Summary of research activity
Over many years now our laboratory has studied a wide range of topics related to chromosome biology, including gene regulation and promoter nucleosome architecture, telomere length regulation and chromosome end protection ("capping"), and the control of DNA replication initiation and elongation. At present we have projects aimed at understanding:
- how transcription factors and nucleosomes interact to bring about regulated expression of genes that drive cell growth, particularly ribosomal protein genes and genes required for ribosome assembly;
- how ribosome assembly impacts cellular protein homeostasis, and
- how promoter nucleosome architecture is established and maintained genome-wide.
We use the budding yeast Saccharomyces cerevisiae as our experimental system because it allows us to employ a powerful combination of genetic, molecular, biochemical and cell biological approaches to address specific mechanistic questions.
The ribosome biogenesis transcriptional network
Ribosome biogenesis is an energy-intensive process that drives cell growth. Underlying this point is the remarkable fact that in rapidly growing yeast cells about 50% of all initiation events on protein-coding genes by RNAPII occur on a ribosomal protein (RP) gene. Consequently, ribosome biogenesis is tightly regulated by nutrient and stress signals. Our initial studies focused on how promoter DNA sequence determines transcription factor and nucleosome assembly at the large suite of RP gene promoters, and how the resulting promoter architecture influences transcription initiation rates at these genes. This early work uncovered two major categories of RP gene promoters (Figure 1), both of which are characterized by binding of the pioneer transcription factor (pTF), Rap1, together with a pair of activator proteins, Fhl1/Ifh1 that require Rap1 for stable RP promoter association.
Most of the 138 RP gene promoters (>90%) fall into one of two categories according to the arrangement of TFs (Rap1, Fhl1 and Ifh1) and MNase-sensitive (“fragile”) nucleosomes (FNs) upstream of the transcription start site (TSS). Category I promoters bind in addition the HMG-B type non-histone chromosomal protein Hmo1 and are comprised by two FNs upstream of the TSS-associated +1 nucleosome, whereas Category II promoters do not bind Hmo1 and have only 1 upstream FN.
Of these three proteins, only the binding of Ifh1 is correlated with transcription rate, suggesting that Ifh1 is the direct activator in the system, even though its precise mechanism of action is still unknown.
In characterizing Ifh1 we noted that it co-purifies with UTP-C subcomplex (consisting of Casein Kinase 2, Utp22 and Rrp7) of the 90S preribosome complex. Remarkably, we showed that the CK2-Utp22-Rrp7-Ifh1 (CURI) complex is necessary for the stable release of Ifh1 from RP gene promoters in response to stress. In its absence, although Ifh1 is still rapidly released from RP gene promoter following stress induction (e.g. following inactivation of TORC1 kinase by rapamycin treatment), it soon returns to the promoters, even in the continued presence of stress. Interestingly, rapid inactivation of rRNA transcription leads to a release of Ifh1 from RP gene promoters that also depends upon formation of the CURI complex. These findings, taken together, suggest that the CURI complex acts to balance RP and rRNA production in stress conditions or the return to rapid growth following stress abrogation ("long timescale" regulation in Figure 5). Still unresolved by this work is the mechanism by which Ifh1 is rapidly released from RP gene promoters in response to stress ("short timescale" regulation in Figure 2).
Ifh1 release from ribosomal protein (RP) gene promoters is controlled through two separate mechanisms following stress (in the case shown, growth inhibition due to TORC1 kinase inactivation by rapamycin treatment). A short timescale (< 5 min) mechanism rapidly removes Ifh1 from promoters, causing strong down-regulation of transcription. Interaction of Ifh1 in the CURI complex is required to prevent Ifh1 re-binding at RP gene promoters (> 20 min) under conditions of continued stress (long timescale mechanism). The short timescale mechanism is induced by the Ribosome Assembly Stress Response (RASTR; see below) and involves CURI-independent condensation of Ifh1. The figure is taken from Albert et al. (2016) Molecular Cell
More recently we have clarified the role of Sfp1, a previously enigmatic TF that had been implicated in the activation of both RP and ribosome biogenesis (RiBi) genes as well as regulation of cell size, through an unknown mechanism (see Figure 3 and Albert et al.  Genes & Dev). Using the novel ChEC (chromatin endogenous cleavage) method coupled to deep sequencing (ChEC-seq) we discovered that Sfp1 binding is robustly detected at the promoters of RiBi genes, an equally large group of other growth related genes and many snoRNA genes. Remarkably, these Sfp1 binding sites are missed by the conventional ChIP-seq method. Conversely, ChIP detects Sfp1 promoter binding at two different classes of genes (RP genes and a set of genes implicated in the G1 to S cell cycle transition, known as "START") that give little or no signal in the ChEC-seq assay. Interestingly, ChIP-detected Sfp1 binding requires a co-factor (Ifh1 at RP genes and Swi4 at the G1/S regulon genes) whereas ChEC-detected binding occurs at sites matching a consensus sequence recognized by Sfp1 in vitro (Zhu et al.  Genome Research 19: 556-66). Sfp1 thus has two distinct modes of chromatin binding.
