Functional tug of war between kinases, phosphatases, and the Gcn5 acetyltransferase in chromatin and cell cycle checkpoint controls

Abstract Covalent modifications of chromatin regulate genomic structure and accessibility in diverse biological processes such as transcriptional regulation, cell cycle progression, and DNA damage repair. Many histone modifications have been characterized, yet understanding the interactions between these and their combinatorial effects remains an active area of investigation, including dissecting functional interactions between enzymes mediating these modifications. In budding yeast, the histone acetyltransferase Gcn5 interacts with Rts1, a regulatory subunit of protein phosphatase 2A (PP2A). Implicated in the interaction is the potential for the dynamic phosphorylation of conserved residues on histone H2B and the Cse4 centromere-specific histone H3 variant. To probe these dynamics, we sought to identify kinases which contribute to the phosphorylated state. In a directed screen beginning with in silico analysis of the 127 members of yeast kinome, we have now identified 16 kinases with genetic interactions with GCN5 and specifically found distinct roles for the Hog1 stress-activated protein kinase. Deletion of HOG1 (hog1Δ) rescues gcn5Δ sensitivity to the microtubule poison nocodazole and the lethality of the gcn5Δ rts1Δ double mutant. The Hog1–Gcn5 interaction requires the conserved H2B-T91 residue, which is phosphorylated in vertebrate species. Furthermore, deletion of HOG1 decreases aneuploidy and apoptotic populations in gcn5Δ cells. Together, these results introduce Hog1 as a kinase that functionally opposes Gcn5 and Rts1 in the context of the spindle assembly checkpoint and suggest further kinases may also influence GCN5's functions.


Introduction
Eukaryotic genomes are packaged into the complex known as chromatin, which contains DNA and associated proteins. Dynamic modification, movement, and assembly of chromatin allow efficient organization and flexible genomic architecture to provide structural support and modulate accessibility. Such dynamic structural changes are indispensable for many essential cellular processes, including transcriptional regulation, cell division, and DNA replication and repair.
The basic repeat unit of chromatin is the nucleosome, containing two heterodimers of histones H2A, and H2B, a tetramer of histones H3 and H4, and a unit length of DNA wrapped around these core histones (Kornberg and Lorch, 1999;McGinty and Tan, 2015). Variant histones provide distinct functions, such the yeast H3 centromere-specific variant Cse4 (Choy et al. 2012). Histones undergo extensive post-translational modifications, mediated by chromatin modifying enzymes which are responsible for the covalent attachment and removal of small chemical or protein marks (Strahl and Allis 2000;Kouzarides 2007;Henikoff and Shilatifard 2011;Tessarz and Kouzarides 2014;Millan-Zambrano et al. 2022). These marks promote regulation of chromatin structure, interactions between histones and DNA, and recruitment of nonhistone proteins to specific locations in the genome (Nickel et al. 1989;Lu et al. 2008;Mishra et al. 2016), thereby contributing significantly to chromatin dynamics for cellular functions.
Enhancing their complexity, histone modifications do not function in isolation. Crosstalk between histone modifications adds a level of control to chromatin modifications. As one example, previous studies linked methylation of histone H3 residues to ubiquitination of H2B (Wyce et al. 2007). Dynamic ubiquitination and de-ubiquitination of H2B-K124 at promoters and within open reading frames modulate the accumulation of methylation on histone H3 throughout the transcribed region, ensuring optimal transcription. Such dynamic chromatin modifications and the enzymes that mediate them together form a complex network that regulates growth and mediates responses to stress conditions and nutrient availability. (Altheim and Schultz, 1999;Viéitez et al. 2020;Hsieh et al. 2022).
Among modifying enzymes, Gcn5 is a conserved lysine acetyltransferase (K/HAT) that modifies histones H2B and H3 (Grant et al. 1997;Zhang et al. 1998;Suka et al. 2001), while functioning as an enzymatic subunit of several distinct complexes (Grant et al. 1997;Eberharter et al. 1999;Sterner et al. 2002;Helmlinger et al. 2021). Since histone acetylation can disrupt interactions between the negatively charged DNA and the positively charged lysine residues, Gcn5-mediated histone acetylation is crucial for many processes, notably transcriptional regulation (Imoberdorf et al. 2006;Xue-Franzén et al. 2013), and responses to DNA damage (Pai et al. 2014). Gcn5 is also an important player in cell cycle progression. Gcn5-influenced histone gene expression and nucleosome assembly are crucial for DNA replication and chromatin organization during early stages of the cell cycle Gong et al. 2020). Indeed, previous research confirmed that GCN5 mutants suffer slow progression through the G1/S phase transition (Petty et al. 2016). Later in the cell cycle, Gcn5 interacts with centromeres (Vernarecci et al. 2008) and regulates gene expression crucial for mitotic progression (Krebs et al. 2000). In addition, GCN5 mutants display phenotypes such as sensitivity to microtubule destabilization and slow progression through the G2/M transition (Zhang et al. 1998;Howe et al. 2001;Vernarecci et al. 2008;Petty et al. 2018).
Whereas the functions and biological significance of Gcn5-mediated histone acetylation are well established, defining the significance of interactions between Gcn5 and other proteins remains an area of active investigation (Petty and Pillus, 2021). Genetic screens provide a powerful tool to identify factors that functionally interact with one another, revealing both positive and negative relationships. In S. cerevisiae, deletion of GCN5 is tolerated but causes a variety of phenotypes (Cherry et al. 2012;Petty et al. 2016). By introducing altered gene dosage or additional mutations in a gcn5Δ mutant and selecting for alleviation of gcn5Δ phenotypes, we previously identified RTS1, a gene encoding a regulatory subunit of protein phosphatase 2A (PP2A) (Healy et al. 1991;Shu et al. 1997), as a high copy suppressor of a distinct set of gcn5Δ phenotypes (Petty et al. 2016).
Overexpression of RTS1 in gcn5Δ mutants rescues temperature sensitivity, DNA damage sensitivity, poor growth on nonfermentable carbon sources and chromosome segregation defects (Petty et al. 2016). Notably, histone H2B threonine residue 91 (H2B-T91) is required for the Rts1-Gcn5 interaction. Whereas the nonphosphorylatable H2B-T91A mutant hinders the rescue of gcn5Δ phenotypes by RTS1 overexpression, the phosphomimic H2B-T91D or H2B-T91E mutations cause lethality, strongly suggests that orchestrated dynamic histone phosphorylation is crucial for growth and stress-response functions in gcn5Δ mutant cells.
A closer look at the Gcn5-Rts1 interaction in the context of chromosome segregation indicated that deletion of GCN5 abolishes the centromeric localization of Rts1, which occurs during the spindle assembly checkpoint to promote proper tension sensing and faithful chromosome segregation. A serine residue on the centromere-specific histone Cse4-S180 was found to play a role in Rts1 centromeric localization (Petty et al. 2018). The mutant cse4-S180A restores Rts1 centromeric localization whereas the phosphomimetic mutant cse4-S180EE causes synthetic lethality in gcn5Δ (Petty et al. 2018). These results suggest that tightly regulated dynamic phosphorylation of Cse4 residues is crucial for proper response to nocodazole in gcn5Δ mutants.
Kinases and phosphatases together create dynamic changes in histone phosphorylation states, exerting diverse effects on the structure and function of chromatin (Rossetto et al. 2012). Since the functional interaction between PP2A Rts1 and Gcn5 requires intact phosphorylatable residues, we hypothesized that a kinase may also function as part of a Rts1-Gcn5 interaction network. (Fig. 1a). Since overexpression of Rts1 potentially biases the chromatin phosphorylation balance toward the dephosphorylated state, loss of a kinase that shares a substrate with PP2A Rts1 might achieve the same effect. Therefore, the deletion of a kinase involved in the Rts1-Gcn5 interaction might suppress similar phenotypes in the gcn5Δ background as does overexpression of RTS1. To test this hypothesis, a screen considering all of the annotated yeast kinases in the yeast genome was performed.
By examining genetic interactions between kinase mutants and gcn5Δ, we have identified the Hog1 mitogen-activated protein kinase (MAPK) to be centrally involved in the Rts1-Gcn5 interaction. MAPKs are deeply conserved serine/threonine kinases in eukaryotes critical for signal transduction. In yeast, Hog1 functions in the high osmolarity glycerol (HOG) pathway and in governing cellular responses to changes in extracellular osmolarity (Brewster et al.,1993). When extracellular stresses are detected by membrane sensors, a cascade is activated, resulting in the phosphorylation and activation of Hog1 stress-response functions (Maeda et al. 1995;Posas et al. 1996;Maayan et al. 2012).
During the intracellular response to hyperosmotic stress, Hog1 interacts with cytoplasmic proteins (Bouwman et al. 2011), chromatin modifiers (Pérez-Martínez et al. 2020, and transcription factors (Rep et al. 1999(Rep et al. , 2000, to modulate the concentration of intracellular osmolytes. Furthermore, Hog1 also interacts with components of cell cycle checkpoints to induce a transient cell cycle arrest at G1 (Escoté et al. 2004), S (Yaakov et al. 2009), G2 (Clotet et al. 2006), and M phases (Tognetti et al. 2020), until the stressful condition is resolved. Whereas the stress-response-related interactions of the MAPK and its partners are well documented in budding yeast, defining the functional role of MAPK chromatin modification and potential crosstalk remains an area of active investigation.
In this study, hog1Δ was identified as a suppressor of gcn5Δ mutants' sensitivity to microtubule destabilization by nocodazole and the synthetic lethality of the gcn5Δ rts1Δ double mutant. The previously pinpointed histone H2B-T91 residue was required for the Hog1-Gcn5 interaction, further indicating that Hog1 and PP2A Rts1 together mediate dynamic modifications crucial for gcn5Δ viability. Cell cycle analysis suggests that hog1Δ alleviates defective cell cycle progression in cells lacking GCN5 under microtubule destabilizing conditions. Together, these results point to a functional interaction between Hog1, Gcn5, and PP2A Rts1 , in which Hog1 activity opposes Gcn5 and PP2A Rts1 in the context of the spindle assembly checkpoint.

