Introduction

Genomes of higher eukaryotes are organized into multiple hierarchical levels. Chromosomes are segregated within individual territories, and within chromosomal territories active and inactive regions are separated into topologically associated domains (TADs) (Wendt & Grosveld, 2014). Within TADs, DNA loops are formed to permit the interaction between enhancers or silencers and the promoters of their target loci. Although this hierarchy suggests a static view of the genome, recent studies have suggested that the genome is quite dynamic, as multiple points of interaction form in response to developmental cues and other stimuli. This dynamic restructuring ranges from the interactions between co-regulated genes (Fanucchi et al., 2013; Papantonis et al., 2012; Park et al., 2014; Schoenfelder et al., 2010), to the convergence of enhancers dispersed across multiple chromosomes into a hub that regulates a single gene (Monahan & Lomvardas, 2015), to the reorganization of the genome that occurs in the zygote (Schulz & Harrison, 2019).

Despite its evolutionary distance, the yeast Saccharomyces cerevisiae also possesses an organized genome. Centromeres are located in a cluster at the spindle pole body, while chromosomal arms are extended with telomeres and the nucleolus located at the opposite side of the nucleus (Duan et al., 2010), resulting in a Rabl-like configuration (Taddei & Gasser, 2012). And as in higher eukaryotes, the budding yeast genome is organized into TAD-like structures (Eser et al., 2017) subdivided into smaller loop domains (Hsieh et al., 2015). Also as is the case in mammalian cells, the yeast genome is not static. Genes, including INO1, GAL1 and GAL10, have been observed to reposition from the nuclear interior to the nuclear periphery upon their activation (Brickner et al., 2019; Brickner & Walter, 2004; Cabal et al., 2006; Casolari et al., 2004; Dieppois et al., 2006; Green et al., 2012). In response to methionine starvation, several Met4-regulated genes have been observed to cluster (Du et al., 2017; Lee et al., 2024).

More dramatic still are Heat Shock Response (HSR) genes under the regulation of Heat Shock Factor 1 (Hsf1). Upon exposure to acute thermal stress (heat shock [HS]), HSR genes dispersed across multiple chromosomes transcriptionally activate and engage in novel cis- and trans-intergenic interactions between one another – principally although not exclusively involving coding regions – that culminate in their coalescence into intranuclear foci (Chowdhary et al., 2017, 2019). These stress-induced foci, comprised of Hsf1 and components of the transcriptional machinery, form quickly, rearrange rapidly and dissolve suddenly (Chowdhary et al., 2022). As such, Hsf1 clusters exhibit properties of biomolecular condensates, defined as self-organized membrane-free compartments enriched in specific macromolecules that may or may not be phase-separated (Banani et al., 2017). Hsf1 condensate formation elicited by heat shock, like that of HSR gene interactions, parallels the kinetics of induction and attenuation of Hsf1-dependent genes (Chowdhary et al., 2017, 2022).

The Hsf1-driven heat shock response is a fundamental, evolutionarily conserved transcriptional program characterized by the gene-specific transcription factor (TF) Hsf1, its DNA recognition element (heat shock element (HSE)) and a core set of target genes encoding molecular chaperones and co-chaperones (reviewed in (Verghese et al., 2012)). In absence of proteotoxic stress, yeast Hsf1 is bound by Hsp70 and its co-chaperone Sis1 in the nucleoplasm (Feder et al., 2021; Krakowiak et al., 2018; Peffer et al., 2019; Zheng et al., 2016). Upon encountering stress, Hsp70 is titrated by unfolded proteins, particularly orphan ribosomal proteins located in the nucleolus and nascent polypeptides in the cytosol (Albert et al., 2019; Ali et al., 2023; Tye et al., 2019; Tye & Churchman, 2021). This results in the release of Hsf1 which then trimerizes and binds to HSEs located upstream of ∼50 genes whose transactivation is dependent on this factor (Pincus et al., 2018). Once proteostasis is reestablished, excess Hsp70 binds Hsf1, inactivating it, thereby closing the negative feedback loop that regulates Hsf1 transcriptional activity (Krakowiak et al., 2018; Zheng et al., 2016). Note that in yeast, a fraction of Hsf1 is constitutively trimeric and it is this species that binds high-affinity HSEs even under non-stressful conditions (Giardina & Lis, 1995; Pincus et al., 2018).

In addition to thermal stress, Hsf1 can be activated by chemical stressors such as ethanol. Ethanol is a metabolite of glucose breakdown that budding yeast cells secrete into their surroundings. Ethanol production helps yeast outcompete other microbes in the environment (Liti, 2015; Piškur et al., 2006; Rozpedowska et al., 2011). Furthermore, once glucose is depleted, ethanol serves as an alternative carbon source (Piškur et al., 2006). Given this strategic use of ethanol, it is of fundamental importance for yeast to have a mechanism in place to respond to the stress that ethanol elicits. Similar to thermal stress, exposure to ethanol causes a large number of cellular perturbations including disruption of the plasma membrane (Piper et al., 1994); disruption of the H⁺ ATPase and intracellular acidification (Rosa & Sá-Correia, 1991, 1996; Triandafillou et al., 2020); production of reactive oxygen species (Bandas & Zakharov, 1980; Davidson et al., 1996); depolymerization of the actin cytoskeleton (Homoto & Izawa, 2018; Tan et al., 2017); cell cycle arrest (Johnston & Singer, 1980; Kubota et al., 2004); disruption of mRNP transport to the daughter cell and formation of stress granules (Grouši et al., 2009; Kato et al., 2011); global inhibition of transcription and translation (Bresson et al., 2020; Gasch et al., 2000); and formation of protein aggregates (Piper, 1995; Plesset et al., 1982; Stanley et al., 2010). The cell counteracts many of these perturbations through the enhanced production of molecular chaperones (Altamirano et al., 2024; Verghese et al., 2012).