Sfp1 binding is detected at RP genes by ChIP, but not at RiBi, RiBi-like or snoRNA genes. Conversely, ChEC robustly detects Sfp1 at the latter groups of genes, but not at RP genes. In contrast, ChEC is able to detect Ifh1 binding at RP gene (left panels). Schematic (right) summarizes classes of Sfp1 target genes and their mode of detection (ChIP or ChEC). Taken from Albert et al.  Genes & Dev
Current studies are now aimed at understanding the action of Sfp1 at G1/S regulon genes, which we hypothesize to be critical for Sfp1’s role in cell size determination. Part of this work is being carried out in collaboration with Martí Aldea’s laboratory in Barcelona.
Ribosome biogenesis and cellular proteostasis
In addition to the massive transcriptional load imposed by ribosome biogenesis (involving all three RNAP complexes), it is important to keep in mind that each of the 79 proteins that constitute the yeast ribosome are produced at a prodigious rate (~2,000 copies of each protein per minute) and contribute to ~30% of total protein mass in the cell. Considering that the RPs have evolved to assemble together with the highly charged rRNA molecules in a ribosome and that unassembled RPs are highly aggregation-prone, it is easy to imagine that perturbations to the ribosome assembly process could seriously challenge cellular proteostatis systems.
We recently discovered a remarkable transcriptional response to perturbations in ribosome biogenesis. Our first hint of what we call the "Ribosome Assembly Stress Response" (RASTR) came from the analysis of a strain in which Topoisomerase I can be rapidly degraded following addition of auxin to cells (using the auxin-induced degron, or AID system). Twenty minutes following auxin addition to a Top1-AID strain we noted a strong down-regulation of nearly all RP genes and the up-regulation of heat shock factor 1 (Hsf1) target genes, with all other protein-coding genes essentially unaffected (Figure 4).
Transcriptional response (measured by RNAPII ChIP-seq) to auxin-induced degradation of Top1, at 20 and 60 min following auxin addition, as indicated. Ribosomal protein (RP), ribosome biogenesis (RiBi), and stress-induced genes targeted by Hsf1 and Msn2/4 are indicated in green, yellow, red and purple, respectively. All other genes are in grey. Expression levels given in arbitrary units (log10), comparing auxin to vehicle (mock) treatment. Taken from . Albert et al. (2019) eLife
Strikingly, RP gene down-regulation following Top1 degradation was associated with a rapid release of promoter-bound Ifh1, similar to that observed upon TORC1 inhibition. Curiously, this very specific effect on RNAPII-mediated transcription (both RP gene down-regulation and Hsf1 target gene up-regulation) disappeared by 60 minutes, due at least in part to compensation by Topoisomerase II. In addition to the effect on RNAPII, we noted that Top1 degradation caused an immediate and very strong block of rRNA production, consistent with previous findings in top1-Δ top2-ts cells. We hypothesized, based upon this latter finding, that perturbation of ribosome biogenesis might be the root cause of the RNAPII transcriptional effect we observed both on RP and Hsf1 target genes. This hypothesis is strongly supported by the finding that blocking ribosome assembly at more downstream steps, either by rapid degradation of two different ribosome assembly factors or treatment with the drug diazaborine, which blocks a late cytoplasmic step in ribosome assembly, all lead to a very similar RNAPII transcriptional response.
The above observations beg the question of how blocks to ribosome assembly could activate such a specific transcriptional response at the level of protein-coding genes. We hypothesized that RASTR results from the accumulation of unassembled RPs which in turn overload the proteostasis system. More specifically, we propose that unassembled RPs titrate Hsp70, which normally binds to and inhibits the activity of Hsf1, thus leading to activation of Hsf1 target genes. This claim is strongly supported by the finding that the specific inhibition of RP production, brought about by nuclear depletion of the RP-specific activator Ifh1, abrogates Hsf1 target gene activation following initiation of RASTR (Figure 5).
