Chromatin stability safeguards mitochondrial homeostasis and prevents mTORC1 hyperactivation

Abstract

Cells dynamically regulate chromatin in response to nutrient flux which promotes the transcriptional changes necessary for adaptation. The mechanistic target of rapamycin complex 1 (mTORC1) kinase integrates nutrient signaling with chromatin regulation, yet whether chromatin stability feeds back to mTORC1 activation and stress adaption remains unknown. We previously identified histone H3 at lysine 37 (H3K37) as essential for the response to mTORC1 stress such that mutation of H3K37 to alanine (H3K37A) causes cell death upon mTORC1 inhibition. Herein, we show that H3K37-dependent chromatin stability prevents proteasome-mediated histone degradation, restricts mTORC1 signaling, and safeguards mitochondrial homeostasis during mTORC1 stress. Genetic interaction analyses reveal that H3K37A combined with mutants that destabilize chromatin, including loss of the Set2 H3K36 methyltransferase, Rpd3S histone deacetylase, or multiple histone deposition pathways, causes synthetic lethality when mTORC1 is inhibited. Transcriptome analysis indicates that H3K37A misregulates the mitochondrial transcriptome during mTORC1 stress, which increases mitochondrial reactive oxygen species (ROS) and triggers lethal mitochondrial retrograde signaling. Inactivation of retrograde signaling, or ROS neutralization, rescues viability of H3K37A and chromatin stability mutants during mTORC1 stress. These findings establish chromatin stability as a key safeguard that restrains mTORC1 activity and prevents toxic mitochondrial stress during metabolic adaptation.

Introduction

Eukaryotic cells dynamically adjust their growth and metabolism in response to changes in nutrient availability, energy status, and environmental stress¹. A central regulator of this adaptive response is the mechanistic target of rapamycin complex 1 (mTORC1), a conserved kinase complex that integrates nutrient inputs with downstream control of transcriptional regulation, protein synthesis, metabolism, and autophagy inhibition required for cell growth and proliferation^1,2. In all eukaryotes, coordination of mTORC1 activity with the availability of nutrient resources is essential to prevent deregulated growth or activation of cell stress responses that compromise organismal fitness^3–5.

In budding yeast, the mTORC1 complex is a serine/threonine kinase comprised of either the Tor1 or Tor2 kinase subunits along with the additional Lst8, Kog1, and Tco89 subunits^6,7. In yeast, the vacuole (homologous to the mammalian lysosome) is a degradative compartment that stores amino acids and other metabolites⁸. mTORC1 is recruited to the vacuole surface via Kog1 and Tco89-mediated interactions with the Rag GTPase containing EGO complex that consists of the Rag GTPases Gtr1 and Gtr2, and the additional Ego1–3 subunits^2,9,10. Vacuole amino acids activate mTORC1 by promoting Gtr1 loading with GTP and Gtr2 loading with GDP. Interaction of mTORC1 with EGO and the additional vacuole V-ATPase complex, which is the resident vacuole H⁺-pump that acidifies the vacuole, results in full mTORC1 activation¹¹. Once activated, mTORC1 phosphorylates a complex network of direct substrates that promote anabolism and cell proliferation while inhibiting nutrient stress responses, including autophagy^2,12–14. Among the multiple downstream kinases phosphorylated by mTORC1 is the Ypk3 kinase that in turn phosphorylates ribosomal protein S6 (S6ph)¹⁵. Emerging evidence indicates that mTORC1 also directly controls gene transcription in the nucleus, and it is a known regulator of epigenetic signaling^16,17. However, the underlying mechanisms by which mTORC1 controls transcription and the epigenome remain poorly defined. Deciphering this regulation will ultimately be essential for determining how mTORC1 dysregulation causes disease.

The regulation of chromatin structure and histone post-translational modifications (PTMs), collectively referred to as epigenetic regulation, controls gene expression and genome stability, and this epigenetic regulation is highly intertwined with cellular metabolism^18,19. The abundance of many histone PTMs that affect gene expression, including histone methylation and acetylation, is controlled by the availability of their metabolic donor molecules (S-adenosylmethionine (SAM) for histone methylation and acetyl-CoA for histone acetylation), which orchestrate chromatin accessibility and transcriptional output with the cellular metabolic state²⁰. Importantly, recent studies suggest that chromatin not only responds to metabolic cues, but that it also can signal in reverse to potentially modulate nutrient-responsive signaling pathways such as mTORC1^21,22. These studies suggest that reciprocal communication may exist between chromatin and mTORC1, yet the mechanisms explaining such bidirectional signaling have yet to be defined.

We previously performed a chemical genetic screen utilizing the mTORC1-specific inhibitor rapamycin against a library of histone H3 and H4 (H3/H4) mutants to identify functional relationships between mTORC1 and the epigenome²³. This screen found that mTORC1 exhibits strong genetic interactions with multiple histone H3 N-terminal residues, many of which are sites of PTM. Notably, we found that mutation of H3K37, which is located at the nucleosomal DNA entry and exit point²⁴, to alanine (H3K37A) causes cytotoxicity upon mTORC1 inhibition^21,23. H3K37 is adjacent to H3K36, which is methylated by the Set2 methyltransferase required for maintaining transcriptional fidelity and chromatin stability during gene transcription^25,26. Importantly, both the H3K37A mutant, as well as cells lacking Set2 (set2Δ), exhibit elevated mTORC1 signaling^21,22, suggesting that the H3K36/K37 di-lysine motif may contribute to chromatin-mediated mTORC1 regulation. Set2 catalyzes mono-, di-, and trimethylation of H3K36 (H3K36me1/2/3) via its conserved SET domain, whereas its conserved C-terminal Set2-Rpb1 interacting (SRI) domain binds to the dually phosphorylated serine 2 and serine 5 residues of the RNA polymerase II (Pol II) C-terminal domain (CTD)^25–27. Set2 binding to Pol II promotes targeting of H3K36me1/2/3 to transcribed chromatin, which maintains transcriptional fidelity and chromatin integrity through the recruitment of epigenetic reader proteins that selectively bind the H3K36me2/3 modifications²⁶. Specific readers of H3K36me2/3 include the Eaf3 subunit of the Rpd3S histone deacetylase (HDAC) complex^28–30, the Ioc4 subunit of the Isw1b chromatin remodeling complex³¹, and the Pdp3 subunit of the NuA3 histone acetyltransferase complex³². To date, no known reader protein has been identified that selectively binds H3K36me1. Importantly, H3K36me2/3 also maintains chromatin stability during transcription by suppressing histone replacement mediated by histone chaperone proteins such as Asf1³³.

In this study, we used the H3K37A mutation as a tool to define how chromatin stability constrains mTORC1 signaling and supports cellular adaptation to mTORC1 stress. We show that H3K37A destabilizes histones, leading to proteasome-dependent histone degradation that is associated with mTORC1 hyperactivation. Genetic and biochemical analyses reveal that Set2-mediated H3K36me2/3 and the Rpd3S HDAC complex cooperates with H3K37 to prevent chromatin instability and sustain viability during mTORC1 stress. Transcriptome analyses indicates that mTORC1 inhibition in H3K37A selectively deregulates genes controlling mitochondrial function and metabolism, driving mitochondrial reactive oxygen species (ROS) and cytotoxic mitochondrial retrograde signaling. Inhibition of mitochondrial retrograde signaling or ROS neutralization fully rescues H3K37A viability under mTORC1 stress. Compromising additional chromatin assembly pathways in the H3K37A background triggers a similar lethal mitochondrial retrograde program, underscoring that maintaining chromatin stability prevents toxic activation of this pathway during mTORC1 stress. We also find that combining H3K37A with either Set2 loss or chromatin assembly defects further hyperactivates mTORC1 activity, supporting a causal link between histone degradation and mTORC1 deregulation. These findings establish chromatin stability as a key safeguard that restrains mTORC1 signaling, preserves mitochondrial homeostasis, and shapes metabolic stress adaptation.