Yeast methods
Standard protocols for yeast growth were used (Guthrie and Fink, 1991). All yeast strains were grown at 30°C, except where indicated. For dilution assays, yeast cultures were adjusted to 1.0 OD 600 /ml and 1:5 serial dilutions were plated. Plates were photographed after 2-5 days incubation, as indicated in the figure legends. To test the effects of specific histone mutations, histone shuffle strains were constructed, with the WT histone shuffle strain (LPY 14461 hht1-hhf1Δ::kanMX hta1-htb1Δ::natMX hta2-htb2Δ::HPH) crossed to hog1Δ gcn5Δ (LPY 21367) to produce the hog1Δ hht1-hhf1Δ::kanMX hta1-htb1Δ::natMX hta2-htb2Δ::HPH (LPY 23129) (Petty et al. 2016). Genetic screens demonstrated that specific H2B histone residues (blue) are important for the Rts1-Gcn5 interaction. Together, these data suggest that loss of Gcn5-mediated histone acetylation potentially disrupts the dynamic phosphorylation balance of specific histone residues which can be restored by the overexpression of RTS1. Thus, there may be a kinase or kinases (denoted as KinX) functioning at the RTS1-GCN5 interaction nexus. The Kinase-Rts1-Gcn5 interaction nexus could take place in a variety of contexts, such as chromosome segregation, transcriptional regulation, and the DNA damage response. Further, these interactions may be mediated directly through histones or other protein substrates (ProX). We hypothesize that deletion of such a kinase would rescue the same set of gcn5Δ phenotypes suppressed by RTS1 overexpression, whereas overexpression of the gene encoding the kinase may result in lethality in gcn5Δ cells. b) 13 kinases of interest were selected based on nuclear localization, cellular functions, interaction with GCN5 and/or RTS1, and viability of the gcn5Δ kinXΔ double mutant. An in silico screen of the 127 member yeast kinome started with determination of nuclear localization as curated in the Saccharomyces Genome Database. Thirty-eight kinases were identified. Because RTS1 overexpression specifically suppresses various gcn5Δ sensitivities the kinase candidates were further categorized based on their involvement in nutrient and stress-response, cell cycle progression, DNA damage, and interactions with GCN5 and RTS1. Eighteen kinases met these criteria. Since the deletion of the nuclear kinase is hypothesized to rescue gcn5Δ phenotypes, the gcn5Δ kinxΔ double mutant must be viable. Therefore, two essential kinases were excluded. Of the remaining 16, published or preliminary results indicated that three caused severe sickness or lethality in gcn5Δ background, yielding 13 nuclear kinases for extended phenotypic and genetic analysis. c) Distinct patterns of phenotypic interaction with gcn5Δ are observed. A heat map to summarize the screen results is shown. Semiquantitative dilution assays were performed with WT (LPY 5), gcn5Δ (LPY 13319), and the gcn5Δ kinxΔ mutant under a variety of stress conditions. Methyl methanesulfonate (MMS, 0.015 and 0.02%), hydroxyurea (HU, 0.05 and 0.1 M), and camptothecin (CPT, 3 μg/ml and 6 μg/ml) were used to induce distinct forms of DNA damage. Nocodazole (NOC, 1 μg/ml and 2 μg/ml) was used to induce chromosome segregation defects, calcofluor white (10 μg/ml) was used to produce cell wall-related stress, and 1 M KCl and sorbitol media were used to test for the sensitivity of salt and hyperosmolarity stress. The cells were scored after growth at 30°C for 3 days. A suppression of gcn5Δ phenotypes is shown as a blue cell in the heat map, with dark blue indicating strong suppression, and light blue indicating partial suppression. Exacerbation of gcn5Δ phenotypes is shown in red, with similar quantitative comparisons. When the gcn5Δ kinxΔ growth is comparable to gcn5Δ, a white cell is used to indicate no interaction. Dark and light gray cells denote essential gene deletions and gene deletions that cause lethality in the gcn5Δ background, respectively. The nonessential kinase deletions are shown in alphabetical order.
Cells analyzed for budding index were taken from log-phase cultures and fixed with 70% cold ethanol. In independent experiments, more than 500 cells were examined for each variable.