Here, we investigate activation of the HSR in yeast exposed to 8.5% ethanol (ethanol stress (ES)) and compare it to the response induced by exposure to 39°C (HS). We find that like HS, exposure to ES induces transcription of Hsf1-regulated HSR genes and induces concerted intergenic interactions between them. In contrast to HS, however, Hsf1 condensate formation and HSR intergenic interactions cells peak well before transcription. The delay in transcriptional induction may be linked to a widespread increase in nucleosome occupancy that occurs upon exposure of cells to ethanol. At longer times of ethanol exposure, HSR gene transcript accumulation continues to increase while HSR gene coalescence has already dissipated. Likewise, ES-induced Hsf1 condensates are present for a prolonged period, in contrast to HS-induced condensates that begin to dissipate shortly after they appear. Collectively, our data indicate that different stimuli drive distinct transcription, chromatin, topologic and condensation phenomena, yet all are dependent on Hsf1.

Results

Thermal stress and ethanol stress elicit distinct proteotoxic responses in yeast

Saccharomyces cerevisiae in the wild metabolizes glucose and other sugars into ethanol, which the yeast secretes into the environment to suppress microbial competition. Therefore, it is likely that yeast has evolved mechanisms to contend with ethanol toxicity. Indeed, a common laboratory strain (W303) retains viability when cultivated in the presence of 8.5% ethanol, although its ability to proliferate is diminished (Figure 1A, B). As assessed by the presence of Hsp104-containing foci (a measure of protein aggregation (Liu et al., 2010)), the rate of cytosolic protein aggregation is similar in cells exposed to either 8.5% ethanol or 39°C thermal stress (Figures 1D and 1E; see Figure 1C for experimental design). However, it is notable that the number and volume of Hsp104 foci, and by extension protein aggregation, is substantially higher in cells subjected to ethanol stress.

Figure 1.

We next investigated the subcellular localization of the Hsp70 co-chaperone, Sis1, in cells exposed to ethanol versus thermal stress. Previous work has shown that Sis1 is diffusely localized within the nucleus under control conditions where it promotes binding of Hsp70 to Hsf1, thereby repressing the HSR (Feder et al., 2021). In response to acute HS, Sis1 relocates to nucleolus where it forms a perinucleolar ring, and to the cytoplasm where it forms cytosolic clusters that colocalize with Hsp104 and spatially associate with the endoplasmic reticulum (Feder et al., 2021). Consistent with these previous observations, we observed rapid relocalization of Sis1 in response to a 39°C HS. It formed a ring-like structure within 2.5 min consistent with a perinuclear location. Such relocalization lasted at least 60 min and spatially separated Sis1 from Hsf1 (Figures 2A and 2B). A similar result was obtained when cells were exposed to 42°C. In response to ES, Sis1 remained largely co-localized with Hsf1 within the nucleoplasm although some enrichment at the nuclear periphery was evident. This was the case whether cells were exposed to 5% or 8.5% ethanol (Figures 2A and 2B). In addition, a 60 min exposure to 8.5% ethanol resulted in the formation of prominent cytosolic Sis1 puncta that were less evident in the HS sample (Figure 2A). These observations suggest that these two stressful treatments elicit a qualitatively distinct response. Experiments described below provide further support for this possibility.

Figure 2.

Ethanol stress induces transcriptional activation of Hsf1-dependent genes but with delayed kinetics and reduced expression compared to thermal stress

To gain further insight into the mechanism by which S. cerevisiae contends with ethanol-induced proteotoxicity, we assessed the kinetics of transcriptional activation of HSR genes in cells exposed to ethanol stress. Cells were cultivated to early log phase in rich YPD medium, then ethanol was added to a final concentration of 8.5% and cell aliquots were removed at 0-, 10-, 20- and 60-min. Transcription was terminated through addition of sodium azide (see Materials and Methods). A parallel culture was exposed to an instantaneous 30° to 39°C heat shock and cells were removed at the corresponding time points. Transcription was terminated as above.

While cells exposed to heat shock displayed a rapid and substantial increase in HSR mRNA levels (typically >10-fold increase within 10 min of thermal upshift), those exposed to ethanol stress only weakly induced the same cohort of genes (Figure 3A). However, while HS induced a transient increase in RNA expression, ES induced a sustained increase that was evident at all Hsf1-dependent genes tested (Figure 3– figure supplements 1 and 2). A corresponding immunoblot analysis of two Hsf1 targets, Hsp104 and Btn2, strengthens the notion that HS induces a rapid yet transient response whereas ES induces a delayed, yet far more sustained increase in gene expression (Figure 3–figure supplement 3). In the case of HSP12 and HSP26, whose transcription is under the dual regulation of Msn2 and Hsf1, exposure to heat shock resulted in a high level of induction as previously observed (Chowdhary et al., 2019) yet exposure to 8.5% ethanol failed to cause detectable activation during the first 20 min (Figure 3B; Figure 3–figure supplement 1). Nonetheless, as described below, both HSP12 and HSP26 respond to acute ethanol stress but do so through their inducible and dramatic 3D genomic repositioning.

Figure 3.

Pol II recruitment and histone eviction are delayed in ethanol-stressed cells and this correlates with a transient, widespread increase in nucleosome density

The delayed transcriptional response of HSR genes in ES-versus HS-treated cells prompted us to investigate occupancy of Hsf1, RNA Pol II and histones at these genes over a time course. A possible explanation for the delay in activation in cells exposed to ethanol stress is less rapid (or reduced) Hsf1 binding to the genes’ upstream regulatory regions. To explore this possibility, we exposed cells to either thermal or chemical stress and processed them for chromatin immunoprecipitation (ChIP) analysis. As previously observed (Kim & Gross, 2013; Pincus et al., 2018; Sekinger & Gross, 2001), occupancy of Hsf1 at its target loci increases many-fold following a brief heat shock (Figure 4B, left, dark red; see Figure 4A for location of primers). Hsf1 occupancy typically declines after 60 min of continuous thermal stress and in the case of TMA10, dissociation begins much sooner. In response to ethanol stress, Hsf1 occupancy steadily increased, in most cases reaching maximal levels by 20 min and plateauing thereafter (Figure 4B, left, black). These results suggest that ethanol stress induces binding of Hsf1 to its cognate HSEs to a degree similar to heat shock, yet such binding is more gradual. Moreover, given the above, Hsf1’s binding to these sites fails to elicit a corresponding transcriptional response.