Results

H3K37 prevents histone destabilization and sensitivity to mTORC1 stress

We previously found that mutation of H3K37 to alanine (H3K37A) resulted in a loss of HMGB protein binding to chromatin, and it also increased both mTORC1 signaling and sensitivity to mTORC1 stress^21,23. H3K37 is located at the nucleosome DNA entry/exit point, and a recent nucleosome crystal structure suggests that H3K37 may contact nucleosomal DNA to help stabilize epigenetic reader binding to the adjacent methylated H3K36 residue (Fig. 1A)³⁴. To delineate the functional relationship between H3K37 and the mTORC1 pathway, we initially tested the idea that the H3K37A hypersensitivity to mTORC1 stress might be due to a histone defect. H3 wild-type (H3WT) and H3K37A were transformed with either a control vector or a vector expressing all four core histone genes. Cells were then spotted to control media or media containing the mTORC1 inhibitor rapamycin. While H3K37A vector control cells exhibited a profound growth defect compared to H3WT, histone expression rescued H3K37A growth under mTORC1 stress (Supplementary Fig. S1A). Intriguingly, immunoblot (IB) analysis of both soluble and denatured cell extracts revealed that histone protein levels were reduced in the H3K37A vector control relative to the H3WT vector control, and this histone deficiency was rescued by the histone expression vector (Fig. 1B–E and Supplementary Fig. S1B-E). These data indicate that H3K37A causes both a global reduction in histone levels, and an increased sensitivity to mTORC1 stress, that can be rescued by histone overexpression.

Figure 1. The H3K37A mutant reduces histone expression and causes lethality during mTORC1 stress.

(A) The crystal structure of nucleosome-Rpd3S interaction. Red- H3K37; blue- H3K36me3; yellow– DNA; white- histones; green– Eaf3. Structure is PDB ID 8TOF. (B) H3WT and H3K37A were transformed with a control vector or a vector expressing all four core histone genes. Soluble protein extracts were prepared and analyzed by IB with the indicated antibodies. (C-E) Quantification of histones H3, H4, and H2A and normalized to G6PDH signal. n=3; *p<0.05, **p<0.01. (F) The H3K37A cells transformed with the HHT2-HHF2, hht2K37A-HHF2, or hht2K37R-HHF2 plasmid, along with the control vector in H3WT and H3K37A, were diluted 5-fold and spotted onto the indicated plate for 3–5 days. (G) IBs for the total histone H3 and H4 for the cells in F. (H-I) The graph indicates IB band intensity of H3, H4, and was normalized to G6PDH. n=4; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

We next confirmed that the H3K37A mutation was solely responsible for the mTORC1 stress sensitivity and histone reduction. H3WT and H3K37A were transformed with a control vector, while H3K37A was also transformed with a vector expressing either the wild-type HHT2-HHF2 histone H3-H4 gene locus, an H3K37A mutation (hht2K37A-HHF2) that disrupts both the size and charge of the native H3K37 residue, or an H3K37R change (hht2K37R-HHF2) that restores charge and size at this position. Cells were then spotted to control media or rapamycin media. None of the strains exhibited growth defects in control media, whereas the H3K37A vector control cells were severely growth-impaired under rapamycin-induced mTORC1 stress (Fig. 1F). The HHT2-HHF2 and hht2K37R-HHF2 expression vectors rescued this growth deficiency, while hht2K37A-HHF2 did not (Fig. 1F). Additionally, IB analysis of H3K37A cells expressing either HHT2-HHF2 or hht2K37R-HHF2 revealed that histone protein levels were restored to those of H3WT, while histone levels in hht2K37A-HHF2 expressing cells were similar to the H3K37A control (Fig. 1G–I). Our previous report that employed individual yeast strains expressing these histone H3 mutants integrated into the genome also found that H3WT, H3K37R, and H3K37Q restored growth during mTORC1 stress, whereas H3K37A did not²¹. Collectively, these findings demonstrate that the single H3K37A mutation mediates hypersensitivity to mTORC1 stress and globally reduces histone levels.

Since hht2K37A-HHF2 failed to restore histone levels, we next determined if H3K37A reduced histone gene expression to explain the lower histone protein levels. The H3WT and H3K37A strains express a single genomically integrated synthetic H3-H4 cassette³⁵, while the endogenous HHT1-HHF1 and HHT2-HHF2 alleles have been deleted. Using HHT1-specific primers that recognize the synthetic H3 cassette, we analyzed HHT1, HTA1, and HTB1 mRNA levels by quantitative real-time PCR (qRT-PCR) in H3WT and H3K37A expressing control vector, and also in H3K37A expressing either HHT2-HHF2 or hht2K37A-HHF2. No differences in HHT1, HTA1, and HTB1 mRNA levels were found between H3WT and either H3K37A control vector, HHT2-HHF2, or hht2K37A-HHF2 expressing cells (Supplementary Fig. S1F-H), therefore indicating the histone H3 reduction by H3K37A is independent of mRNA changes. Additionally, no difference in HHT2 or hht2K37A expression from the exogenous histone vector was detected in H3K37A (Supplementary Fig. S1I). Based on these data and additional results discussed below, we conclude that H3K37A reduces histone levels through an mRNA-independent mechanism.

Histones are synthesized during the S-phase where they are then incorporated into nascent chromatin³⁶. We assessed whether cell cycle differences exist between H3WT and H3K37A that could explain the histone reduction in H3K37A. Asynchronous H3WT and H3K37A cultures were stained with SYTOX Green and analyzed by flow cytometry, which revealed that H3WT and H3K37A exhibited nearly identical cell cycle profiles (Fig. 2A and Supplementary Fig. S2A-B). Therefore, H3K37A inhibits histone expression independently of changes to its cell cycle. We next determined whether H3K37A enhanced histone degradation to account for the lower histone levels. Protein degradation occurs predominantly through either autophagy or the proteasome. We blocked autophagy in H3WT and H3K37A using the autophagy inhibitor phenylmethylsulfonyl fluoride (PMSF)³⁷, and histone IB analysis revealed that autophagy inhibition failed to restore histone expression (Fig. 2B–C). However, proteasome inhibition in H3WT and H3K37A using the proteasome inhibitor MG132 restored H3K37A histone levels to those of H3WT (Fig. 2D–F)³⁸. These data suggest that H3K37A increases histone instability and subsequent degradation by the proteasome. Since a reduction in histone stability may promote chromatin decompaction³⁹, we analyzed global chromatin accessibility in H3WT and H3K37A by performing micrococcal nuclease (MNase) digestion. Even though H3K37A quantifiably reduces histone levels (Fig. 1B–E, 1G, and Supplementary Fig. S1B-E), no difference in chromatin accessibility was detectable between H3WT and H3K37A (Supplementary Fig. S2C). In total, these findings indicate that while H3K37A increases histone instability and degradation by the proteasome, the histone reduction observed is insufficient to substantively alter global chromatin compaction.

Figure 2. The H3K37A mutant causes histone instability and degradation via the proteasome.