Media preparation
Yeast media were prepared with standard protocols, with modifications as noted (Guthrie and Fink, 1991;Sherman, 2002). Hydroxyurea (HU) plates were prepared by adding filter sterilized 1 M HU stock solution to synthetic complete (SC) media to a final concentration of 0.1 M or 0.05 M. Methyl methanesulfonate (MMS) plates were prepared at final concentrations of 0.01-0.025% in SC. Camptothecin (CPT) was added from a 3 mg/ml stock dissolved in DMSO to Yeast extract Peptone Adenine Dextrose (YPAD) media potassium phosphate buffered to pH 7.5 to a final concentration of 3-12 µg/ml. DMSO plates were prepared as a solvent control. Nocodazole (NOC) plates were prepared by adding 10 mg/ml of nocodazole dissolved in DMSO to YPAD to a final concentration of 1-2 µg/ml. KCl and sorbitol plates were prepared by adding filter sterilized KCl or sorbitol solutions to YPAD media to a final concentration of 1 M KCl or 1 M sorbitol. 5-FOA plates were prepared by adding filter sterilized 5-FOA solution to SC media to a final concentration of 0.5 × (∼5.7 mM). 0.5 × of standard 5-FOA concentrations was used due to increased sensitivity in the gcn5Δ background. We note that variable concentrations of nocodazole and 5-FOA were used in some experiments, due to the significantly differing sensitivities of wildtype and mutant strains. Concentrations used are noted for individual experiments in the figure legends.

Flow cytometry
Samples were taken from exponentially growing cell cultures and fixed with cold 70% ethanol at 4°C overnight. Following fixation, the samples were treated with RNAse A (1 mg/ml) at 37°C overnight and stained with propidium iodide (PI) at 4°C overnight. All experiments used a BD ACCURI C6 flow cytometer for analysis. Fluorescence was detected with the FL2A channel with fluidics set to slow and an event limit of 30,000 events. During measurement, collected data were visualized in the BD ACCURI C6 flow cytometer histogram software. Microsoft Excel was used for statistical analysis and presentation. For analysis of sub-G1 populations, significance was evaluated with single-factor ANOVA. The significance of the data in 4(c) is P < 0.05 for all comparisons.

Nocodazole sensitivity assay
10 mg/ml nocodazole dissolved in DMSO was added to log-phase cultures for a final concentration of 1-2 µg/ml. Samples were collected every hour for 6-8 hours following the nocodazole treatment. The samples collected were analyzed with flow cytometry as described above.

CRISPR mutagenesis
The hog1-T174A catalytic mutation was introduced via CRISPR-mediated mutagenesis in a protocol adapted from Ran et al. (2013). CRISPR Direct (https://crispr.dbcls.jp/) was used for the identification of PAM sequences and guide RNA location within the HOG1 sequence. The plasmid pML104, which encodes the Cas9 protein (a gift from L. McDonnell, UCSD), was digested with BclI and SwaI. Oligonucleotides containing a BclI 5′ overhang, HOG1 gRNA sequence, and Cas9 scaffold sequence were hybridized and subsequently ligated into pML104. The homology directed repair template was synthesized via PCR with a pair of overlapping oligonucleotides containing the hog1-T174A mutation and the PAM disrupting silent mutation. The PAM (NGG) sequence was disrupted with a silent mutation, and the newly introduced codon at the PAM site and on threonine 174 were confirmed to be one of the preferred codons for yeast. The constructed CRISPR plasmid (1 µg) and the HDR templates (200 ng) were transformed into yeast cells via the lithium acetate method. hog1-T174A cells were selected based on positive growth on URA-media and a lack of growth on 1 M sorbitol media. The genotype was confirmed by sequencing PCR amplified products from genomic DNA (adapted from Hoffman and Winston, 1987). Oligonucleotides used for CRISPR mutagenesis are presented in Supplementary Table 3.