Figure 4.

In light of this disconnect between Hsf1 binding and HSR mRNA production, we evaluated abundance of the large subunit of Pol II at representative genes over the same time course. In response to heat shock, Rpb1 is rapidly recruited to the promoters and coding regions of each HSR gene, peaking within 2.5 min and then gradually declining over the next ∼60 min (Figure 4B, middle, red and pink traces). By contrast, its occupancy is noticeably delayed in cells exposed to ethanol stress (blue traces), consistent with reduced transcript levels (Figure 4–figure supplement 1B). The increase in HSR mRNA in heat-shocked cells parallels, yet consistently lags, the abundance of Pol II within HSR gene coding regions (Figure 4–figure supplement 1A). This delay in reaching peak accumulation may reflect contributions beyond RNA synthesis, such as transient enhanced stability of HSR transcripts during the acute phase of heat shock.

To obtain further insight into the chromatin landscape present during the two stresses, we assayed histone H3 abundance as a measure of nucleosome density. In response to heat shock, nucleosomes are rapidly displaced over the promoter and coding regions of strongly expressed HSR genes (HSP82, HSP104 and SSA4); a similar, albeit less robust, response is observed for TMA10 and HSP12. Following this initial phase (∼20 min), nucleosomes reassemble over these genes and in certain cases return to their original density by 60 min (Figure 4B, right, red and pink traces). This dramatic and dynamic remodeling has been previously observed (Kremer & Gross, 2009; Zhang et al., 2014; Zhao et al., 2005). In contrast, ethanol stress elicits a transient increase in H3 abundance over all five genes (Figure 4B, right, blue traces). This apparent increase in chromatin compaction is not restricted to HSR genes; a variety of unrelated loci, including a stress-responsive, Msn2-regulated gene (PGM2), two constitutively expressed genes (TUB1, ACT1), two genes assembled into SIR-dependent heterochromatin (HMLα1, YFR057w) and a non-transcribed region (ARS504) also exhibit a transient increase in nucleosome density in ethanol-exposed cells (Figure 4– figure supplement 2A).

In agreement with this idea, chromatin volume (as assessed by the signal arising from an H2A-mCherry protein fusion) decreases upon exposure of cells to ethanol stress (and much more transiently to heat stress; Figure 4–figure supplement 2[B, C]). These data suggest that the increase in nucleosome density antagonizes Pol II recruitment and its subsequent release into the coding regions of HSR genes in ES cells. Once Pol II has been stably recruited, elongation and concomitant histone eviction ensues (Figure 4B). Altogether, our ChIP data indicate that Hsf1 binds to its target enhancers less readily in ethanol-stressed than in thermally stressed cells. This impediment to Hsf1 occupancy is magnified by a corresponding, and more severe, hindrance to Pol II recruitment resulting in a pronounced delay in HSR gene transcription.

Acute ethanol stress induces rapid and profound 3D genomic repositioning of HSR loci

An intriguing feature of HSR genes is the fact that they coalesce into discrete intranuclear foci in response to heat shock. Such interactions have been documented using both molecular (chromosome conformation capture (3C)) and imaging (fluorescence microscopy) approaches (Chowdhary et al., 2017, 2019, 2022; Rubio & Gross, 2023). Intriguingly, physical interactions specifically involve Hsf1 targets irrespective of their location in the genome. Other loci, including adjacent, transcriptionally active genes, show little or no tendency to interact with Hsf1-dependent genes. Such cis- and trans-interactions principally involve gene coding regions and are highly dynamic, typically peaking at 2.5 min and dissipating by 30- to 60-min. The kinetics of HS-induced coalescence often, although not always, correlate with kinetics of transcriptional induction; they also parallel the formation of Hsf1 condensates (Chowdhary et al., 2017, 2019, 2022) as discussed further below.

Given these previous observations, we wished to know if ethanol stress induced similar 3D genome restructuring. It seemed unlikely that such topological changes would occur during the initial phase of ES since only weak Pol II occupancy and low HSR transcript levels are observed (Figures 3, 4). However, as shown in Figure 5, exposure to 8.5% ethanol triggered frequent intergenic interactions between Hsf1 targets during the first 10 min as revealed by Taq I-3C, a highly sensitive, quantitative version of 3C (Chowdhary et al., 2020) (see Figure 5 – figure supplement 1 for location of 3C primers). Both intra- and interchromosomal interactions can be detected. Moreover, the interaction frequencies following this exposure in most cases equaled, and in some instances exceeded, those detected in cells heat-shocked for 2.5 min (Figures 5A, 5B), when peak 3C interactions occur in thermally stressed cells (Chowdhary et al., 2017). A detailed kinetic analysis revealed that intergenic interactions elicited by ethanol stress, similar to those induced by thermal stress, are highly dynamic: detectable within 2.5 min, peak shortly thereafter (within 10-20 min) and largely attenuate by 60 min (Figures 5 and 6A).

Figure 5.

Figure 6.

It has been previously suggested that a functional link exists between gene looping and transcriptional activation (Ansari & Hampsey, 2005; O’Sullivan et al., 2004). Indeed, gene loops and other intragenic interactions (Chowdhary et al., 2017) are readily detected within HSR genes in cells exposed to ethanol. However, as is the case with intergenic interactions, these topological changes are kinetically uncoupled from both transcription and Pol II occupancy. They are detected quite early (within 2.5 min), peak soon thereafter (at 10 min) and attenuate by 60 min (Figure 6B and Figure 5 – figure supplement 2). Taken together, our 3C and expression analyses indicate that 3D genomic repositioning and intragenic looping of HSR genes precedes the maxima of transcript levels and Pol II occupancy. Moreover, for certain loci (e.g., HSP12 and HSP26), they argue that neither transcription nor Pol II recruitment is required to drive 3D topological changes in these genes. Further evidence in support of this is provided below.