(A) Graphical representation of the cell percentage for each phase of the cell cycle in H3WT and H3K37A. n=3; ns- no significance. (B) The H3WT and H3K37A were treated with the autophagy inhibitor PMSF (1 mM) for four hours before analyzing histone H3 by IB. (C) Quantification of histone H3 levels normalized to G6PDH from B. n=4; **p<0.01, ***p<0.001. (D) H3WT and H3K37A are treated with the proteosome inhibitor MG132 (20 μM) for two hours, and then histones H3 and H2A were analyzed by IB. (E-F) Quantification of histones H3 (E) or H2A (F) normalized to G6PDH. n=3–4; *p<0.05, ***p<0.001.

H3K37 and the Set2 methyltransferase act synergistically to sustain viability during mTORC1 stress.

H3K37 is adjacent to H3K36 (Fig. 1A), which is methylated co-transcriptionally by the Set2 methyltransferase. We previously found that Set2 regulates mTORC1 signaling and the response to mTORC1 stress²². To assess if loss of Set2 in H3K37A increases sensitivity to mTORC1 stress, we combined a set2Δ into H3WT and H3K37A. Set2 is one of three histone H3 methyltransferases with critical roles in transcriptional control, so we also expanded our analysis to examine inactivating mutants of these additional histone H3 methyltransferases. H3WT and H3K37A strains were generated that either lack H3K4 methylation due to a Set1-containing COMPASS complex inactivating mutant (cps50Δ) or lack Dot1-dependent H3K79 methylation (dot1Δ)^40,41. All strains were scored for their mTORC1 stress sensitivity by plating onto control media or media containing rapamycin. Neither the H3WT, nor the individual methyltransferase mutants in the H3WT background, affected mTORC1 stress sensitivity (Fig. 3A). As expected, the H3K37A mutant exhibited sensitivity to mTORC1 stress, and this effect was further exacerbated by set2Δ (Fig. 3A). Since the only known Set2 substrate is H3K36²⁶, the increased mTORC1 stress sensitivity of H3K37A set2Δ indicates that Set2-mediated H3K36 methylation remains at least partially intact in H3K37A. Importantly, while the H3K37A cps50Δ caused only a negligible increase in mTORC1 stress sensitivity relative to H3K37A, the H3K37A dot1Δ modestly rescued the H3K37A growth defect (Fig. 3A). Although both COMPASS and Set2 were proposed to methylate H3K37⁴², the rescue of the H3K37A mTORC1 growth defect by either H3K37R (Fig. 1F) or H3K37Q²³, combined with the negligible effect that H3K37A cps50Δ has (Fig. 3A), indicates that loss of H3K37 methylation is not responsible for the increased sensitivity to mTORC1 stress. Set2-mediated H3K36 methylation prevents nucleosome hyperacetylation and disruption of nucleosome stability during transcription²⁶, while Dot1-mediated H3K79 methylation is associated with an open and transcriptionally active chromatin environment⁴³. The opposing effects that Set2 and Dot1 loss have on the H3K37A mTORC1 stress response suggest that maintaining chromatin stability is essential for viability during mTORC1 stress.

Figure 3. Set2 maintains viability in H3K37A during mTORC1 stress through methylation-dependent and -independent mechanisms.

(A) The set2Δ, cps50Δ, and dot1Δ were diluted 5-fold and spotted onto the indicated plates for three days. (B) H3WT, H3K37A, and their set2Δ derivatives were mock-treated or 50 nM rapamycin-treated for six hours before YO-PRO-1 and PI staining and analysis by flow cytometry. The cell death percentage is represented in the graph. n=3; **p<0.01, ****p<0.0001 and ^##p<0.01. (C) IB analysis of histone H3 in H3WT, H3K37A, and their set2Δ derivatives. (D) Quantification of histone H3 normalized to G6PDH from C. n=3; *p<0.05, ***p<0.001 and ^#p<0.05. (E) Expression of wild-type Set2-HA and the indicated Set2 mutants in H3K37A set2Δ. (F) Quantification of Set2-HA and the indicated Set2-HA mutants normalized to G6PDH from E. n=4; ns- no significance. (G) Spot assay of the strains transformed with the control vector, or vectors expressing either wild-type Set2-HA or the indicated Set2 mutants. (H) As in G, except cells express either control vector, or vector expressing wild-type Set2, the catalytically inactive Set2 (set2-H199L), or the SRI domain-truncated Set2 mutant (set2 (1–618)). For A, G, and H, images were captured on the indicated day.

We previously found that the H3K37A growth defect during mTORC1 stress is due to cell death²¹. Since H3K37A set2Δ enhanced sensitivity to mTORC1 stress, we tested if it exhibited increased cell death upon mTORC1 inhibition. Asynchronous cultures of H3WT, H3K37A, and their set2Δ derivatives were mock-treated or treated with 50 nM rapamycin for six hours before staining with YO-PRO-1 (YO, stains early apoptotic cells) and propidium iodide (PI, stains late apoptotic and necrotic cells) and then analyzed by flow cytometry (Fig. 3B and Supplementary Fig. S3A). No significant cell death was detected in the H3WT, H3K37A, and the set2Δ mutants under control conditions (Fig. 3B). Importantly, while rapamycin treatment did not affect either H3WT or H3WT set2Δ, it increased cell death in H3K37A as expected²¹, and this cytotoxic effect was enhanced further by H3K37A set2Δ (Fig. 3B). To determine if the increased cell death in H3K37A set2Δ was linked to a further reduction in histone levels, we quantified histone H3 levels in H3WT, H3K37A, and the set2Δ mutants. H3K37A reduced H3 expression as seen above (Fig. 1), while H3WT set2Δ also reduced H3 levels, and this histone reduction was decreased further by H3K37A set2Δ (Fig. 3C–D). These data demonstrate that loss of the Set2 pathway in H3K37A enhances mTORC1-stress induced cytotoxicity and further reduces histone protein levels. However, histone instability alone cannot be the sole cause of cell death under these conditions since the H3WT set2Δ has reduced histone expression but a normal response to mTORC1 stress.

Set2 is solely responsible for mono-, di-, and tri-methylation at H3K36 (H3K36me1/2/3)²⁶. We next determined the role the specific H3K36 methylation states have in sustaining the viability of H3K37A set2Δ during mTORC1 stress. H3K37A set2Δ was transformed with control vector, or vectors expressing either HA-tagged SET2 wild-type (SET2-WT), the set2-Y149F mutant that selectively maintains H3K36me3 without H3K36me1/2, the set2-F234Y that generates H3K36me1/2 but lacks H3K36me3, or the catalytic domain mutant set2-H199L that ablates almost all H3K36 methylation while retaining only minimal residual H3K36me1⁴⁴. The Set2-WT and the different mutants were equivalently expressed in H3K37A (Fig. 3E–F), and they restored the expected H3K36 methylation states as previously reported (Supplementary Fig. S3B)⁴⁴. Under mTORC1 stress, the SET2-WT rescued H3K37A set2Δ viability and growth back to that of H3K37A (Fig. 3G). Although less effective than Set2-WT, expressing either Set2-F234Y, Set2-Y149F, or Set2-H199L in H3K37A set2Δ rescued growth relative to H3K37A set2Δ vector control (Fig. 3G). Set2 is recruited to chromatin in part by binding to the phosphorylated C-terminal domain of Pol II through the Set2 C-terminal SRI domain. Set2 binding to the CTD then alleviates autoinhibition of Set2 catalytic activity to generate H3K36me2/3⁴⁵. To test if Set2 requires Pol II binding for its role in the mTORC1 stress response, we transformed H3K37A set2Δ with either SET2-WT-HA, set2(1–618)-HA that lacks the SRI domain, the catalytically inactive set2-H199L-HA, or a control vector. Spot assays revealed that the SET2-WT rescued H3K37A set2Δ growth back to that of H3K37A control (Fig. 3H). Importantly, while less efficient than SET2-WT-HA, both set2(1–618)-HA and set2-H199L-HA partially rescued growth under mTORC1 stress (Fig. 3H). The SRI truncated Set2 does not bind Pol II or generate H3K36me3, likely due to autoinhibition of its SET domain^27,45, while the set2-H199L generates only minimal residual H3K36me1 with no H3K36me2/3⁴⁴. No epigenetic reader is known to selectively bind H3K36me1, and no functional role for H3K36me1 has ever been defined. Therefore, these results suggest that Set2 may provide both a methylation-dependent and methylation-independent role in maintaining H3K37A viability during mTORC1 stress that only partially depends upon Pol II binding.