A genetic screen identifies kinases with potential interactions with GCN5
To identify kinases that might function in the GCN5-RTS1 interaction, we first surveyed the 127 genes encoding protein kinases (Rubenstein and Schmidt, 2007) in budding yeast via the Saccharomyces Genome Database (Fig. 1b). Considering that the GCN5-RTS1 nexus involves specific histone residues (Petty et al. 2016), we hypothesized that a kinase participating in this interaction was likely to include nuclear functions. Of the yeast kinases, 38 genes encoding nuclear kinases were identified. Analysis of curated information for kinases with functions in DNA damage repair, cell cycle progression, stress/nutrient responses, and interaction with Gcn5 and Rts1 identified 18 candidate genes meeting these criteria (denoted as KINX) ( Fig. 1b): CDC28, CHK1, CKA1, CKA2, CLA4, CTK1, DUN1, HOG1, IPL1, KSS1, PHO85, RAD53, SCH9, SLT2, SNF1, STE20, SWE1, and TEL1 (Cherry et al. 2012).
To test for functional interactions with GCN5, classical genetic analysis was performed in which a double mutant was constructed with gcn5Δ and the nonessential genes encoding the nuclear kinase candidates (denoted as gcn5Δ kinxΔ). Diploids were constructed with the gcn5Δ parent containing a CEN-URA3-GCN5 plasmid. After sporulation and dissection, all viable spores were genotyped. Double mutants were identified and to confirm the viability of gcn5Δ kinxΔ strains prior to phenotypic analysis, they were plated on medium containing 5-Fluoroorotic acid (5-FOA), which is used for robust counter-selection against URA3 expression (Boeke et al. 1987). Growth on 5-FOA thus indicates that the double mutant is viable as it can survive without the CEN-URA3-GCN5 plasmid. Since CDC28 and IPL1 are themselves essential in yeast, the corresponding double null mutants were not included in the initial screen (Fig. 1c). Furthermore, based on preliminary results and available literature pointing to synthetic lethalities and sickness between loss of Gcn5 and the deletion of CLA4, RAD53, or SLT2, analysis of those double mutants was not a Note that five kinases, in parentheses, were not characterized further due to their essential nature or published or preliminary data indicating negative synthetic interactions with gcn5Δ.
Any rescue of gcn5Δ phenotypes by deletion of a gene encoding a nuclear kinase implies a potential interaction between that kinase and Gcn5. The phenotypes of the double mutants were therefore evaluated with various challenges, including elevated growth temperature, hyperosmolarity, DNA damage induction, and compounds that interfere with cell cycle progression (Petty et al. 2016). Figure 1c summarizes the results of this phenotypic survey. Deletions of CKA1, CKA2, SWE1, and KSS1 were identified as suppressors of gcn5Δ mutant's sensitivity to DNA damage induced by hydroxyurea (HU), methyl methane sulfonate (MMS), and camptothecin (CPT) (Fig. 1c). chk1Δ was found to improve gcn5Δ's growth at elevated temperatures and increased resistance to calcofluor white (CFW), a compound that causes cell wall damage (Fig. 1c). Among the mutants, hog1Δ had the strongest rescue of gcn5Δ sensitivity to nocodazole, a compound which can cause chromosome segregation defects (Fig. 1c). Like RTS1 overexpression, which restores a variety of gcn5Δ phenotypes, deletion of the nuclear kinases resulted in unique patterns of resistance and sensitivity, indicating that multiple kinases may contribute to distinct aspects of Gcn5 function.

Hog1 and the Rts1-Gcn5 interaction is linked to spindle function
Previous results demonstrated that increased RTS1 dosage restores growth in nocodazole to strains lacking GCN5, linking the Rts1-Gcn5 interaction to spindle assembly checkpoint functions (Petty et al. 2018). In this study, we tested the growth phenotypes of the gcn5Δ kinxΔ double mutants on medium containing nocodazole. As noted above, deletion of HOG1 was observed to significantly improve growth of gcn5Δ mutants under microtubule destabilizing conditions (Fig. 2a, top). We next tested if loss of Hog1 catalytic activity is the factor that underlies the rescue. The catalytically inactive Hog1 mutant was constructed by replacing the threonine 174 residue on the Hog1 activation loop with alanine via CRISPR-based mutagenesis. The hog1-T174A mutation prevents the activation of Hog1 catalytic activity by the upstream MAPKK (Maeda et al. 1995;Maayan et al. 2012). The hog1-T174A gcn5Δ double mutant was constructed genetically and tested for growth on nocodazole. Indeed, loss of Hog1 catalytic activity is sufficient to restore growth in the gcn5Δ background upon the nocodazole challenge (Fig. 2a, bottom). Further, loss of Hog1 catalytic activity improves nocodazole resistance not only in gcn5Δ mutants, but also has a modest effect in otherwise wildtype (WT) cells (Fig. 2a, bottom). Suppression of gcn5Δ sensitivity to nocodazole was somewhat weaker in hog1-T174A gcn5Δ when compared to the hog1Δ gcn5Δ mutant, an effect likely due to disruption, but not complete loss of catalytic activity in the point mutant. Overall, these results suggest that compromising Hog1 kinase activity generally reduces nocodazole sensitivity.
Concurrent loss of Gcn5 and Rts1 induces synthetic lethality (Petty et al. 2016). Because both RTS1 overexpression and deletion of HOG1 act as suppressors of gcn5Δ nocodazole sensitivity, we hypothesized that these opposing phenotypes may indicate that loss of Hog1 could rescue the rts1Δ gcn5Δ synthetic lethality. The triple mutant hog1Δ rts1Δ gcn5Δ, bearing a WT URA3-GCN5 plasmid was constructed genetically. Triple mutants recovered from the cross were plated onto 5-FOA medium to determine if the WT URA3-GCN5 plasmid could be lost. We found that the hog1Δ rts1Δ gcn5Δ was viable (Fig. 2b), indicating that concurrent deletion of HOG1, indeed, rescues the rts1Δ gcn5Δ synthetic lethality.
We next tested if loss of Hog1 kinase activity provided the specific mechanism for rescue of rts1Δ gcn5Δ synthetic lethality. In this case, the CEN-HIS3-hog1-T174A plasmid was transformed into the CEN-URA3-GCN5-bearing hog1Δ rts1Δ gcn5Δ strain. On 5-FOA medium, the hog1Δ rts1Δ gcn5Δ strain showed growth comparable to WT on 5-FOA medium, indicating that the catalytic inactivation of HOG1 is what drives suppression of the rts1Δ gcn5Δ synthetic lethality (Fig. 2c). Further, the hog1Δ rts1Δ gcn5Δ mutant bearing the CEN-HIS3-HOG1 plasmid showed compromised growth upon loss of the CEN-URA3-GCN5 plasmid (Fig. 2d). Together, these results support the idea that loss of Hog1's catalytic activity, and not another property of the enzyme, is what rescues rts1Δ gcn5Δ synthetic lethality, pointing to a key enzymatic role in the Rts1-Gcn5 interaction.

