12; Figure 2C). The rapamycin-dependent heterodimerization of the FRB domain and the FKB12 protein causes the nuclear target protein to leave the nucleus and accumulate in the cytoplasm. An Sfp1 anchor-away strain was tested by a growth assay on rapamycin-containing plates and by microscopic observation of Sfp1-FRB-GFP protein to assess Sfp1 subcellular redistribution after rapamycin treatment (Figure 2B; Yvonne Gloor unpublished results).
Figure 3A shows the results of Rpb1 ChIP experiments for two selected RP and RiBi, genes, RPL1B and NSR1, respectively, following either overexpression or cytoplasmic anchoring of Sfp1-FRB. The probes used for RT-PCR (listed in Table 2) amplify a region that is in close proximity to the ATG of the selected ORF (Figure 3B), where the ChIP signal for Rpb1 is found
to be higher (Figure 3C). Compared to WT, SFP1 overexpression causes ~7-fold increase in RNAPII binding on the NSR1 promoter after 30 minutes of galactose treatment, whereas a
~2.3-fold increase is observed for RPL1B (Figure 3A, right panel). RNAPII binding increases for both genes during the entire time course of the experiment. We noticed that for both genes RNAPII binding is lower at the time “zero”: this is the consequence of the absence of galactose in the medium (cells are grown in raffinose for two generations), which causes the GAL1 promoter to be inactive, or expressed at a very low level (Figure 2A). These observation are consistent with those published previously demonstrating elevated RP and RiBi gene transcripts levels as a consequence of SFP1 overexpression (Jorgensen et al., 2002).The opposite effect is observed when Sfp1 is removed from the nucleus upon rapamycin treatment:
for both the RPG and RiBi gene promoters analyzed, the binding of RNAPII was negatively affected, with a much stronger effect observed for NSR1 (Figure 3A left panel). Interestingly, already after 5 minutes of rapamycin treatment the binding of RNAPII on NSR1 promoter is almost completely abolished, whereas binding on RPL1B is affected with much slower kinetics. We have shown here for the first time that SFP1 overexpression as well as Sfp1 anchor-away has a rapid effect on RNAPII binding that could be the consequence of direct regulation of RNAPII recruitment by Sfp1.
Figure 2. Altering level of SFP1 expression: different strategies
A. SFP1 overexpression was obtained by the GAL1 promoter upon addition of galactose (YHL38; YHL201). The top graph shows SFP1 mRNA measurement at different time points from galactose addition, whereas Sfp1 protein level is shown at the bottom panel. To be noticed the different time scales used for the two panels. B. Sfp1 Anchor-away strains characterization (HHY168; YYG843; YYG844). Microscopy observation revealed that after 15 minutes upon Rapamycin addition, the nucleus is completely depleted of Sfp1 (Top panel). As expected nuclear depletion of Sfp1 causes a growth defect (Bottom panel;
Yvonne Gloor unpublished results). C. Picture adapted from Haruky et al 2008. Schematic description of the “Anchor-away” technique.
Figure 3. Defective binding of RNAPII on two selected Sfp1 targets
A. Conditional nuclear depletion of Sfp1 (Left panel) or its overexpression (Right panel) causes decreased RNAPII binding or its increase respectively onto the promoter of RPL1B and NSR1 selected Sfp1 targets, as measured by ChIP. B. Schematic representation of the DNA regions amplified by RT-PCR using the probes listed on table 2. C. Rpb1 ChIP signal is higher when measured in the ATG region compared to the promoter. RPL1B gene is used here as example. Fold enrichment calculated using Y’ARS region as an internal control.
Figure 4. The auxin-induced degron system
A. Picture adapted from Nishimura et al., 2009. Schematic representation of the AID system: in presence of auxin, the AID-tagged protein is recruited to the SCF-TIR1 complex that through the E2 ligase causes poly-ubiquitination and therefore degradation of the target protein. B. Characterization of Sfp1-AID strain: growth assay in presence of auxin causes the expected growth defect of strains deleted for Sfp1 (Top panel). Already after 5 minutes form auxin addition, Sfp1 is completely degraded (bottom panel). C. Sfp1 degradation causes a reduction in RNAPII binding on NSR1 and RPL1B gene promoters as shown by ChIP. Fold enrichment calculated using Y’ARS region as an internal control.