We next addressed the functional consequence that loss of H3K36me2/3 demethylase activity (mediated by Rph1) in H3K37A has on the mTORC1 stress response⁴⁶. A rph1Δ was introduced into H3WT and H3K37A, and then H3WT, H3K37A, and both the set2Δ and rph1Δ derivatives were scored on control and rapamycin media. Consistent with the above results, H3K37A set2Δ caused synthetic lethality upon mTORC1 stress, while the H3K37A rph1Δ resulted in a modest growth rescue (Fig. 4A). These results agree with the above data demonstrating that inhibition of H3K37A mTORC1-stress induced cytotoxicity partially depends on H3K36 methylation (Fig. 3G). H3K36me2/3 binds distinct epigenetic reader proteins to regulate chromatin structure during transcription, including the Eaf3 subunit of the Rpd3S histone deacetylase (HDAC) that binds H3K36me2/3^28–30, the Ioc4 subunit of the Isw1b chromatin remodeler that binds H3K36me3³¹, and the Pdp3 subunit of the NuA3b HAT complex that also binds H3K36me3³². To identify the readers relevant to the H3K37A mTORC1 stress response, we individually deleted the genes encoding each reader subunit in H3WT and H3K37A and then scored each for growth on control or rapamycin plates. Loss of either Eaf3 or its associated HDAC subunit (Rpd3) phenocopied Set2 loss (Fig. 4A), indicating that Set2-dependent suppression of mTORC1 stress-induced cell death requires Rpd3S chromatin anchoring and histone deacetylation. Consistent with this concept, we found that H3K23ac levels (an Rpd3S substrate⁴⁷) increased in H3K37A, H3WT set2Δ, and H3K37A set2Δ compared to H3WT (Supplementary Fig. S3C-D). While H3WT set2Δ is expected to increase histone acetylation due to disrupted Rpd3S chromatin binding, these data suggest that H3K37A also may modestly reduce Rpd3S binding. This possibility is not unexpected since H3K37 interacts with DNA to potentially stabilize Eaf3-dependent nucleosome binding (Fig. 1A). Intriguingly, the H3K37A set2Δ mutant exhibits lower H3K23ac relative to H3WT set2Δ (Supplementary Fig. S3C-D), but the reason for this reduction is currently unclear. Analysis of the additional H3K36me2/3 readers revealed that H3K37A ioc4Δ, which disrupts anchoring of Isw1b to transcribed regions and impairs Isw1b-mediated nucleosome positioning and suppression of histone exchange³¹, resulted in a minor synthetic sick interaction compared to H3K37A (Fig. 4B). Importantly, we consistently observed that the H3K37A pdp3Δ, which impairs NuA3 histone acetyltransferase binding to transcribed genes³², rescued growth relative to H3K37A (Fig. 4B). Overall, these data suggest that H3K37A increases sensitivity to mTORC1 stress by impinging on pathways that sustain chromatin stability, including histone deacetylation.

Figure 4. The Set2, HIR, and CAF-1 pathways prevent increased histone instability and mTORC1 stress in H3K37A.

(A-B) The indicated strains were spotted onto control and rapamycin (4–5 nM) plates for 3–7 days. (C) H3WT, H3K37A, and their set2Δ derivatives cells were spotted onto control plates, or plates containing HDAC inhibitors (25 mM NAM and 10 mM SB), 4 nM rapamycin, or all inhibitors combined for 3–4 days. (D) The H3WT and H3K37A cac2Δ and hir2Δ mutants were serially diluted 5-fold and spotted onto the indicated plates for 2–6 days. (E-G) IB analysis and quantification of histones H3 and H4 in H3WT and H3K27A cac2Δ. Histone H3 and H4 levels were normalized to G6PDH. n=3; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 and ^#p<0.05. (H-J) IB analysis and quantification of histones H3 and H4 in H3WT and H3K37A hir2Δ normalized to G6PDH. n=4; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

To independently test this possibility, we spotted H3WT, H3K37A, and the set2Δ mutants to control plates, or plates containing either rapamycin, the sirtuin (class III HDAC) inhibitor nicotinamide (NAM)⁴⁸ and class I/II HDAC inhibitor sodium butyrate (SB)⁴⁹, or all three inhibitors combined. We found that H3K37A and H3K37A set2Δ showed no sensitivity to HDAC inhibition, while both were exceptionally sensitive to mTORC1 inhibition as seen above (Fig. 4C). Combining both mTORC1 and HDAC inhibition increased the cell death of H3K37A and H3K37A set2Δ, whereas H3WT and H3WT set2Δ were unaffected (Fig. 4C). Since global HDAC inhibition causes histone hyperacetylation and can increase nucleosome disassembly and/or histone exchange⁵⁰, these data reinforce the genetic results from above suggesting that mTORC1-stress induced cell death in H3K37A is linked to histone instability.

We probed this possibility further by assessing whether disruption of known chromatin assembly pathways would cause synthetic lethality in H3K37A during mTORC1 stress that was similar to Set2 loss. In both H3WT and H3K37A, we deleted the CAC2 gene that encodes a subunit of the chromatin assembly factor-1 (CAF-1) that deposits newly synthesized H3-H4 during replication⁵¹. Independently, we also deleted the HIR2 gene, which encodes a subunit of the HIR complex that promotes replication-independent histone deposition⁵². Neither H3WT cac2Δ nor H3WT hir2Δ substantially increased rapamycin sensitivity, but disruption of either chromatin assembly pathway in H3K37A exacerbated the mTORC1-stress induced cytotoxicity similar to H3K37A set2Δ (Fig. 4D). Although disruption of either histone deposition pathway in H3WT failed to sensitize cells to mTORC1 stress, histone levels were reduced in both, while histones were decreased even further in the H3K37A background (Fig. 4E–J). However, this decreased histone expression alone cannot explain the increased mTORC1 stress sensitivity in H3K37A, since H3WT set2Δ, H3WT cac2Δ, and H3WT hir2Δ reduced histone levels but exhibited no overt mTORC1 growth defects. These findings suggest that loss of H3K37-dependent chromatin regulation, combined with defects in histone stability through disruption of histone deposition and/or transcription-linked nucleosome stability pathways, collaboratively cause cell death upon mTORC1 stress.

Histone instability is linked to mTORC1 deregulation and sensitivity to mTORC1 stress.