Live cell imaging reveals that HSR genes coalesce to a similar degree under ethanol stress and heat stress conditions

To provide an orthogonal line of evidence for HSR gene interaction, we employed fluorescence microscopy to image live cells bearing LacO-tagged HSP104 and TetO-tagged TMA10 loci in cells expressing GFP-LacI and TetR-mCherry fusion proteins. Both genes are located on Chromosome XII, on opposite arms, and are physically separated by the nucleolus (100-200 rDNA repeats) that lies between them (Duan et al., 2010) (schematically depicted in Figure 7-figure supplement 1A). In the absence of stress, fluorescence signals representing these two genes are typically well-separated (Figure 7-figure supplement 1B, 0 min). Upon heat shock, they rapidly converge, usually within 2.5 min. Upon exposure to ethanol, gene convergence is also observed, albeit less rapidly (Figure 7-figure supplement 1B; see also below). Despite the slight delay, these results demonstrate that HSP104 and TMA10 coalesce in ethanol stressed cells and do so with similar frequency as in thermally stressed cells (Figure 7-figure supplement 1C), consistent with the 3C analysis described above.

Having confirmed the physical interaction of Hsf1-dependent genes under ethanol stress, we assessed the transcriptional status of coalesced genes. Our RT-qPCR analysis indicated that the increase in HSR mRNA levels in ES-induced cells is delayed compared to those in HS-induced cells (Figure 3). To obtain insight into HSR gene transcription kinetics in single cells, we integrated a stem loop array (24xMS2) upstream of HSP104, allowing production of a chimeric transcript visualized upon binding of the MCP-mCherry fusion protein (Haim et al., 2007). This strain also harbored LacO-tagged HSP104 and HSP12 genes and expressed GFP-LacI (schematically illustrated in Figure 7A). We were unable to detect an MCP-mCherry focus adjacent to HSP104 under no stress conditions (Figure 7B; 0 min), consistent with very low HSP104 basal transcript levels (Figure 3A). Heat shock induced rapid coalescence between HSP104 and HSP12, as well as transcription from HSP104. These phenomena were detectable by 2.5 min as a merged signal of the chimeric transcript and the two GFP-labeled genes (Figure 7B). This visualization method allowed us to quantify the percentage of the population that is actively engaged in transcription, revealing that during heat shock, transcription and coalescence are positively correlated (Figure 7[C, F]). Consistent with this, after HSP104 transcription declined (10 min HS), so did HSP104-HSP12 coalescence (Figure 7 – figure supplement 1D).

Figure 7.

A detailed single cell analysis supports the strong spatiotemporal correlation between HSR gene coalescence and transcription in heat-shocked cells (Figure 7D). In contrast, under ethanol stress, HSP104 RNA was not detected until 5-10 min (and in most cases not until 10 min) even though HSP12 and HSP104 coalesced as early as 2.5 min (Figure 7E; Figure 7 – figure supplement 1D). Indeed, such an analysis suggests temporal uncoupling between HSR gene coalescence and HSR gene transcription in ethanol stressed cells (Figure 7F). Underscoring the disconnect between 3D genome repositioning and transcription is the fact that in ethanol stressed cells, enhanced HSP12 RNA levels were undetectable until 60 min (Figure 3B and Figure 3 – figure supplement 1). Collectively, our RT-qPCR, 3C and imaging data argue that ethanol stress induces striking topological changes in HSR genes within the first 10-20 min and that these are accompanied by a minimal increase in transcript levels. This provides a sharp contrast to heat-shocked cells where there exists a strong temporal correlation between HSR gene transcription and HSR gene repositioning (Figure 7F).

Ethanol stress induces rapid formation of long-lived Hsf1 condensates

Recently, TF condensates have been proposed as a mechanism for transcriptional regulation (Boija et al., 2018; Cho et al., 2018; Hnisz et al., 2017; Nair et al., 2019; Sabari et al., 2018). Heat shock-activated Hsf1 forms small nuclear condensates (diameter of ≤300 nm) that localize at HSR gene loci in both human (Zhang et al., 2022) and budding yeast cells (Chowdhary et al., 2022), and their presence positively correlates with HSR gene transcriptional activity. The tendency of Hsf1 to form condensates may be linked to its extensive intrinsically disordered structure (Figure 8A), a feature proposed to be critical for the biomolecular condensation of proteins (reviewed in (Alberti et al., 2019; Banani et al., 2017)). Additionally, nuclear stress bodies comprised of HSF1 bound to satellite III DNA repeats in human cells have been described (Jolly et al., 1997, 2002), although formation of these large foci (diameter of several µm) is independent of the sites of active Heat Shock Protein genes and their presence may repress HSR gene activation (Gaglia et al., 2020).

Figure 8.

Given the temporal link between Hsf1 condensation and transcription established in heat-shocked yeast cells, we anticipated that the appearance of Hsf1 condensates in ethanol-stressed cells would be delayed relative to what is seen in heat-shocked cells. However, we found that ethanol stress induced formation of Hsf1-GFP condensates as rapidly as did heat shock. These were visible in virtually all ES cells as early as 2.5 min, paralleling their rapid appearance in HS cells (Figure 8 B-D). However, in contrast to the rapid dissolution of condensates in HS cells, those formed in cells exposed to 8.5% ethanol showed no evidence of dissipating even after 60 min of continuous exposure (Figure 8B, D). An independent analysis of Hsf1 tagged with a monomeric GFP, mNeonGreen, gave virtually identical results: >90% of cells exhibited Hsf1 puncta as early as 2.5 min and such puncta were stably maintained for 150 min (Figure 8 – figure supplement 1). Moreover, in a further analysis they were still visible at 5.5 h (data not shown). Therefore, in 8.5% ethanol-stressed cells, formation of Hsf1 condensates is uncoupled from HSR gene transcription and their maintenance is uncoupled from HSR gene repositioning. An important implication is that although condensates may initiate or promote HSR gene repositioning, they cannot maintain the 3D restructured state of the genome.