Given the fact that the TORC1 kinase was found to directly interact with and phosphorylate Sfp1 (Lempiainen et al., 2009), we wanted to confirm the strong effect observed on RNAPII binding in the Sfp1 anchor-away strain by using a different strategy. (Nevertheless, we note that the anchor-away system contains the rapamycin-insensitive TOR1-1 allele and should in principle not be subject to any of the normal physiological effects of rapamycin.) For this purpose we took advantage of the “Auxin-Inducible Degron” (AID) system (Nishimura et al., 2009). The AID system is based on the plant hormone auxin that in plant cells induces the ubiquitin-dependent degradation of the AUX/IAA family of transcription repressors by a specific form of the SCF E3 ubiquitin ligase, SCF-TIR1 (Dharmasiri et al., 2005; Kepinski and Leyser, 2005). Briefly, when TIR1 from Oryza sativa is ectopically expressed in yeast using the ADH1 promoter, it is able to form the SCF-TIR1 complex through an interaction with the highly conserved Skp1 F-box protein (Nishimura et al., 2009). Tagging the protein of interest (Sfp1 for this study) with the Iaa17 epitope (atAXR3 codifying for one component of the AUX/IAA family, hereafter called AID) in a strain that ectopically expresses osTIR1, causes proteasome-dependent degradation of this protein upon auxin addition (Figure 4A). We therefore generated a strain where Sfp1 could be conditionally degraded by auxin treatment and initially analyzed Sfp1 degradation by WB using an antibody that recognizes the AID tag (Figure 4B bottom panel). Only after 5 minutes following auxin treatment the WB signal corresponding to Sfp1-AID tagged protein had completely disappeared. However, to confirm that the full-length fusion protein was degraded (and not just the AID tag) a viability assay in the presence of auxin was performed. Sfp1-AID cells grew poorly in presence of auxin, suggesting that indeed Sfp1 is degraded even though we cannot confirm that its degradation is as fast as the AID tag (Figure 4B). ChIP experiments were then performed using the Sfp1-AID strain to measure the effects of Sfp1 degradation on RNAPII binding. As shown in Figure 4C, the obtained results resemble the effect already described for the anchor-away system, confirming that removing Sfp1 from the nucleus either by degradation or cytoplasmic tethering, causes a strong transcriptional defect, at least on the genes analyzed.
Genome-wide analysis reveals Sfp1 binding at both promoters and gene bodies
In parallel to the described RNAPII binding, the Sfp1-DNA interaction at the RPG and RiBi genes was analyzed by ChIP (YHL38; Table 1). The experiment was performed using standard growth conditions in WT cells (logarithmically growing population in the presence of glucose) and upon SFP1 overexpression, driven by the GAL1 promoter (YHL201; Table 1).
Briefly, cells were pre-grown in raffinose containing medium for two generations and consequently treated with galactose for the indicated time points (Figure 5A). Confirming published results, Sfp1-TAP ChIP signal on RPL1B promoter is significant when compared to the untagged control immunoprecipitate (IP). Moreover, when the RPL1B ORF is considered, Sfp1-TAP ChIP signal slightly exceeds the background, suggesting the presence of Sfp1 on RPG ORFs at least under optimal growth conditions (YPAD). On the other hand, Sfp1-TAP ChIP signal on NSR1 is undetectable or only slightly above the background (Figure 5A). This observation is puzzling, since NSR1 transcription is strongly affected by SFP1 overexpression or depletion. However, we show here that Sfp1-TAP ChIP signal on the NSR1 promoter increases above the background after SFP1 overexpression and it becomes significantly stronger on the ORF (Figure 5A).