By analyzing ribosomal S6ph, which is a downstream readout for mTORC1 activity¹⁵, we previously found that mTORC1 signaling is increased in H3K37A²¹. A separate report from our group also showed that mTORC1 activity increases in set2Δ²². These studies further support a functional relationship between epigenetic signaling through the histone H3K36/K37 di-lysine motif and the mTORC1 pathway. To probe the mechanistic basis of this relationship further, we analyzed mTORC1 activity in H3WT, H3K37A, and the set2Δ backgrounds. As previously reported, mTORC1 activity increased in both H3K37A and H3WT set2Δ, while it was enhanced even further in H3K37A set2Δ (Fig. 5A–B). To confirm that the elevated mTORC1 signaling in H3K37A was due solely to H3K37A, we transformed H3WT and H3K37A with a control vector, or H3K37A with a vector expressing either HHT2-HHF2, hht2K37A-HHF2, or hht2K37R-HHF2, and then analyzed mTORC1 activity. While control vector H3K37A exhibited higher mTORC1 signaling compared to H3WT control, both the HHT2-HHF2 and hht2K37R-HHF2 vectors reduced mTORC1 activity, while expression of hht2K37A-HHF2 had no effect (Fig. 5C–D). Because loss of replication-dependent (cac2Δ) and replication-independent (hir2Δ) histone deposition in H3K37A caused enhanced mTORC1 stress sensitivity and reduced histone expression, we also quantified their mTORC1 activity. Similar to H3K37A, we found that cac2Δ and hir2Δ in H3WT increased mTORC1 signaling comparable to that of H3K37A, which was not further increased in the H3K37A background (Fig. 5E–H). Overall, these data indicate that defects in histone stability (H3K37A), transcription-linked chromatin stability (set2Δ), or impaired histone deposition (cac2Δ or hir2Δ) reduce histone levels and deregulate mTORC1.

Figure 5. mTORC1 activity exhibits a negative correlation with the expression of histone proteins.

(A-B) mTORC1 activity was assessed in H3WT, H3K37A, and their set2Δ derivatives by quantifying S6ph normalized to total S6. n=3; *p<0.05, **p<0.01, ***p<0.001 and ^#p<0.05. (C-D) As in A, but mTORC1 activity was quantified in H3WT and H3K37A transformed with either control vector, or vector expressing either H3WT, H3K37A, or H3K37R. n=3; *p<0.05. (E-F) As in A, but mTORC1 signaling was assessed in the H3WT and H3K37A cac2Δ mutants. n=3; *p<0.05, **p<0.01, and ^#p<0.05. (G-H) As in E but mTORC1 activity was measured in the H3WT and H3K37A cac2Δ background. n=6; **p<0.01, ****p<0.0001.

Tco89 is a nonessential mTORC1 subunit that recruits mTORC1 to the vacuole by interacting with the Gtr1/Gtr2 Rag GTPase subunits of the EGO complex⁹. Tco89 loss reduces mTORC1 activity by disrupting mTORC1-EGO interactions, and it results in mTORC1 repositioning from the vacuole to signaling endosomes^53,54. To address whether increased mTORC1 activity alone was sufficient to cause cell death during mTORC1 stress, we transformed H3WT with control vector or a vector overexpressing Tco89 (TCO89OE) and quantified mTORC1 activity. Compared to the control vector, TCO89OE increased mTORC1 signaling (Fig. 6A–B), which is most likely explained by Tco89 enhancing mTORC1 interactions with EGO⁹. However, this increased mTORC1 signaling alone was insufficient to sensitize cells to mTORC1 stress (Fig. 6C). We next determined if expressing H3K37A in H3WT could deregulate mTORC1. H3WT cells were transformed with a vector expressing either HHT2-HHF2, hht2K37A-HHF2, or hht2K37R-HHF2, and the S6ph/S6 ratio was then quantified. Importantly, compared to HHT2-HHF2 expressing cells, the hht2K37A-HHF2 boosted mTORC1 activity, whereas hht2K37R-HHF2 expressing cells showed no increase (Fig. 6D–E). These cells were then scored for growth on control media and media containing a lethal (for H3K37A) concentration of rapamycin. Expressing either HHT2-HHF2 or hht2K37R-HHF2 did not affect rapamycin sensitivity, while hht2K37A-HHF2 expression caused a modest mTORC1 growth defect (Fig. 6F). In this experiment, the hht2K37A-HHF2 competes with the expression of endogenous H3WT, which likely limits the overall impact H3K37A overexpression has on mTORC1 stress sensitivity. However, the results do indicate that H3K37A expression alone is sufficient to increase mTORC1 signaling and sensitize cells to mTORC1 stress. H3WT set2Δ, H3WT cac2Δ, and H3WT hir2Δ all have reduced histone expression and elevated mTORC1 activity, yet they do not sensitize cells to mTORC1 stress (Fig. 5). Therefore, while H3K37A causes these same effects, it must induce additional chromatin changes to cause cell death during mTORC1 stress.

Figure 6. Increased mTORC1 activity alone is not responsible for cell death during mTORC1 stress.

(A-B) H3WT was transformed with a control vector or vector expressing Tco89-FLAG (TCO89OE), and mTORC1 activity then was quantified by S6ph/S6 IB. Two independent transformants are shown. n=4; **p<0.01. The red arrow indicates Tco89-FLAG, and the asterisk indicates a cross-reactive protein. (C) Cells from A were spotted onto the indicated plates for 2–5 days. (D-E) mTORC1 activity was measured in H3WT transformed with either HHT2-HHF2, hht2K37A-HHF2, or hht2K37R-HHF2 expression vectors by quantifying S6ph/total S6. n=3; *p<0.05, ns– no significance. (F) The cells from D, along with the control vector H3WT and H3K37A, were spotted onto the indicated plates for three days.

Increased mTORC1 stress-induced cell death caused by histone instability is due to mitochondrial dysregulation.

To further delineate how mTORC1 stress in H3K37A causes cell death, we isolated total RNA from H3WT and H3K37A that were either mock-treated or treated with 20 nM rapamycin for one hour and performed RNA sequencing (RNA-seq) of total RNA. By comparing each experimental condition independently to the mock-treated H3WT control, we identified differentially expressed genes (DEGs) and non-coding RNAs, including tRNAs, snRNAs, and/or snoRNAs (collectively all are referred to as DEGs), using a 1.5-fold change in expression and an FDR<0.05 as our significance criteria. (Supplementary Table S1). Relative to mock-treated H3WT, the mock-treated H3K37A affected 382 DEGs, the rapamycin-treated H3WT altered 140 DEGs, and the rapamycin-treated H3K37A resulted in 1070 DEGs (Supplementary Table S1). These DEGs were then analyzed via the STRING database to identify overrepresented cellular component gene ontologies (GOs) for each experimental condition⁵⁵. The mock-treated H3K37A altered GOs annotated to the fungal-type cell wall, cell periphery, and anchored component of membrane categories (Supplementary Fig. S4A). The rapamycin-treated H3WT downregulated ribosomal gene related categories (Supplementary Fig. S4B), consistent with mTORC1 inhibition repressing ribosomal gene expression⁵⁶. Interestingly, while the rapamycin-treated H3K37A altered the expression of nucleus, nucleolus, and both intracellular and cellular anatomical entity categories, ribosomal-specific GOs were not overrepresented, which is unlike the rapamycin-treated H3WT (Supplementary Fig. S4C-D). The individual gene lists for each condition are provided in Supplementary Table S2. Although the reason for this difference is not clear, one possibility could be that since H3K37A has higher basal levels of mTORC1 activity, transcriptionally repressing ribosomal genes may be delayed compared to H3WT. To determine if the increased mTORC1 signaling in H3K37A could be caused by altered expression of a known mTORC1 regulator, we manually analyzed the mock-treated H3K37A DEGs, but we did not find an established mTORC1 regulator whose expression was altered (data not shown). These data indicate that H3K37A likely does not enhance mTORC1 signaling through changes in gene expression. We also compared our DEGs to two previously published set2Δ transcriptomes to assess how much overlap exists between the effect that H3K37A has versus Set2 loss^22,57. Venn analysis of these DEGs identified only 47 genes that overlap between these three studies (Supplementary Fig. S4E with gene lists used for comparison provided in Supplementary Table S1). These results further support our genetic analyses that H3K37A does not disrupt the Set2-dependent H3K36me pathway (Fig. 3–4).