Does the enhanced stability of ES-induced puncta arise from the nature of the stress or its intensity? To test this, we examined the behavior of yeast cells exposed to a lower concentration of ethanol. As shown in Figure 8 – figure supplement 1, Hsf1 puncta formed in >95% of 5% ethanol-treated cells within 2.5 min, resembling what was seen in cells exposed to either 8.5% ethanol or 39°C HS. Yet such puncta dissipated in a majority of 5% ES cells within 30 min, resembling the case with HS but strongly contrasting to what was seen with 8.5% ethanol. Therefore, although 5% and 8.5% ES elicit a similar pattern of Sis1 subcellular relocalization, one that is distinct from that caused by either a 39° or 42°C HS (Figure 2), it is the greater intensity of 8.5% ethanol stress, not its intrinsic nature, that most closely correlates with the stability of Hsf1 condensates.

Hsf1 and Pol II are required for HSR gene interactions in response to both heat shock and ethanol stress

The above analyses reveal several unexpected differences in the way yeast responds to ethanol stress versus heat stress. Given these differences, we asked whether either Hsf1 or RNA Pol II are required for the repositioning of HSR genes in response to ethanol stress; both have been shown to be necessary for 3D genome restructuring in response to heat shock (Chowdhary et al., 2019, 2022). To do so, we used the auxin-induced degradation system to conditionally degrade Hsf1 and the Rpb1 subunit of RNA Pol II in appropriately engineered strains. As schematically summarized in Figure 9A, cells expressing degron-tagged Hsf1 or Rpb1 were pre-treated with 1 mM indoleacetic acid (IAA) for 30-40 min, at which time each protein was >90% degraded (Figure 9 – figure supplement 1A). Although minimal short-term growth defects were detectable, loss of Rpb1 and Hsf1 resulted in loss of cell viability at all temperatures (Figure 9 – figure supplement 1 [B, C]).

Figure 9.

Consistent with previous observations (Chowdhary et al., 2019), intergenic interactions between HSR genes, located on the same or different chromosomes, were nearly obviated in cells conditionally depleted of either Hsf1 or Rpb1 and then heat shocked (Figure 9B and data not shown). While a similar strong dependence on Hsf1 is observed in 8.5% ethanol-treated cells, close inspection of the data suggests that residual HSR-HSR gene interactions are retained in Rpb1-depleted cells for certain loci. Particularly noteworthy is the inducible interaction between HSP12 and HSP26, genes co-regulated by Msn2 and Hsf1. Neither is detectably activated following a 10-min exposure to ethanol (Figure 3B), yet the genes engage in an intergenic interaction that is unaffected by prior degradation of Rpb1 (Figure 9B). Together with the kinetic uncoupling of HSR gene repositioning with Pol II recruitment/transcription and the relative permanence of Hsf1 condensates described above, these observations raise the possibility that 8.5% (v/v) ethanol-induced Hsf1 condensates are compositionally different from those formed in response to heat shock and drive HSR gene repositioning and transcription in a mechanistically distinct way (see Figure 10 for model).

Figure 10.

Discussion

Ethanol stress induces HSR gene transcription, HSR gene coalescence and formation of Hsf1 condensates

Here we have shown that exposure of budding yeast to a high, but sub-lethal, concentration of ethanol strongly stimulates the binding of Hsf1 to the upstream regulatory regions of HSR genes. Unexpectedly, such binding – which is evident as early as 2.5 min – does not lead to concurrent recruitment of Pol II and transcription of HSR genes. Instead, Pol II recruitment and transcription are delayed, typically for 10 min or longer. As exposure to 8.5% ethanol causes a global yet transient increase in H3 occupancy – which we interpret as an increase in the compaction of chromatin (see below) – the increase in nucleosome density may present a barrier to both Pol II recruitment and elongation. In addition, another feature of heat shocked-induced Hsf1 activation, repositioning of HSR genes within the 3D genome, is observed in cells exposed to ethanol. However, unlike transcription, this phenomenon occurs rapidly and is transient, resembling what is observed in heat shocked cells. The lack of temporal linkage between HSR gene transcription and HSR intergenic interactions is consistent with the idea that HSR gene coalescence and transcription are distinct phenomena and that Hsf1 can instigate long-range changes in 3D genome structure independently of inducing transcription. Thus, Hsf1 represents an example of a gene-specific TF that has functions independent of regulating transcription (discussed further below).

It has recently been demonstrated that in response to heat shock, inducible transcriptional condensates drive 3D genome reorganization in budding yeast. This conclusion arose from several features, including the tight temporal linkage between Hsf1 condensation and HSR intergenic interactions and the similar sensitivity of these two phenomena to the aliphatic alcohol, 1,6-hexanediol (Chowdhary et al., 2022). Similar to previous observations of heat-shocked cells (and confirmed here), we have found that in cells exposed to 8.5% ethanol, Hsf1 forms discrete puncta within 2.5 min. However, Hsf1-containing condensates formed in response to 8.5% ES are stable – persisting for hours – while such assemblies dissipate within 30 min in cells exposed to HS. In addition, while Hsf1 condensates formed as rapidly in cells exposed to 5% versus 8.5% ethanol, those formed in the presence of 5% ethanol proved to be transient, resembling their counterparts in HS cells. Relevant to this observation is the finding that Hsf1 activation in yeast exposed to 5% ethanol is dependent on the misfolding of nascent proteins, while in cells exposed to 8% ethanol, Hsf1 activation is dependent on the misfolding of both nascent and mature proteins (Tye & Churchman, 2021). Therefore, the persistence of Hsf1 condensates in 8.5% ethanol may derive from widespread unfolded mature proteins that are more difficult to refold than short nascent polypeptides attached to ribosomes.