Downregulation of RP gene expression by Ifh1 nuclear depletion prior to RASTR initiation strongly dampens Hsf1 target gene activation. Schematic of protocol for Ifh1-FRB nuclear depletion (0–60 min of rapamycin treatment) followed by Top1-AID degradation (auxin treatment, 40–60 min; left panel). Scatter plots comparing RNAPII (Rpb1) ChIP-seq in Top1-AID Ifh1-FRB cells either auxin-treated (y-axis, Aux, middle panel) or auxin- plus rapamycin-treated (y-axis, Aux / -Ifh1, right panel) treated, which provokes Ifh1 nuclear depletion, as shown in the left panel, versus untreated cells (x-axis, vehicle, both middle and right panels). Taken from . Albert et al. (2019) eLife
What then leads to RP gene downregulation during RASTR? We argue that this is also due to an overload of the proteostasis system, based largely upon the observation that Ifh1 rapidly accumulates in visible condensates when RASTR is initiated in a process that depends upon continuing protein synthesis (see Figure 3 and Figure 5 in Albert et al. ). Perhaps not surprisingly, RPs also accumulate in punctate nuclear structures upon RASTR initiation, which is again dependent upon continued protein synthesis. Remarkably, we found that treatment of cells with cycloheximide alone induces a transcription response exactly opposite to and epistatic to that provoked by RASTR (see Figure 5 in Albert et al.  eLife).
Analysis of the insoluble protein fraction following RASTR initiation by mass spectrometry (MS) has revealed a large group of proteins, including not just Ifh1 and RPs, but many proteins involved in ribosome biogenesis and in translation elongation or termination. Indeed, protein aggregation during RASTR appears similar to that observed in so-called “stress granules”. Consistent with this, RASTR appears to be the cycloheximide-sensitive component of the early RNAPII response to TORC1 inhibition (see Figure 6 in Albert et al.  eLife). In addition, the downregulation of RP genes following heat shock may also be a RASTR-like response, since it is also blocked by cycloheximide (see Figure 7 in Albert et al.  eLife).
Our current view is that many forms of cellular stress impinge upon ribosome biogenesis and generate an excess of unassembled RPs that in turn activate a complex proteostasis response involved regulation at the transcriptional, translational and post-translational levels. Our current working model for this system is partly outlined in Figure 6.
Schematic of current view of the ribosome biogenesis – proteostasis connection. The top half indicates flow through the system under normal growth, the bottom under stress conditions. Numbers indicate steps we have blocked by genetic or chemical methods: (1) rRNA transcription by Top1 degradation; (2) ribosome assembly by Utp8 or Utp13 degradation; (3) ribosome assembly by diazaborine (DZA); (4) RP translation, by cycloheximide (CHX). Block of RP transcription by Ifh1 anchor-away is not indicated here. The proteasome is shown as a barrel of brown protomers. INQ: intranuclear quality control compartment; JUNQ: juxtanuclear quality control compartment. The relationship between INQ and JUNQ is still unclear.
Role of promoter nucleosome architecture and general co-activators in transcription of protein-coding genes
The basic unit of chromatin structure, the nucleosome, is generally recognized to inhibit transcription initiation. The precise location of nucleosomes at promoter regions, where RNA polymerase must bind to initiate transcription, is thus of critical importance in understanding gene regulation. We discovered that the promoter nucleosome architecture at a large set of protein-coding genes, most of which are highly transcribed and implicated in cell growth, contain one or two dynamic ("fragile") nucleosomes (FNs), characterized experimentally by an unusual sensitivity to micrococcal nuclease digestion. Our analysis of these fragile nucleosomes uncovered two short sequence motifs whose number and distribution are strongly correlated with promoter nucleosome positions at all genes (Figure 7, left; see Kubik et al. ).