Manual inspection of the RNA-seq dataset revealed a modest reduction (~1.5–1.7 fold) in multiple histone genes in the mock-treated H3K37A, and also in the rapamycin-treated H3WT and rapamycin-treated H3K37A (data not shown). We independently validated this finding by RT-qPCR and found that multiple histone genes, including HHO1, which encodes linker histone H1, were downregulated in H3K37A compared to H3WT (Supplementary Fig. S5A-F). In contrast, in Supplementary Fig. 1F–I, our data indicated there is no significant difference in the expression of histone genes between the control vector expressing H3WT and H3K37A cells. Currently, the explanation for these differences remains unclear, but they most likely relate to changes in the media nutrient environment required to select for plasmid maintenance in Supplementary Fig. S1F-I. We have found that nutrient-selective media required to sustain plasmid maintenance can impact mTORC1-related cell phenotypes (data not shown). Regardless, these histone mRNA decreases do not explain the reduced histone content of H3K37A, since overexpression of either HHT2-HHF2 or hht2K37A-HHF2 in H3K37A resulted in equivalent mRNA expression, while no difference in HHT2 expression from the exogenous histone expression plasmid is detected (Supplementary Fig. 1I). Most importantly, histone levels are restored in H3K37A specifically by proteasome inhibition (Fig. 2D–F), thus indicating that the histone reduction in H3K37A is due to increased histone degradation and not reduced histone mRNA repression. To specifically test if the reduction in linker histone mRNA could explain the increased mTORC1 sensitivity, we deleted HHO1 in H3WT and H3K37A, but HHO1 loss did not modify the mTORC1 stress response (Supplementary Fig. S5G). As such, the reduction in HHO1 mRNA does not explain the mTORC1-stress induced cell death in H3K37A.

We next performed GO analysis of the DEGs between the mock-treated H3K37A and rapamycin-treated H3K37A to identify GO categories that could explain the mTORC1-stress induced cytotoxicity. This analysis revealed that the rapamycin-treated H3K7A GOs were dominated by mitochondrial-related categories, including GOs annotated to the respiratory chain complex, oxidoreductase complex, mitochondrial respirasome, respirasome, mitochondrion, and cytochrome complex (Fig. 7A). Manual inspection of the genes in each of these GOs revealed that the vast majority were upregulated in the rapamycin-treated H3K37A relative to mock-treated H3K37A (data not shown). We validated this increased mitochondrial gene expression by targeted RT-qPCR analysis of some of the most overexpressed mitochondrial genes (CRC1, COX26, FMP16, and FMP45). All four genes were upregulated in rapamycin-treated H3K37A relative to the other experimental conditions (Supplementary Fig. S6A-D). These data indicate that during mTORC1 stress, H3K37A specifically causes deregulation of mitochondrial gene expression, which is linked to increased cell death.

Figure 7. Histone instability deregulates mitochondrial retrograde signaling to cause cell death during mTORC1 stress.

(A) Bubble blot analysis of the cellular component gene ontology (GO) categories for H3K37A (mock vs rapamycin). GO analysis was performed through the STRING database. FDR – False discovery rate. (B) H3WT and H3K37A combined with either set2Δ, rtg1Δ, or both mutations were diluted 5-fold and spotted onto the indicated plates for 2–6 days. (C) The indicated strains from B were cultured to the exponential phase and stained with DHE before quantifying DHE fluorescence. A.U.– Arbitrary Units of DHE fluorescence. n=7; ns– no significance. (D) As in C, except the cells were treated with 50 nM rapamycin for 24 hours before staining with DHE. n=8; **p<0.01, ***p<0.001, ****p<0.0001 and ^#p<0.05, ns– no significance. (E) H3WT and H3K37A combined with the set2Δ, cps50Δ, or hir2Δ mutations were diluted 5-fold and spotted onto control plates, or plates containing 50 mM NAC, 20 nM rapamycin, or both inhibitors for 2–5 days.

Histone instability deregulates mitochondrial retrograde signaling to cause cell death during mTORC1 stress.

The inhibition of mTORC1 leads to transcriptional activation of nuclear-encoded mitochondrial genes (retrograde signaling) through the transcription factors Rtg1 and Rtg3⁵⁸. This retrograde signaling leads to increased mitochondrial content and enhanced mitochondrial function during nutrient stress⁵⁸. We previously found that H3K37A increased ROS production compared to H3WT, and that the gene CIT2, a known retrograde-activated gene, was induced to significantly higher levels in rapamycin-treated H3K37A relative to rapamycin-treated H3WT²¹. Our RNA-seq analysis also revealed that both CIT1 (another retrograde-activated gene) and CIT2 genes were upregulated to higher levels in rapamycin-treated H3K37A cells compared to rapamycin-treated H3WT (Supplementary Fig. S6E-F), suggesting that H3K37A may deregulate the retrograde response. To test if deregulated retrograde signaling in H3K37A causes mTORC1 stress-induced cell death, we deleted RTG1 in H3WT, H3K37A, and their set2Δ derivatives to block retrograde activation. The growth of the H3WT, H3K37A, and these additional mutants was then scored on control or rapamycin-containing media. Importantly, while H3K37A and H3K37A set2Δ exhibited substantial cell growth defects (due to increased cell death, Fig. 3B), the viability of both was restored by rtg1Δ (Fig. 7B). This rescue occurred even though both H3K37A rtg1Δ and H3K37A set2Δ rtg1Δ are profoundly slow growing during mTORC1 stress compared to the matched H3WT rtg1Δ and H3WT set2Δ rtg1Δ (Fig.7B). These data indicate that deregulated retrograde signaling in H3K37A and H3K37A set2Δ causes the cytotoxic effect of mTORC1 stress.

A retrograde-mediated increase in mitochondrial activity can elevate ROS production, and we previously found evidence that mTORC1 inhibition increases ROS in H3K37A²¹. To assess if the rescue of H3K37A and H3K37A set2Δ viability by retrograde inactivation was due to ROS inhibition, we mock-treated or 50 nM rapamycin-treated cells for 24 hours and then quantified ROS by DHE staining. While H3K37A and H3K37A set2Δ increased ROS in the rapamycin-treated samples, this ROS was reduced by rtg1Δ (Fig. 7C–D). To test if deregulated ROS in mTORC1-inhibited H3K37A is the direct cause of cell death, we scored H3WT, H3K37A, and their set2Δ derivatives on control media, media containing either 20 nM rapamycin (which is lethal for H3K37A)^21,23, 50 mM of the antioxidant N-acetyl cysteine (NAC), or both conditions. We also included the H3WT and H3K37A histone deposition mutants (hir2Δ or cac2Δ) to assess if mTORC1-stress induced ROS deregulation is a general phenomenon associated with histone instability. All H3WT backgrounds grew equivalently under mTORC1 stress, NAC exposure, or with both inhibitors (Fig. 7E). Under mTORC1 stress alone, all H3K37A backgrounds failed to grow, consistent with this level of mTORC1 stress being lethal in H3K37A; however, combined mTORC1 and ROS inhibition completely restored their viability (Fig. 7E). To confirm that NAC did not prevent mTORC1 inhibition by rapamycin, we assessed mTORC1 activity by analyzing S6ph/S6 and determined that rapamycin still inhibits mTORC1 in the presence of NAC (Supplementary Fig. S6G).