Additional features might account for the way in which these two proteotoxic stresses elicit the HSR in budding yeast. For example, while ethanol denatures proteins similar to thermal stress (Furutani & Izawa, 2022; Kato et al., 2011), it impacts the cell at other levels. Actively growing yeast preferentially consume sugars and even in the presence of oxygen can produce ethanol (Dashko et al., 2014). This change in carbon source, termed the diauxic shift, is marked by a reduction in growth rate due to the less efficient generation of energy as well as accumulation of waste products and depletion of nutrients (Herman, 2002). Consistent with activation of the HSR, this physiological adjustment is accompanied by increased chaperone synthesis (Piper, 1995; Piper et al., 1994) as well as an increase in the activity of alcohol metabolism-related enzymes. Such an adaptation likely takes time. In contrast to the gradual increase in ethanol that occurs under natural conditions, here we have imposed an instantaneous ethanol shock. The addition of ethanol elicits a heat shock transcriptional response and synthesis of chaperones. The cell growth arrest seen here (Figure 1A) may arise as part of the adaptive program deployed upon encountering ethanol stress, such as a transition into quiescence. However, we fail to see activation of Msn2 targets, a signature of quiescence (Breeden & Tsukiyama, 2022), early in the exposure to ES at a time when Hsf1 itself is activated (Figure 3B). While other possibilities exist, our results suggest that the activation of programs involved in proteome protection are likely the first line of defense against sudden ethanol stress. Salient differences by which yeast cells respond to heat versus ethanol stress are summarized in Table 1; a model summarizing the Hsf1 response under both HS and ES conditions is provided in Figure 10.

Table 1.

Exposure to ethanol transiently induces the global compaction of chromatin

An important observation is that exposure of cells to 8.5% ethanol leads to a widespread increase in H3 ChIP signal that we interpret as increased compaction of chromatin. Although this effect is temporary, both euchromatic and heterochromatic regions are impacted, and this is consistent with measurements of total chromatin volume that reveal a decrease lasting nearly 60 min. A similar outcome was recently reported for both yeast and mammalian cells exposed to aliphatic di-alcohols (Itoh et al., 2021; Meduri et al., 2022). It is probable that the effect of ethanol exposure is nearly instantaneous since, as discussed above, ethanol is known to denature proteins (Kato et al., 2019), likely by dehydration (disruption of biomolecules’ hydration shell). While such denaturation may contribute to chromatin compaction, the effect is reversible, possibly due to refolding / renaturation mediated by Hsp70 and other molecular chaperones whose intracellular concentration increases during ethanol exposure. As mentioned above, this temporary increased density of chromatin could suppress Pol II recruitment and subsequent elongation. Also not ruled out is a direct effect of 8.5% ethanol on one or more components of the Pol II machinery. Possible advantages of chromatin compaction could be to downregulate global transcription, as well as to limit chromatin damage by reactive oxygen species or other potentially damaging molecules present in the cell during ethanol stress (Bradley et al., 2021; Costa et al., 1997; Davidson et al., 1996; Davidson & Schiestl, 2001; Shen et al., 2020; Voordeckers et al., 2020).

A novel function for a TF that is uncoupled from regulating the transcription of its target gene

A key finding is that in response to ethanol stress, HSR genes reposition and Hsf1 condensates form well before transcription of Hsf1-dependent genes peaks, and in certain cases, is even detected. This suggests that Hsf1 has a function distinct from transactivating its target genes: it drives HSR gene repositioning that culminates in their physical coalescence. Likewise, deletion of the N-terminal IDR (NTA, amino acids 1-140; see Figure 8A) was observed to have little effect on HSR gene transcription during an acute heat shock yet formation of Hsf1 condensates and intergenic 3C interactions were suppressed (Chowdhary et al., 2022). Furthermore, as discussed above, recruitment of Hsf1 to its target HSEs is delayed in response to ES versus HS. Nevertheless, formation of Hsf1 condensates occurs with similar kinetics under both conditions, well prior to the peak of Hsf1 DNA binding in ES cells. This suggests the possibility that two types of Hsf1 clusters exist: one that is DNA-bound and the other that hovers over the HSR genes. It is possible that this cloud-forming Hsf1 cluster aids in target searching, by analogy with Gal4 that can also form clusters without binding DNA (Meeussen et al., 2023).

Conclusion

While Hsf1 is known to induce interactions between transcriptionally active HSR genes in response to heat shock, here we have demonstrated a similar role for this transcription factor in response to ethanol stress. Despite minimal HSR gene transcription during the initial 10 min exposure to ethanol, HSR genes engage in robust physical interactions, rivaling those seen for 2.5 min HS. This interaction correlates with an increase in definition of Hsf1 condensates yet is not accompanied by concurrent recruitment of Pol II to promoters, perhaps due to the compaction of chromatin in ethanol-stressed cells. Hsf1 therefore forms condensates and drives 3D repositioning of its target genes without appreciably activating these genes. Furthermore, Hsf1 does this through formation of condensates that may be materially different than those that form in response to HS. Our results also argue that formation of TF clusters, while sufficient for TF DNA binding and 3D genome restructuring, are not sufficient to drive transcription. Additional factors and/or activities – such as the opening of chromatin – are necessary. Further research into the biophysical properties, molecular regulation, and functional consequences of HSF1 condensates will deepen our understanding of how cells respond to both thermal and chemical stress and maintain cellular homeostasis.

Materials and Methods

Yeast Strain Construction

The HSF1-mNeonGreen (HSF1-mNG) diploid strain LRY033 expressing a yeast-optimized version of mNeonGreen, derived from Branchiostoma lanceolatum (Shaner et al., 2013), and co-expressing Sis1-mKate and Hsp104-BFP (Blue Fluorescent Protein; mTagBFP2) was created as follows. First, DPY1561 (Feder et al., 2021) was crossed to W303-1B to create strain LRY031. This diploid was sporulated and a strain homozygous for HSF1 and retention of one allele each of SIS-mKate and HSP104-mTagBFP2 was obtained after back crossing. The resultant diploid was named LRY032. LRY032 was transformed with a PCR amplicon containing 50 bp of homology sequences flanking the stop codon, targeting the mNG tag to HSF1 at its C-terminus flanked by the HIS3 selectable marker. The plasmid template for this amplification was pFA6a-link-ymNeonGreen-SpHis5 (Botman et al., 2019). LRY033 is heterozygous for HSF1-mNG, SIS1-mKate and HSP104-mTagBFP2.