These results prompted us to analyzed the genome wide distribution of Sfp1 by ChIP seq (Johnson et al., 2007) using the same strategies to alter SFP1 expression and Sfp1 subcellular localization (pGAL1-mediated SFP1 overexpression and Sfp1 nuclear depletion by the anchor-away system, respectively). Briefly, the genome-wide localization of Sfp1-TAP (YHL38) was analyzed under standard growth conditions (exponentially growing culture in YPAD medium) and the untagged WT strain was used as control (YDS2; Table 2). In addition, YHL38 and YHL201 (Gal1p-SFP1-TAP) were grown in raffinose-containing medium for two generations and subsequently treated for 1 hour with 2% galactose to induce SFP1 expression. Sfp1-TAP was analyzed under these conditions using TAP-specific antibody (Figure 6A). The reads obtained upon sequencing, and all of the mapping parameters, are
described in Materials and Methods. As an example, sequencing results (reads mapped to the genome) from Sfp1-TAP ChIP for chromosome III are represented (Figure 7A).
Figure 5. Sfp1-DNA interaction measured by ChIP on some selected gene targets A. Tap ChIP experiment performed using YHL38 and YHL201 strains that allow Sfp1-Tap ChIP signal to be measured during normal growth conditions (glucose) and upon SFP1 overexpression (GAL1-Sfp1-Tap). The represented ChIP signals are normalized to the background signal obtained by treating the untagged strain with the TAP antibody (YDS2;
“glucose Sfp1” into the graph). B. Schematic representation of the DNA regions amplified
Figure 6. Genome-wide analysis of Sfp1 and RNAPII distribution: experimental setup A. YHL38 was grown in YPAD and YHL38 and YHL201 (Table 2) were grown in YP plus 2% raffinose. During logarithmic phase of growth 2% galactose was added to induce SFP1 expression and cell were harvested after 1 hour for crosslink and immunoprecipitation.
YHL38 YPAD grown cells were collected after 2 generations without any treatment to measure Sfp1 and Rpb1 genome-wide distribution under standard growing conditions. B.
For the “Anchor-away” strategy, (HHY168; YYG843; Table 2) cells where treated with rapamycin after 2 generations of growth in YPAD medium and collected upon 5, 20 and 60 minutes from the treatment for crosslink and immunoprecipitation using an Rpb1 specific antibody. Time lines are not in scale.
Figure 7. Analysis of Sfp1-TAP genome-wide occupancy by ChIP-Seq
A. Mapped sequencing reads on chromosome III for Sfp1-TAP are represented. The X-axis shows chromosome III from 5’ to 3’ end with base pair numbers indicated above. The height on the Y-axis represents the number of reads (400 is fixed as maximum). From top to bottom, samples are ordered as follows: Untagged, GAL1p-SFP1-TAP and SFP1-TAP strains. Growth condition used for each sample is listed on the right part of the image. Red arrows indicates picks obtained by Tap immunoprecipitation of chromatin from the untagged strain and therefore they are non-specific.
We started our ChIP-seq results analysis by categorizing genes according to Sfp1 binding.
The Sfp1 ChIP signal obtained from the Sfp1-TAP sample (YHL38) grown in the presence of glucose was normalized to the signal obtained from the untagged strain grown in the same conditions (YDS2 treated with Tap antibody). Sfp1 was considered to be significantly bound on a specific region (gene promoter or gene ORF) when subtraction of the obtained “Sfp1 ChIP-reads” to the reads obtained using the untagged strain, was > 100 for promoters or > 25 for ORFs ((ReadsSfp1-Tap – ReadsUntag) ≥ 100 or 25)). According to this arbitrary threshold, we were able to observe that Sfp1 does not display significant binding on the vast majority of yeast genes (4488) under standard growth conditions. Analysis of the remaining genes reveals that Sfp1 is not exclusively found on gene promoters, but, to our surprise, can be significantly bound to gene ORFs. We have considered here as a promoter the 400 bp region upstream of the Transcription Start Site (TSS) as defined by Jiang et al. (Jiang and Pugh, 2009), whereas the ORF is considered to be from the TSS to the Stop codon. We could therefore identify three different gene categories according to the distribution of Sfp1 between promoter and ORF (Figure 8A). Namely, 177 genes have Sfp1 exclusively on their promoters (Category I; Cat I), 180 genes have Sfp1 on both promoters and ORFs (Cat II) and 198 genes have Sfp1 only in their ORFs (Cat III). Most of the RPGs were found to belong to Cat II, whereas under normal growth conditions, Sfp1 is mostly undetectable by ChIP on RiBi genes promoter and ORF.