Discussion

In this report, we demonstrate that histone H3K37 maintains chromatin stability, mTORC1 signaling regulation, and transcriptional induction of mitochondrial genes required for maintaining mitochondrial homeostasis during mTORC1 stress. Conversion of H3K37 to a small hydrophobic residue (H3K37A) increases histone instability and reduces histone levels independently of changes to histone mRNA expression or cell cycle status, indicating that the histone reduction occurs post-translationally. Indeed, proteasome inhibition fully restores histone levels in H3K37A, thus revealing the histone reduction to be due solely to increased proteasome-mediated degradation. These histone destabilizing effects were reversed by restoring an amino acid that is either positively charged (H3K37R) or polar (H3K37Q), indicating H3K37 sustains histone stability by maintaining a key histone H3 N-terminal tail contact, as we previously suggested²¹. While H3K37 is critical for histone stability and prevents histone degradation, the reduced histone levels due to H3K37A are insufficient to increase global chromatin accessibility. This seeming incongruity is similar to the results found in a recent preprint report demonstrating that other histone destabilizing mutants also fail to increase global chromatin accessibility even though they do cause marked changes to gene-specific and lineage-specific transcriptional programs⁵⁹. Collectively, these results indicate that cells can sustain global chromatin compaction under conditions where histone expression is reduced.

We do not yet know if H3K37A reduces Set2-mediated H3K36 methylation levels, as immunodetection methods are untrustworthy due to the underlying antibody epitope change caused by H3K37A. However, H3K37A clearly does not abolish either Set2-mediated H3K36 methylation or H3K36me reader binding since the set2Δ, eaf3Δ, and rpd3Δ in H3K37A cause synthetic lethality during mTORC1 stress. This possibility is further supported by the crystal structure of the Rpd3S Eaf3 chromodomain reader subunit bound to the nucleosome³⁴. In this structure, H3K37 is oriented away from the H3K36me3-Eaf3 interaction where it instead contacts nucleosomal DNA, perhaps acting to stabilize H3K36me3-Eaf3 binding. Within this context, H3K37A would disrupt this stabilizing histone-DNA interaction but presumably would not interfere substantially with either Set2-mediated methylation or Eaf3-Rpd3S binding. Intriguingly, our genetic analyses of Set2 indicate that both the catalytically deficient set2-H199L and the SRI-truncated Set2 mutant that prevents Pol II binding partially rescue H3K37A set2Δ mTORC1 stress sensitivity. Both set2-H199L and the set2-SRI mutant lack endogenous H3K36me2/3, while H3K36me1 has no known epigenetic readers. As such, these results suggest that Set2 may have both methylation-dependent and methylation-independent functions in promoting a normal mTORC1 stress response. Intriguingly, a previous report found that mice deficient in the Set2 homolog SETD2 also have deregulated mTORC1 signaling^60,61, suggesting that dysregulation of the Set2/SETD2 pathway may have evolutionarily conserved activating effects on mTORC1. Although yeast Set2 is not known to methylate non-histone substrates, SETD2 is documented to methylate some non-histone substrates^62–64. An intriguing possibility could be that Set2, analogous to SETD2, methylates a non-histone substrate independently of Pol II binding, and this non-histone substrate contributes to the mTORC1 stress response. In support of this possibility, Set2 interacts with a diverse array of non-histone proteins as annotated in the BioGRID database, including several ribosomal proteins that would be directly relevant to mTORC1-dependent translational regulation⁶⁵. Recent work has demonstrated that a SET domain methyltransferase can methylate ribosomal proteins⁶⁶. An intriguing future area for investigation will be to test if Set2, and perhaps even SETD2, methylates ribosomal proteins as part of the mTORC1 pathway’s response to nutrients and/or nutrient stress.

The mTORC1 pathway is integrated into mitochondrial regulation through the retrograde signaling pathway⁶⁷. In nutrient-rich mTORC1-activating conditions, retrograde signaling is repressed due to retention of the Rtg1/Rtg3 heterodimer in the cytoplasm. Upon mTORC1 inhibition due to nutrient insufficiency or rapamycin treatment, Rtg1/Rtg3 translocates to the nucleus and activates transcription of mitochondrial-related genes to enhance mitochondrial function⁵⁸. We find that during mTORC1 stress the H3K37A mutant dramatically dysregulates this mitochondrial retrograde transcriptional response since mitochondrial GOs are overwhelmingly overrepresented in the rapamycin-treated H3K37A mutant compared to the H3K37A mock-treated cells. Importantly, this H3K37A mitochondrial transcriptional dysregulation ultimately is the cause of the mTORC1-stress-induced cytotoxicity since blocking retrograde activation by genetic means, or suppressing mitochondrial-derived ROS, restores H3K37A viability during mTORC1 stress. This H3K37A mitochondrial dysregulation is exacerbated further by additional chromatin destabilizing mutants, including loss of Set2 or the histone deposition complexes CAF-1 or HIR. The exact details explaining this transcriptional deregulation have yet to be established. However, one possible explanation may involve nucleosome stability at the promoters of these mitochondrial genes. Constitutively expressed (housekeeping) genes have a canonical nucleosome-free, or nucleosome-depleted, promoter structure. However, many stress-induced genes, including metabolic stress genes, have nucleosomes positioned in their promoters that must be remodeled for full transcriptional induction⁶⁸. A previous report found that mutants that disrupt chromatin structure also deregulate mitochondrial genes in nutrient-rich conditions where these genes would normally be repressed⁶⁹. Since our data indicate that H3K37A, as well as H3K37A Set2, CAF-1, and HIR mutants, cause chromatin instability, these promoter nucleosomes at mitochondrial genes may either be reduced or fragile and more easily remodeled upon mTORC1 stress. Such chromatin-dependent promoter changes could cause deregulation of the mitochondrial transcriptional response after mTORC1 inhibition. Because retrograde activation in wild-type cells does not normally induce cytotoxicity, this retrograde transcriptional induction in the chromatin destabilizing mutants may be uneven and result in imbalanced mitochondrial gene expression. Such altered stoichiometries of mitochondrial components ultimately could cause mitochondrial metabolic stress and ROS production. Indeed, unbalanced expression of oxidative phosphorylation pathway subunits previously was shown to increase mitochondrial ROS and reduce chronological longevity⁷⁰. Future analysis of this mitochondrial transcriptional control in these chromatin destabilizing mutants will be required to understand how they result in mitochondrial transcriptional deregulation.

The observed increase in mTORC1 activity in H3K37A appears to be a consequence of general chromatin destabilization, since we find this occurs not just in the H3K37A and set2Δ as we previously reported^21,22, but also in the CAF-1 and HIR mutants. Inspection of the mock-treated H3K37A transcriptome data did not reveal gene expression changes that readily explain the elevated mTORC1 activity. Instead, one possible explanation for this enhanced mTORC1 activity could be that increased histone instability, and the resulting histone degradation by the proteasome, boosts intracellular free amino acid levels. These free amino acids could then either be funneled into the vacuole to signal for increased mTORC1 activation by the EGO and V-ATPase complexes, or they could activate mTORC1 through less well-characterized, vacuole-independent pathways⁷¹. Because of the large complement of histones needed for genome packaging, it is not unreasonable to speculate that even a small percentage increase in chromatin instability-mediated histone degradation boosts intracellular amino acid levels sufficiently to enhance mTORC1 activity. Although mTORC1 is a known epigenetic regulator, such a mechanism would imply that changes to the epigenome that cause histone destabilization also could signal back to cause a corresponding increase in mTORC1 activity. Future efforts will need to pursue this possibility further to confirm these predictions.