The distribution of polyA and G/C-rich motifs (top, left) at all 5019 RNAPII promoter regions with a well-defined TSS is shown in red and blue, respectively (left). Grey indicates regions of protection from high levels of MNase digestion. All gene promoters are aligned to the position of their +1 (TSS-associated) nucleosome dyad axis. Schematic diagrams at the right indicate one possible model to explain the action of RSC at promoters with or without a GRF binding site. Taken from Kubik et al. (2015) Molecular Cell
We later showed that these two motifs act in a proximity- and orientation-dependent manner to drive the binding and/or action of the only essential nucleosome remodeler in yeast, called RSC (Remodels Structure of Chromatin) to generate a nucleosome-depleted region upstream of nearly all promoters. At highly transcribed genes one of a small set of general regulatory factors (GRFs), with properties similar to those of mammalian “pioneer transcription factors” (pTFs), work together with RSC to generate a nucleosome depleted region (NDR). At particularly large NDRs we imagine that RSC engulfs and destabilizes one or in some cases two FNs. Ongoing studies in collaboration with Beat Fierz’s laboratory at EPFL are addressed at understanding the structure and dynamic properties of FNs, how pTFs influence their formation and the mechanisms by which they affect transcription initiation.
More recently, we surveyed all chromatin remodeler complexes in yeast known to affect promoter nucleosome architecture, using rapid nuclear depletion or degradation methods to measure their direct effect genome-wide on both transcription initiation and nucleosome positioning (see Kubik et al. Kubik et al. ). We combined this functional analysis with measurement of sites of remodeler binding, using the novel ChEC-seq method. Whereas remodeler binding studies using the more established ChIP-seq method have been notoriously problematic, we found that ChEC-seq measurements correlate well with the sites of remodeler action as defined by our functional analyses. Our studies provided a coherent view of remodeler action genome-wide in which one class of complexes that we call the “pushers”, consisting of RSC and SWI/SNF remodelers, consistently act to move the +1 and -1 promoter nucleosomes away from each other, thus expanding the so-called nucleosome-free or nucleosome-depleted region (NFR or NDR). In contrast, a second set of remodelers, the “pullers”, consisting of ISW2 and INO80, act to pull the +1 nucleosome towards the TATA or TATA-like element, thus shrinking the NFR/NDR. Interestingly, we found, in collaboration with the Libri laboratory at Institut Monod in Paris, that puller action often suppresses transcription initiation at sites downstream from the primary site of initiation observed in wild-type cells. These results are summarized in Figure 8. Current studies are addressing the functional consequences of remodeler action during regulatory transitions provoked by nutrient changes or stress conditions.
Effect of rapid nuclear depletion or degradation of six different chromatin remodelers know to affect nucleosome positions at or near promotes (left panel). Data shown are MNase-seq analyses +/- individual remodelers (grey/color) over a ~10 kbp region on Chromosome XVI (see Figure 2, Kubik et al. ). The schematic to the right depicts our view of the combined action of “pusher” (RSC and SWI/SNF) and “puller” (INO80 and ISW2) remodelers, that together determine +1 and -1 nucleosome positions at many promoters genome-wide, in the process regulating PIC assembly and transcription start site selection. Taken from Kubik et al. (2019) Nature Structural and Molecular Biology
We have also explored the role of three general and ubiquitous transcriptional co-activator complexes genome-wide. Two of these complexes, SAGA and NuA4, contain a histone acetyltransferase (HAT) enzyme, whereas the third, called “Mediator”, is thought to work as a general bridging element that helps to connect transcriptional activators to the RNAPII holoenzyme. NuA4 and Mediator are required for cell viability. We found that SAGA, contrary to a generally held view, acts at nearly all protein-coding gene promoters, though its action is quantitatively more important, as expected, at TATA box-containing promoters. In contrast, we found that the action of Esa1, the HAT in NuA4, though widespread, is particularly important for transcription of a group of growth-promoting genes, including most genes required for ribosome biogenesis. Finally, although Mediator is required for full expression of nearly all genes, we found that its effect is most strong at a relatively small group of highly expressed TATA box-containing genes. These results are summarized in Figure 9.
Heat map showing the effect of rapid nuclear depletion of the indicated co-activator proteins on RNAPII binding at the coding regions of 5036 protein-coding genes (left panel). Genes have been k-means clustered into five groups and are displayed according to increased average transcription rate under optimal growth conditions (rich medium, 2% glucose). Cluster 3 is overrepresented by RiBi genes, cluster 4 by RP genes and cluster 5 by TATA-box containing genes. Note the strong dependence of cluster 3 genes on Esa1 (HAT in NuA4) and cluster 5 genes on full Mediator activity (Med17 anchor-away generates a partial loss-of-function phenotype). Taken from Bruzzone et al. (2018) Genes & Dev