H2B threonine 91 is required for hog1Δ rescue of gcn5Δ nocodazole sensitivity
Our previous studies established that H2B-T91 plays an important role in the Gcn5-Rts1 interaction. This conserved residue is phosphorylated in vertebrates (phosphosite.com), and in yeast, the phosphodeficient H2B-T91A mutant hinders RTS1 dosagedependent rescue of gcn5Δ phenotypes (Petty et al. 2016). We tested if the same histone residue is involved in the Gcn5-Hog1 interaction. Confirming previous observations, we found that H2B-T91D/E phosphomimetic mutations cause lethality in all backgrounds (Fig. 3a). The htb1-T91A allele was found to hinder the growth of hog1Δ gcn5Δ growth on nocodazole-containing media, indicating that the phosphodeficient histone mutation partially abolishes the hog1Δ rescue of gcn5Δ sensitivity to nocodazole (Fig. 3b). Moderate rescue of gcn5Δ sensitivity to nocodazole remains in the hog1Δ gcn5Δ htb1-T91A strain at lower nocodazole concentrations (Fig. 3b). Together, these results confirmed that mutation of histone H2B-T91 exerts an effect on the Hog1-Gcn5 interaction, indicating that the phosphorylatable H2B-T91 residue plays an important role in the Hog1-Gcn5 interaction. Both Gcn5-Rts1 (Petty et al. 2016) and Gcn5-Hog1 (Fig. 3b) functional interactions rely on H2B-T91, further supporting the hypothesis that Hog1 and PP2A Rts1 together mediate dynamic histone modifications crucial for gcn5Δ mutants.
Beyond the canonical core histones, residues have been identified in the centromere-specific histone Cse4 that hinder RTS1 dosage-dependent suppression of gcn5Δ DNA damage sensitivity but improve gcn5Δ resistance to nocodazole (Petty et al. 2018). Therefore, the effects of these, cse4-S180A and cse4-S135A, were tested in hog1Δ gcn5Δ mutants. Strains of hog1Δ gcn5Δ containing the integrated cse4-S180A and cse4-S135A alleles or WT CSE4 were constructed and plated onto nocodazole-containing media. No difference was observed between the growth of the hog1Δ gcn5Δ cse4-S135A, hog1Δ gcn5Δ cse4-S180A, and hog1Δ gcn5Δ CSE4 strains (Fig. 3c). Thus, we concluded that phosphodeficient cse4 alleles do not negatively affect the Hog1-Gcn5 interaction.
hog1Δ specifically restores gcn5Δ viability in response to nocodazole RTS1 overexpression restores cell cycle progression in gcn5Δ mutants at both the G1/S transition and the transition between mitosis and G1 (Petty et al. 2016(Petty et al. , 2018. Considering Hog1 as a new player in the Rts1-Gcn5 interaction, we hypothesized that deletion of HOG1 rescues the gcn5Δ sensitivity to nocodazole by improving cell cycle progression. To test this possibility, asynchronous cell cultures of gcn5Δ and hog1Δ gcn5Δ were incubated overnight and diluted to 0.1 OD 600 with fresh media and monitored after the resumption of exponential growth. The DNA content of these cell populations was evaluated via flow cytometry of propidium iodide-stained cells. Both gcn5Δ and hog1Δ gcn5Δ show an increased G2/M population compared to WT (Fig. 4a, left), indicating that under normal growth conditions, both gcn5Δ and hog1Δ gcn5Δ have defective cell cycle progression.
In S. cerevisiae, the budding cycle is directly related to the cell cycle (Zettel et al. 2003). Abnormal budding has been reported in gcn5Δ mutants in our earlier studies, along with its documentation in genome-wide studies and those directed at understanding GCN5's role in centromere and kinetochore functions  , and hog1Δ gcn5Δ (LPY 23128) strains were constructed, bearing a URA3-HTB1 plasmid and each of three HIS3-htb1 plasmids (htb1-T94A, htb1-T91E, or htb1-T91D). Strains were plated onto HIS-URA-(growth) and HIS-and incubated at 30°C for three days. The H2B-T91A mutation was found to be viable in all strains. b) htb1-T91A partially interferes with hog1Δ suppression of gcn5Δ nocodazole sensitivity. Strains from above with HTB1 or htb1-T91A were plated onto HIS-(growth) and YPAD + NOC plates following overnight growth in HIS-media. The HIS-plate is used as a control to ensure retention of histone shuffle plasmid. Cells were incubated at 30°C for three days. In this case, htb1-T91A abolishes hog1Δ increased resistance to NOC in WT cells. c) Phosphodeficient cse4-S180A and cse4-S135A mutations have no negative effects on hog1Δ gcn5Δ. To test effects of the centromeric specific histone mutants, cse4-S135A and cse4-S180A WT strains were crossed with hog1Δ gcn5Δ (LPY 21366) to produce the hog1Δ cse4 double mutant and the hog1Δ gcn5Δ cse4 triple mutant. Strains were plated onto YPAD (Growth) and YPAD + NOC medium at 1.5 µg/ml, 1.75 µg/ml, 2 µg/ml, and 2.25 µg/ml concentrations. Cells were incubated at 30°C for three days. Here, cse4-S180A itself increases gcn5Δ nocodazole sensitivity, confirming previous results (Petty et al. 2018).
We evaluated budding phenotypes microscopically to determine the percentage of each cell type in an asynchronous cell culture. Unbudded cells, those with small buds, and those with large buds roughly correspond to populations undergoing G1, S, and G2/M phases of the cell cycle, respectively. Cells with elongated buds and other abnormal budding patterns likely result from disruption of the budding cycle or other mutant phenotypes. In WT cultures, unbudded cells and those with small buds and large buds are found at approximately 50, 25, and 25% of the total population, respectively (Fig. 4a, right). Only a small fraction of the WT population showed elongated (1%) or abnormal buds (1.2%). Compared to the WT, gcn5Δ and hog1Δ gcn5Δ cultures contained an increased proportion of elongated buds (24.0 and 22.4%) and decreased numbers of small and large buds (Fig. 4a, right), indicating that these two mutants exhibit similar budding defects in growing populations of cells.
We next tested if deletion of HOG1 improves gcn5Δ progression through specific stages of the cell cycle. Mitotic arrest in growing cultures was induced by 1.5-3.75 µg/ml nocodazole. In this experiment, mitotic arrest in cells lacking GCN5 was induced with a lower concentration of nocodazole to avoid excessive cell death due to the strain's sensitivity. After the cells were released from the arrest (0 mins), samples were collected and stained with propidium iodide for detection of DNA content. In gcn5Δ and hog1Δ gcn5Δ cultures, accumulation of cells in G1 at 60 mins and a decrease in the G2 population at 90 mins were observed (Fig. 4b). In both and hog1Δ gcn5Δ strains also showed an increased number of dead cells compared to WT and hog1Δ, indicated by the cell population located on the left corner of the display. Budding indices were also determined for 500 cells of each strain (right). The hog1Δ mutation alone caused a slight increase in the percentages of elongated buds, with gcn5Δ having an even larger increase. No significant differences of cells with different bud shapes were observed between gcn5Δ, and hog1Δ gcn5Δ. Data shown are representative of three independent experiments. b) Exponentially growing WT (LPY 5), hog1Δ (LPY 21375), gcn5Δ (LPY 13319), hog1Δ gcn5Δ (LPY 21368) cultures were treated with nocodazole (3.75 µg/ml for WT and hog1Δ, and 1.5 µg/ml for gcn5Δ and hog1Δ gcn5Δ. Note that strains deleted for GCN5 are particularly sensitive, therefore must be treated with lower concentrations to avoid excessive cell death. These concentrations are not adequate to arrest WT cells). At T = 0 mins, cultures were released from the arrest, then samples taken every 30 mins and prepared for flow cytometry with a total of 30,000 events collected. Following nocodazole arrest and release, hog1Δ gcn5Δ shows a decreased sub-G1 population compared to gcn5Δ. Those cells with genetic content lower than 1C (fluorescence intensity < 800 AU) are denoted as sub-G1, likely corresponding to apoptotic or aneuploid cells. Whereas cell cycle progression between hog1Δ gcn5Δ and gcn5Δ is similar following nocodazole-induced mitotic arrest, a reduction in the sub-G1 population was observed in hog1Δ gcn5Δ when compared to gcn5Δ. c) Boxplots quantifying the percentages of the sub-G1 populations shown in (b). A lower percentage of sub-G1 cells was observed in hog1Δ gcn5Δ and hog1Δ strains when compared to gcn5Δ and WT. The statistical significance is P-value < 0.05 for all comparisons. asynchronous and G2/M synchronized cell cultures, we have found no significant difference in the DNA content and budding index between gcn5Δ and hog1Δ gcn5Δ ( Fig. 4a and b). These results indicate that under normal growth conditions in rich medium, hog1Δ gcn5Δ exhibit similar cell cycle defects as gcn5Δ. Thus, hog1Δ specifically rescues gcn5Δ sensitivity to end-point growth on nocodazole, without restoring normal cell cycle progression.
However, in the flow cytometry analysis, we noticed a significantly lower sub-G1 population in hog1Δ gcn5Δ compared to gcn5Δ following the nocodazole-induced mitotic arrest (Fig. 4b).
When the percentage of the sub-G1 population is plotted at each time point, the median percentages of this population in gcn5Δ and hog1Δ gcn5Δ are approximately 10 and 6%, respectively (Fig. 4c). hog1Δ greatly decreases the percentage of sub-G1 cells following nocodazole-induced arrest in both WT and gcn5Δ backgrounds (Fig. 4c). These results suggest that although hog1Δ failed to improve gcn5Δ cell cycle progression in rich media, hog1Δ otherwise alleviates the negative effects caused by nocodazole arrest in the gcn5Δ mutant. Therefore, we suspected that hog1Δ might contribute to an increased resistance to nocodazole independently of a direct role in the regulation of cell cycle progression.