Therefore, RiBi genes were not included in these categories except few examples such as NSR1 that was positioned on Cat III. The genes identified by these criteria to be Sfp1 targets are listed in Tables 3, 4 and 5 respectively with the corresponding GO enrichments (when they could be identified). Cat I genes were not significantly enriched in any gene ontology group and their functions are generally required for metabolism, cell wall and membranes homeostasis, cell cycle progression and stress related responses. The vast majority of Cat II genes is represented by ribosomal protein genes (as already stated) and genes required for translation elongation. Finally, genes involved in oxido-reductase activities, hydrogen and monovalent inorganic cation membrane transport, and RNA processing comprise Cat III. As positive control for this analysis, we have measured Sfp1 binding on RPG promoters and
found that out of 139 total, 119 RPG are indeed occupied by Sfp1 in their promoter confirming previously published data (Reja et al., 2015). Moreover, when our ChIP seq data are compared to previously published DNA microarray results (Jorgensen et al., 2002) we found that the expression of some of the genes identified as new Sfp1 ChIP targets, was affected (either increased or decreased) upon SFP1 overexpression (data not shown).
We therefore started to analyze more in detail the different aspects related to Sfp1 binding that could be extrapolated from our ChIP-seq experiment. First of all, as revealed by the average line plots shown on Figure 9A, Sfp1 occupies the overall promoter region defined as
“Nucleosome Free Region” (NFR; (Jiang and Pugh, 2009)). When the averaged Sfp1 signal is compared to the previously reported genome-wide characterization of promoter nucleosome landscapes (Kubik et al., 2015), we noticed that the distance between the +1 and the stable -1 nucleosome in Cat I and II genes is bigger than that of Cat III genes in a statistically significant manner (Figures 9B), indicating that Sfp1 promoter occupancy strongly correlates with what has been previously defined as NFR width. We have also measured that after overexpression Sfp1 occupancy increases both on promoters and ORFs of Cat I, II and III genes: when Sfp1 and Rpb1 binding (obtained by ChIP seq under the same conditions; see next paragraph) are compared, no significant correlation was found (R=0.0011 for Cat I; R=0.3 for Cat II; R=0.22 for Cat III) suggesting that the increased amount of Sfp1 found either on the promoter or through the ORF may not be sufficient to increase RNAPII recruitment and perhaps expression (see Figures 9A and 10 for an overall quantification of this effect and its gene by gene representation by the heat maps).
Figure 8. Characterization of gene that are occupied by Sfp1 either on the promoter or on the ORF
A. Pie chart summarizing the Sfp1 genome-wide distribution. The vast majority of yeast genes are not occupied by Sfp1. According to our established arbitrary threshold, Sfp1 gene targets can be classified into three different categories: Cat I, II and III with 177, 180 and 198 genes respectively. B. Sfp1 binds RPGs promoters but it is also significantly enriched on the ORF of most of these genes.
A B
Figure 9. Characterization of Sfp1 binding on the identified gene categories
A. Line plot for the averaged Sfp1-TAP ChIP-seq signals quantified on gene promoters and/or ORF according to the category. On the X-axis the bp numbers are shown relative to the TSS with positive values indicating downstream sequences and negative values upstream sequences. The Y-axis represents the ChIP-seq signal obtained considering a region spanning from 1000 bp upstream and downstream the TSS and reported as relative values compared to the untag for Sfp1 and using reads as unit for nucleosomes ChIP. B.
Box plots representing the distances between +1 and stable -1 nucleosome (bp). The three different identified categories are represented. The “none” group refers to all of the yeast genes where Sfp1 occupancy was not significant according to the established threshold.
Genome-wide analysis of the Sfp1-dependent changes in RNAPII distribution
Sfp1 chromatin binding measurements on a genome-wide level, were followed by a ChIP-seq analysis of RNAPII binding under normal or elevated SFP1 expression or upon conditional nuclear depletion of Sfp1 by the anchor-away technique. SFP1 overexpression was obtained
Sfp1 chromatin binding measurements on a genome-wide level, were followed by a ChIP-seq analysis of RNAPII binding under normal or elevated SFP1 expression or upon conditional nuclear depletion of Sfp1 by the anchor-away technique. SFP1 overexpression was obtained