Other strains were created as follows. LRY037 was constructed using the HSF1-targeted mNeonGreen amplicon to transform strain W303-1B. LRY040 was constructed by transforming LRY037 with an amplicon containing RPB3-mCherry::hphMX6, obtained using genomic DNA from strain SCY004 (Chowdhary et al., 2022) as template. LRY100 and LRY102 were constructed using LRY016 (Rubio & Gross, 2023) as recipient of the mini-degron tag amplified from pHyg-AID*-9myc (Morawska & Ulrich, 2013), targeted to the C-terminus of HSF1 and RPB1, respectively.

A complete list of strains as well as plasmids and primers used in strain construction are listed in Supplemental File 1 – Tables 1, 2, 3.

Yeast Culture and Treatment Conditions

Cells were grown at 30°C in YPDA (1% w/v yeast extract, 2% w/v peptone, 2% w/v dextrose and 20 mg/L adenine) to mid-log density (OD₆₀₀= 0.6 – 0.8). For ethanol stress, the cell culture was mixed with an equal volume of YPDA containing 17% v/v ethanol (yielding a final concentration of 8.5%) and incubated at 25°C for different lengths of time as indicated in the figures. For heat shock, the mid-log culture was mixed with an equal volume of 55°C YPDA medium to achieve an instantaneous temperature upshift to 39°C, and the culture was maintained at 39°C for the indicated times. The no stress samples were diluted with an equivalent volume of YPDA and maintained at 25°C. Samples were kept at their respective temperatures using a water bath with constant shaking.

Cell Viability and Growth Assays

Cell Viability Assay

Cells were grown at 30°C in YPDA to OD₆₀₀=0.6 and then diluted to OD₆₀₀=0.4 using an equivalent volume of medium as described above for ethanol stress, heat shock or the no stress control (YPDA at 25°C). Cells were kept under these conditions for 3 h; during this time aliquots were taken at different timepoints and diluted 1:26,000 for plating onto YPDA. Plates were incubated at 30°C for 3 days then scanned. Colonies were quantified using ImageJ/Fiji (v. 1.53t) (Schindelin et al., 2012) - “Analyze Particles” option. The number of colony-forming units (CFUs) obtained in stress samples were normalized to the no stress samples and expressed as a percentage of the number of CFUs obtained in the no stress sample.

Growth Assay

Cells were grown in liquid culture and subjected to the same treatments as described above. OD₆₀₀ readings of each sample were taken at intervals over 3 h (see Figure 1). The average OD₆₀₀ from two samples was plotted versus time.

Spot Dilution

Spot dilution was conducted as described previously (Rubio & Gross, 2023).

Auxin Induced Degradation

Cells expressing the F-box protein osTIR1 in combination with a degron-tagged protein were grown in YPDA medium to mid-log phase and indole-3-acetic acid (IAA) was then added to a final concentration of 1 mM IAA stocks [10 mg/mL (57 mM)] were prepared fresh in 95% ethanol and filter-sterilized before use. For immunoblot analysis, cells were treated for varying times up to 1 h prior to metabolic arrest using 20 mM sodium azide, followed by cell harvesting. 0 min control samples were treated with vehicle alone (1.7% v/v ethanol). For growth curve analysis, samples were kept at 30°C with constant shaking and aliquots were removed at various timepoints to monitor OD₆₀₀. For 3C analysis, cells were similarly grown in YPDA to OD₆₀₀=0.6, then treated with 1 mM IAA for either 30 min (LRY016 and LRY100) or 40 min (LRY102) prior to cell cross-linking and subsequent cell harvesting.

Reverse Transcription-qPCR (RT-qPCR)

RT-qPCR was conducted as previously described using 25 mL cell culture aliquots (Rubio & Gross, 2023). Transcription was terminated via addition of sodium azide to a final concentration of 20 mM (Lee & Garrard, 1991). PCR primers used are listed in Supplemental File 1 – Table 4.

Immunoblot Analysis

Immunoblotting was performed as previously reported using 20 mL cell culture aliquots (Rubio & Gross, 2023).

Chromatin Immunoprecipitation

Chromatin immunoprecipitation was performed as previously described (Chowdhary et al., 2019) with modifications. Briefly, cells from a 50 mL mid-log culture were exposed to 8.5% v/v ethanol and fixed using 3.01% formaldehyde (HCHO), resulting in a net concentration of 1%. Glycine was then added to 0.363 M glycine to quench excess formaldehyde. We note that HCHO is consumed in a reaction with ethanol. Since 8.5% ethanol equals 1.46 M, acetal formation at 1:2 stoichiometry consumes 0.730 M (2.01%) of HCHO, leaving an effective concentration of 0.363 M (1%). The reaction resulting in formation of an acetal is illustrated in Figure 11.

Figure 11.

For heat shock, cells from a 50 mL mid-log culture were fixed with 1% (0.363 M) formaldehyde following 39°C upshift for the times indicated (Figure 4; Figure 4 – figure supplements 1 and 2). Glycine was then added at a final concentration of 0.363 M to quench unreacted formaldehyde. Chromatin lysates, prepared as previously described (Chowdhary et al., 2019), were incubated using 20% of the lysate with one of the following antibodies: 1.5 µL of anti-Hsf1 (Erkine et al., 1996), 1.5 µL anti-Rpb1 antiserum (Zhao et al., 2005), or 1 µL of H3 antibody (Abcam, ab1791). Incubation of lysate with antibody was done for 16 h at 4°C. Chromatin fragments bound to antibody were captured on Protein A-Sepharose beads (GE Healthcare) for 16 h at 4°C. Wash, elution, and DNA purification were conducted as described (Chowdhary et al., 2019).

The ChIP DNA template was quantified by qPCR (7900HT Fast Real Time PCR System, Applied Biosystems). A standard curve was generated using genomic DNA and ChIP DNA quantities were deduced by interpolation. The qPCR signal for each primer combination was normalized relative to the corresponding signal arising from the input DNA. Primers used in ChIP analysis are listed in Supplemental File 1 – Table 5.

Taq I chromosome conformation capture (Taq I-3C)

TaqI-3C was performed as previously described (Rubio & Gross, 2023). A master cell culture was grown at 30°C in YPDA from OD₆₀₀=0.15 to a final OD₆₀₀=0.8. Aliquots of 50 mL were used for each condition. Heat shock and ethanol stress were conducted as described above. Primers for analysis of 3C templates are listed in Supplemental File 1 – Table 6.