hog1Δ plays a context-dependent role in gcn5Δ cell cycle progression
Nocodazole can be used in two different ways: to induce cell cycle arrest at the G2/M phase or to monitor the sensitivity of a strain to microtubule destabilization. Previous results showed that hog1Δ gcn5Δ cultures contained fewer sub-G1 cells in response to nocodazole-induced G2/M arrest. Based on these results, we hypothesized that deletion of HOG1 specifically rescues growth of cells lacking GCN5 in microtubule destabilizing conditions. Thus, using lower concentrations of nocodazole as a microtubule destabilizing agent, we tested if deletion of HOG1 improves gcn5Δ cell cycle progression. Log-phase cultures growing in rich media were treated with nocodazole at 1 µg/ml, a concentration that does not induce cell cycle arrest, and were monitored for 6 hours following nocodazole addition. In asynchronous cultures, gcn5Δ cell cycle progression was disrupted, indicated by an increase in both the sub-G1 and greater than 2C populations (Fig. 5a). Deletion of HOG1 in the gcn5Δ mutant resulted in a significant reduction in sub-G1 and greater 2C populations, consistent with a potential decrease in aneuploidy and apoptotic cells (Fig. 5a). Upon nocodazole treatment, hog1Δ gcn5Δ cultures exhibit an increased 2C population, indicating that cells are accumulating at G2/M (Fig. 5a). Similar increases in resistance to nocodazole and accumulation at G2/M were also observed in hog1Δ when compared to WT at 2 µg/ml nocodazole (not shown). These results indicate that loss of Hog1 improves cell cycle progression under microtubule destabilizing conditions. Because deletion of HOG1 improves gcn5Δ cell cycle progression under nocodazole conditions which hinder but do not halt mitotic progression, we tested if the Hog1-Gcn5 interaction might be further defined in the context of other cell cycle checkpoint functions.
In S. cerevisiae, the cell cycle and bud morphology are coupled by the morphogenesis checkpoint during the normal G2 phase. Hsl1, a key kinase regulator of the morphogenesis checkpoint, recruits Hsl7 and Swe1, a negative regulator of Cdc28, to the septin ring (Shulewitz et al. 1999). Hsl1-dependent Swe1 phosphorylation facilitates Swe1 degradation and cell cycle progression (McMillan et al. 1999). It has been observed that constitutive activation of the morphogenesis checkpoint resulting from deletion of either HSL1 or HSL7 is lethal in the gcn5Δ background (Ruault and Pillus, 2006).
Deletion of the cyclin-dependent kinase inhibitor Swe1 alleviates the synthetic lethality in gcn5Δ hsl1Δ and gcn5Δ hsl7Δ. To test if loss of HOG1 also rescues this lethality, the triple mutants hog1Δ gcn5Δ hsl1Δ and hog1Δ gcn5Δ hsl7Δ were constructed, initially supported with a CEN-URA3-GCN5 plasmid. The mutants were then plated onto 5-FOA. Whereas gcn5Δ hsl1Δ and gcn5Δ hsl7Δ were inviable on 5-FOA, hog1Δ gcn5Δ hsl1Δ and hog1Δ gcn5Δ hsl7Δ showed growth comparable to WT strains (Fig. 5b). This suppression of gcn5Δ's synthetic lethality with morphogenesis checkpoint mutants by hog1Δ suggests that the Hog1-Gcn5 interaction extends beyond the scope of mitosis and spindle assembly checkpoints.
Since both hog1Δ and swe1Δ rescue gcn5Δ hsl1Δ and gcn5Δ hsl7Δ synthetic lethality, these results indicate that in the context of the morphogenesis checkpoint, Hog1, like Swe1, may function in inhibiting cell cycle progression. Whereas in the context of the morphogenesis checkpoint, Hog1 plays a role in inhibiting the cell cycle via checkpoint activation, in the presence of nocodazole, deletion of HOG1 seems to restore normal checkpoint function. Therefore, we conclude that hog1Δ plays a diverse and contextdependent role in gcn5Δ cell cycle progression.

Discussion
Based on the previously established Gcn5-Rts1 functional interaction implicating H2B phosphorylation (Petty et al. 2016), we set out to determine if there were nuclear kinases which participated in the interaction, potentially through opposing the PP2A Rts1 contribution. Of the kinases screened, we found distinct patterns of suppression, and exacerbation of gcn5Δ phenotypes. Among the kinase mutants tested, HOG1 was found specifically to suppress gcn5Δ sensitivity to the microtubule poison nocodazole as well as its synthetic lethality with rts1Δ (Fig. 2). Furthermore, we have identified distinct HOG1-GCN5 interactions in the context of the spindle assembly checkpoint and the morphogenesis checkpoint. Like Gcn5 (Downey, 2021), Hog1 targets a large number of substrates active in a variety of cellular processes (Janschitz et al. 2019), thus it is likely that the context-dependency observed is caused by the collective effects exerted by multiple Hog1 substrates.

A role for Hog1 in the Gcn5-PP2A Rts1 nexus
Hog1 plays a part in the Gcn5-Rts1 interaction in microtubule destabilizing conditions. To determine if phosphorylatable histone residues might be required in the Hog1-Gcn5 interaction, we focused on H2B-T91, previously found to be involved in the Rts1-Gcn5 interaction. In the gcn5Δ hog1Δ mutant, we discovered that the phosphodeficient htb-T91A mutation dampens the suppression of gcn5Δ sensitivity to nocodazole by hog1Δ (Fig. 3b). These results are consistent with potential joint roles for Hog1 and PP2A Rts1 in mediating dynamic phosphorylation on H2B-T91, which is crucial for the survival of gcn5Δ mutants.
We note that the sequence surrounding H2B-T91 does not meet the established Hog1 phosphorylation consensus, which ordinarily consists of a serine or a threonine preceding a proline residue (Mok et al. 2010). Thus, Hog1 may well activate another nuclear kinase, which in turn participates in the dynamic phosphorylation of H2B-T91. Indeed, H2B-T91 in yeast has been predicted as a target of Ste20, a member of the p21-activated kinase family (Petty et al. 2016, Wang et al. 2020. Additionally, Ste20 was found to mediate histone phosphorylation involved in a number of processes. Ste20-mediated H2B-S10 phosphorylation, which plays a role in apoptosis, was found to depend on the deacetylation of H2B-K11 by Hos3 (Ahn et al. 2005). Therefore, there is precedent for Ste20 histone phosphorylation function interacting with enzymes regulating histone acetylation. Furthermore, Ste20-mediated phosphorylation of H4-S47 plays a role in the modulation of osmotic stress-responsive genes (Viéitez et al. 2020). Previous studies have shown that Ste20 interacts with Hog1 by regulating the activation of Hog1 in response to osmotic stress (Raitt et al. 2000;O'Rourke and Herskowitz, 2002). One hypothesis is that deletion of Hog1 might lead to loss of Ste20-mediated histone phosphorylation, restoring the phosphorylation balance vital for the gcn5Δ mutant. However, loss of STE20 instead was found to exacerbate gcn5Δ sensitivity to nocodazole (Fig. 1c), indicating that such a simple hypothesis does not capture the complete functional relationship between Hog1, Gcn5 and Ste20.