Fluorescence Microscopy

Widefield Fluorescence Microscopy

For Figure 7 – figure supplement 1 [B, C] and Figure 8, cells were grown at 30°C in Synthetic Complete Dextrose (SDC) medium supplemented with 0.1 mg/mL adenine to early log phase. From this culture, 90 µL were removed and cells were immobilized onto a concanavalin A (ConA, Sigma Aldrich, 100 μg/mL in ddH₂O)-coated coverslip for 20 min. The medium was then removed and replaced by either SDC or SDC + 8.5% v/v ethanol. The coverslip was mounted onto a concave microscope slide to reduce media evaporation (2-Well Concavity Slide, Electron Microscopy Sciences). Images were acquired using an AX70 Olympus epifluorescence microscope across 11 z-planes with 0.5 µm of interplanar distance. Filter set 89021 (Chroma Technology) and a Photometrics Prime 95B camera were used to image GFP and mCherry. SlideBook Software version 6.0.15 (Intelligent Imaging Innovations) was used for image capturing and z-axis stepping motor operation (Ludl Electronic products).

For heat shock, No Stress (NS) control images were taken using an Olympus Ach 100/1.25-NA objective coupled to a heating device (Bioptechs objective heater system). Heat-shocked sample imaging was done by rotating the objective away from the coverslip, switching the heating system on, and allowing it to reach 38°C before returning the objective to the coverslip for an instantaneous heat shock. The same field of view on the coverslip was imaged for both NS and HS timepoints. For ethanol stress, an Olympus UPlanFl 100/1.4-NA objective was used for image acquisition. Analysis of the images after acquisition was done using ImageJ (v. 1.52).

In the HSP104-TMA10 coalescence analysis, cells in which the fluorophore-tagged genes had undergone replication (two green or two red fluorescent spots, indicative of late S/G2 phase) were excluded from the analysis. The locations of the two tagged loci were analyzed over 11 different z-planes, encompassing the whole nucleus. Cells were deemed coalescence positive if they displayed colocalized, non-resolvable fluorophore signals with a distance of <0.4 µm between centroids.

Spinning Disk Confocal Microscopy

For all other imaging, cells were grown as described above and image acquisition was done using an UPlan Apo 100x/1.50 NA objective in an Olympus Yokogawa CSU W1 Spinning Disk Confocal System coupled to sCMOS cameras (Hamamatsu Fusion) controlled by cellSens Dimension software. A 50 µm pinhole disk was used for imaging in combination with 10% of laser power employed for excitation using 405 nm, 488 nm, and 561 nm lasers. Z-stacks were captured as for widefield fluorescence. For heat shock, cells were attached to a VAHEAT substrate (Interherence GmbH, Germany) (Icha et al., 2022) using ConA as above. The substrate was mounted onto the VAHEAT holder and control images were captured before heat was applied. Samples were instantaneously heated to 39° or 42°C. The substrate reservoir was covered with a coverslip to prevent media evaporation. Imaging was done over multiple timepoints as indicated in the figures. For ethanol stress, cells were mounted onto a coverslip as above, petroleum jelly was used to hold the coverslip against the concave slide and to reduce media evaporation. The slide was inverted and placed on the stage for visualization at room temperature.

Image reconstruction and analysis was done using FIJI/ImageJ (v. 1.53t) (Schindelin et al., 2012). Hsp104-BFP (LRY033) and Hsf1-mNeonGreen (LRY040) foci count and volume were performed using the “Cells” feature in Imaris v.10.0.0. The background fluorescence was used to delimit the region of interest (ROI). This ROI was then used to quantify the number of foci per cell. For Hsp104, the ROI (whole cell) was delimited to volumes between 20 and 140 µm³ (Uchida et al., 2011). Foci size was measured to an average of 0.4 µm in diameter, which was used to train Imaris to recognize foci of distinct sizes [“Different Spot Sizes (Region Growing)” option]. A similar approach was taken to identify the nucleus (as the cell) and the Hsf1 foci (as nucleus) using the “Cells” feature in Imaris. Nuclei with volume of 10 µm³ ±3.7 µm³ were used for the analysis (Larson et al., 2011). Data obtained from these analyses were plotted using GraphPad Prism 8. Plot profile for signal arising from Hsf1-mNeonGreen, Sis1-mKate, and Hsp104-BFP was done using ImageJ/FIJI.

For analysis of transcription in VPY705, 24xMS2-HSP104 mRNA was visualized using the signal arising from MCP-mCherry binding to the chimeric transcript. Cells were interrogated for transcription by assaying the presence of an mCherry focus adjacent to HSP104-LacO₂₅₆ bound by LacI-GFP. The high background signal from MCP-mCherry made it difficult to assess the localization of transcripts once they dissociated from the HSP104-tagged gene.

Statistical Tests

The statistical significance of the differences in mean values for a variety of assays was determined as specified in the figure legends using GraphPad Prism 8.

Acknowledgements

We thank Drs. Kelly Tatchell, Lucy C. Robinson, Rini Ravindran, Eric First, Surabhi Chowdhary and David Pincus for helpful discussions and technical advice; Drs. Vickky Pandit and Amoldeep Kainth for strain construction; Paula Polk for help with quantitative PCR; and Drs. Donna and Jason Brickner, Surabhi Chowdhary, Amoldeep Kainth and David Pincus for generously providing yeast strains. This work was supported by NIH grants R15 GM128065 and R01 GM138988 awarded to D.S.G. and an Ike Muslow predoctoral fellowship awarded to L.S.R.

Author Contributions

Linda S. Rubio, Conceptualization, Investigation, Visualization, Methodology, Formal Analysis, Writing – original draft, Writing – review and editing; Suman Mohajan, Investigation, Visualization, Methodology, Formal Analysis; David S. Gross, Conceptualization, Resources, Visualization, Validation, Supervision, Funding Acquisition, Writing – original draft, Writing – review and editing.