Hog1 functionally opposes Gcn5 and Rts1 centromeric functions during the spindle assembly checkpoint
During investigation of the mechanism underlying the Hog1-Gcn5 interaction, we found that hog1Δ suppresses potential chromosome segregation defects as evidenced by fewer cells with abnormal DNA content in both WT and gcn5Δ backgrounds in microtubule destabilizing conditions (Figs. 4d and 5a). Furthermore, hog1Δ gcn5Δ cells contain increased G2/M populations under microtubule destabilizing conditions when compared to WT strains (Fig. 5a).
During the transition from metaphase to anaphase, proper attachment of spindle fibers to the kinetochore ensures faithful segregation of sister chromatids and prevents aneuploidy and cell death (Li and Murray, 1991). Nocodazole disrupts dynamic microtubule structures leading to activation of the spindle assembly checkpoint during mitosis (Vasquez et al. 1997;Blajeski et al. 2002). Low nocodazole concentrations mainly affect the spindle fiber interaction with the kinetochore without major structural changes to the spindle fibers (Wang and Burke, 1995). Therefore, we propose that under microtubule destabilizing conditions induced by the low dose 1 µg/ml nocodazole treatment, deletion of HOG1 slows the kinetics of cell cycle progression, allowing for proper kinetochore-microtubule attachments to stabilize under otherwise less-than-ideal conditions. Accordingly, lower numbers of apoptotic and aneuploid cells were observed. The increased G2/ M population in hog1Δ gcn5Δ further supports this interpretation.
In S. cerevisiae, anaphase onset and mitotic exit follow the faithful segregation of sister chromatids. The mitotic exit network (MEN) controls the completion of mitosis and cytokinesis by coordinating molecular events resulting in the inactivation of cyclin-dependent kinase (CDK), reversal of CDK-mediated phosphorylation, accumulation of CDK inhibitors, and promotion of cyclin destruction (Visintin et al. 1998;Jin et al. 2008;Campbell et al. 2019). Under conditions where the kinetochore-microtubule attachment is destabilized, proper MEN inhibition is required for appropriate spindle assembly checkpoint function to ensure correct orientation and segregation of sister chromatids (Caydasi et al. 2010). Reiser and colleagues (2006) reported that sustained activation of the HOG pathway rescues the mitotic exit defects in MEN mutants. The results reported here align with these observations in that no distinct improvement was observed in hog1Δ gcn5Δ cell cycle progression when compared to gcn5Δ under normal growth conditions. However, improvement in cell cycle progression was observed when nocodazole was introduced to growth media, akin to inhibiting mitotic exit. Furthermore, hog1Δ mutants showed a significant delay in the spindle disassembly checkpoint under normal growth condition (Pigula et al. 2014). Together, these results imply that in the context of mitosis, Hog1 may have a role in promoting mitotic progression and exit, acting in opposition to checkpoints that halt cell cycle progression. Based on our results and previous studies, we suspected that this specific function of Hog1 might be independent of the cell cycle modulation mediated by Hog1 upon osmotic shock. Indeed, Hog1 was found to delay mitotic exit by inhibiting anaphase onset during osmotic stress (Tognetti et al. 2020).
Previous studies have shown that Gcn5 plays an important role in the centromeric localization of Rts1, which in turn promotes the tension sensing function of the spindle assembly checkpoint (Petty et al. 2018). Since the spindle assembly checkpoint plays a crucial role in the survival of cells in microtubule destabilizing conditions, in preliminary studies we examined if GCN5 had genetic relationships with the spindle assembly checkpoint mutants bub1Δ (Farr and Hoyt 1998;Musacchio 2015) and mad2Δ (Hardwick et al. 1999;Musacchio 2015) by testing the viability of bub1Δ gcn5Δ and mad2Δ gcn5Δ double mutants. No synthetically lethal interaction between the deletion of GCN5 and the loss of Mad2 or Bub1 was observed. This result suggests that the Gcn5-Rts1 interaction and the spindle assembly checkpoint components Bub1 and Mad2 might function in distinct aspects of the spindle assembly checkpoint. Further investigation of the spindle assembly checkpoint is in order, including analysis of genetic interactions with other players (Musacchio 2015) or a future focus on their localization or phosphorylation in its regulation in hog1Δ and gcn5Δ mutants.
We hypothesize that Hog1 functionally opposes the Gcn5 and Rts1-mediated checkpoint. Upon microtubule destabilization, Hog1 may be activated to promote mitotic exit and oppose the cell cycle arrest induced by the spindle assembly checkpoint mediated by Gcn5 and PP2A Rts1 (Fig. 6). In WT cells, a balance between the effects imposed on the cell cycle by Gcn5/Rts1 and Hog1 ensures the cells undergo faithful chromosome segregation without prolonged mitotic stalling or arrest. However, in gcn5Δ mutants where the checkpoint is defective, Hog1-related induction of mitotic exit could exacerbate the defective checkpoint (Fig. 6). Thus, deletion of HOG1 alleviates the chromosome segregation defect in gcn5Δ cells and increases the overall G2/M population when treated with nocodazole.
Previous studies demonstrated that a balance of dynamic phosphorylation at the centromere has a profound consequence on the interaction between the kinetochore and spindle fibers (Liu et al. 2010, Sherwin andWang, 2019). One possibility is that Hog1 directly opposes the spindle assembly checkpoint and promotes mitotic exit by phosphorylating Cse4. A single threonine residue, Cse4-T133, aligns with the phosphorylation consensus site of Hog1. Hog1 and PP2A Rts1 thus potentially mediates a balance of phosphorylation at the centromere, modulating the spindle assembly checkpoint. Additionally, Hog1 might indirectly promote mitotic exit via interaction with cell cycle checkpoint components since Hog1 has been shown to mediate cell cycle arrest via modulation of cyclin expression and CDK inhibition under osmotic stress (Escoté et al. 2004;Clotet et al. 2006;Yaakov et al. 2009;Tognetti et al. 2020). Other studies have also shown that Hog1 recruits the histone deacetylase Rpd3 to Fig. 6. A functional tug of war between Hog1, Gcn5, and Rts1 in mitotic progression. Gcn5 promotes Rts1 localization to centromeres, which is crucial for the activation of the spindle assembly checkpoint (SAC). We propose that under microtubule destabilizing conditions, Gcn5 acts to promote Rts1 localization to the centromere, which results in spindle assembly checkpoint (SAC) activation and cell cycle arrest (Petty et al 2018). At the same time, activated Hog1 promotes mitotic exit via either direct interaction with the centromere or indirect mechanisms involving other interacting partners. We hypothesize that under such conditions, Hog1 functionally opposes Rts1 and Gcn5. A balance between Gcn5/Rts1-mediated SAC and Hog1-promoted mitotic exit ensures faithful segregation of chromosomes and undisturbed progression of the cell cycle. Such a balance between spindle assembly checkpoint activation and mitotic exit is crucial under microtubule destabilizing conditions. The balance between forces driving the SAC and mitotic exit result in gcn5Δ sensitivity to nocodazole. In hog1Δ gcn5Δ, deletion of HOG1 causes a lack of driving force for mitotic exit during microtubule destabilizing conditions, restoring the balance between SAC activation and mitotic exit. osmotic stress gene promoters to modulate transcription of stressresponsive genes (De Nadal et al. 2004). Therefore, loss of Hog1 in gcn5Δ cells might restore the acetylation balance on key histone residues, thus improving gcn5Δ cell cycle progression.

The many faces of hog1Δ in cell cycle progression in the absence of Gcn5
Beyond defective mitotic progression in gcn5Δ mutants, as evidenced by gcn5Δ sensitivity to the overexpression of Clb2 (Krebs et al. 2000), gcn5Δ disturbs other stages of cell cycle progression, including a significant reduction in the centromeric localization of Rts1 during the spindle assembly checkpoint, and slower progression of G1/S and G2/M (Petty et al. 2016(Petty et al. , 2018. Based on the results from the present study, we propose that Hog1 plays different roles in gcn5Δ cell cycle progression based on its specific interaction partners and the context of the interaction. We hypothesize that Hog1 plays a role in promoting mitotic exit in its interaction with Gcn5 and Rts1 in the context of spindle assembly checkpoint. However, hog1Δ suppression of gcn5Δ synthetic lethality with morphogenesis checkpoint mutants suggests that in the context of that checkpoint, Hog1 plays a role in promoting cell cycle arrest. During the G2/M transition, timely Swe1 degradation depends on Hsl1 and Hsl7 and promotes the progression of the cell cycle. Loss of either Hsl1 or Hsl7 result in a constitutively activated morphogenesis checkpoint (McMillan et al. 1999). Previous studies identified deletion of SWE1, a gene encoding a CDK inhibitor that regulates G2/M progression, as a suppressor of gcn5Δ lethality under a constitutively activated morphogenesis checkpoint (Ruault and Pillus, 2006). Deletion of either HOG1 or SWE1 rescues gcn5Δ hsl1Δ and gcn5Δ hsl7Δ synthetic lethality, indicating that in the context of the morphogenesis checkpoint in gcn5Δ mutants, Hog1 and Swe1 play similar roles in the negative regulation of cell cycle progression. Indeed, upon sensing osmotic stress, Hog1 phosphorylates Hsl1, promoting Swe1 stabilization and G2/M arrest (Clotet et al. 2006). Together, these results suggest that there are multiple aspects of the Hog1-Gcn5 functional interaction contributing to cell cycle progression. The stage of the cell cycle, as well as the interacting partners, determines the role played by Hog1 and Gcn5, respectively, potentially as a function of their distinct substrates throughout the cell cycle.
In conclusion, through genetic analysis, we identified a previously unknown Hog1-Rts1-Gcn5 interaction in the context of histone modification and cell cycle progression. Our data support a hypothesis in which Hog1 and Rts1/Gcn5 functionally oppose each other at the spindle assembly checkpoint, pointing to a new role of the HOG pathway in the context of cell cycle checkpoint functions beyond classically defined responses to hyperosmolarity stress.

Data availability
Yeast strains and plasmids will be provided upon request. We affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables. Supplemental tables detailing strains, plasmids, and oligonucleotides are available at G3 online.
Supplemental material available at G3 online.