Chromatin destabilizing histone mutants alter gene expression or chromosome function to negatively affect cellular development and stress resilience⁵⁹. Mitochondrial activity also appears highly sensitive to changes in chromatin stability as mutants that perturb chromatin deregulate mitochondrial function in both yeast and metazoans^69,72,73. Many of the reported phenotypic changes that occur in histone destabilizing mutants mirror those changes that occur during aging since both worm and fly models expressing histone destabilizing mutants exhibit increased functional decline indicative of accelerated aging^59,72,73. Yeast and human cells expressing histone destabilizing mutants also exhibit defective responses to environmental stressors⁵⁹, further supporting the idea that chromatin stability is necessary for environmental stress responsiveness⁷⁴. During aging, epigenome integrity deteriorates while histone expression is reduced, and these chromatin changes are known to reduce longevity⁷⁵. Intriguingly, mTORC1 activity in many organisms increases as a function of aging, whereas pharmacological mTORC1 inhibition is a geroprotective intervention that increases longevity⁷⁶. Although it remains to be experimentally tested, we speculate that the histone loss associated with aging-dependent epigenome deterioration may fuel this increased mTORC1 activity if the histones are degraded by the proteasome and their amino acids are then recycled into the mTORC1 activation pathway. One consequence of such aging-induced increase in mTORC1 activity would be repression of nutrient-inhibited stress responses such as autophagy. Such repression would also make aging cells less resilient to nutrient stress, which is known to reduce longevity⁷⁶. Future experiments will determine if such mTORC1-epigenetic bidirectional communication occurs, and whether deregulation of this process contributes to aging or diseases associated with mTORC1 dysregulation, such as cancer, neurodegenerative diseases, or metabolic disorders.

Methods

Yeast strains, plasmids, and cell growth conditions

The YNL406 (H3WT) and YNL403 (H3K37A) strains used in this study are ura3Δ derivatives from the histone H3/H4 library described by Dai et. al.³⁵, which is ultimately derived from S288C. Yeast strain engineering, including gene deletion or epitope tagging, was performed as previously described⁷⁷. All yeast strains used in this study are listed in Supplementary Table S3. Unless otherwise specified, yeast strains were cultivated in either nutrient-rich YPD media (1% yeast extract, 2% peptone, and 2% glucose) or in nutrient-defined SC media formulated with 0.2% yeast nitrogen base devoid of amino acids, 0.5% ammonium sulfate, 2% dextrose, and 0.19% amino acids. Dropout media to select for plasmid maintenance was prepared by leaving out the specific nutrient. Plasmids used in this study are listed in Supplementary Table S4. For solid media preparation, 2% agar was incorporated into the media, and specific inhibitors, such as rapamycin, nicotinamide (NAM), sodium butyrate (SB), or N-acetyl cysteine (NAC), were added just before pouring plates. In all experiments conducted, cells were cultivated in an incubated shaker at 30°C and harvested during the mid-exponential phase (OD₆₀₀= 0.8–1.0) as determined by using a SmartSpect^™ Plus spectrophotometer. All yeast growth media and reagents were obtained from Research Products International, US Biologicals, Fisher Scientific, and Millipore Sigma.

Spot assays

Cells were cultured overnight at 30 °C in either YPD or SC media, or SC media lacking the appropriate nutrient to maintain plasmid selection. Following overnight culturing, the OD₆₀₀ was determined, and an equivalent of OD₆₀₀=1.0 number of cells was harvested and serially diluted at a 1:5 ratio before being spotted onto the designated plates. The plates were then incubated at 30 °C, and images were captured using a MultiDoc-It Digital Imaging System at the indicated time in days.

Site-directed mutagenesis

We cloned the HHT2-HHF2 genetic locus into the 2μ plasmid pRS425. The resulting HHT2-HHF2 plasmid was then used as a template for site-directed mutagenesis as follows. Primers were designed using the online software NEBaseChangerTM to change H3K37 to H3K37A and H3K37R (see Supplementary Table S5 for primer details). Exponential PCR using the high-fidelity NEB Q5 enzyme was performed for each primer set to create mutations at H3K37A and H3K37R. The reaction mixes included 1X Q5 reaction buffer, 0.5 μM of the forward and reverse primers, 200 μM dNTPs, 0.02 U/μL of Q5 high-fidelity DNA polymerase, 25 ng of plasmid template, and nuclease-free water up to 25 μL. The mixture was incubated at 95 °C for 5 minutes for the initial denaturation, then subjected to 25 cycles at 95 °C for 1 minute, 64 °C for 1 minute, and 72 °C for 3 minutes and 40 seconds. Final extension was performed at 72 °C for 10 minutes, and the PCR products were kept at 4 °C. To remove the template plasmid, the PCR products were treated with Kinase, Ligase, & DpnI (KLD). The KLD mix contained 1 μL of the PCR product, 1X KLD reaction buffer, 1X KLD enzyme mix, and nuclease-free water to 10 μL. Samples were mixed well by pipetting, then incubated at room temperature for five minutes. After incubation, 5 μl of the KLD mix was added to ice-thawed 50 μl of JM109 competent cells and mixed, and the mixture was then incubated on ice for 30 minutes. Samples were heat-shocked at 42 °C for 30 seconds and then placed back on ice for five minutes. Subsequently, 950 μl of room temperature LB medium was added, and the mixture was incubated at 37 °C for 60 minutes with shaking. Cells were then pelleted by centrifugation, resuspended in 100 μl of LB medium, spread onto a 100 mg/mL ampicillin plate, and then incubated overnight at 37 °C. Colonies were picked, the plasmid was purified, and then screened for positive clones by Sanger sequencing.

RNA isolation

RNA was isolated using the hot acid phenol approach⁷⁸. Briefly, yeast cells were grown in 10 mL of SC media to OD₆₀₀=0.8 – 1.0, pelleted, and the supernatant discarded. Pellets were resuspended in 1 mL of ice-cold water, transferred to a clean 1.5 mL microcentrifuge tube, pelleted again, and then the supernatant was removed. Cells were resuspended in 400 μL of TES solution (10 mM Tris HCl (pH 7.5), 10 mM EDTA, and 0.5% SDS), and then 400 μL of acid phenol was added before vortexing vigorously for 10 seconds. Samples were then incubated for 45 minutes at 65 °C with occasional brief vortexing, and then they were placed on ice for five minutes before centrifuging at 4 °C at 15000×rpm for five minutes. The aqueous phase was transferred to a clean 1.5 mL microcentrifuge tube, and the acid phenol treatment was repeated. After isolating the aqueous phase from the second acid phenol extraction, 400 μL of chloroform was added, the samples were vortexed vigorously, and then centrifuged for five minutes at 15,000×rpm at 4 °C. The aqueous phase was transferred to a new tube, ethanol precipitated for 30 minutes and then pelleted at 4 °C. Pellets were washed with ice-cold 70% ethanol, pelleted again, and then resuspended in 40 μL of RNase-free water and incubated at 65 °C for 3–5 minutes to dissolve the RNA pellets. All RNAs were digested with RQ1 DNase (Promega) in a 50 μL total reaction volume, incubated at 37 °C for 45 minutes. RNA was purified by phenol/chloroform extraction, precipitated, and then resuspended in 80 μL of nuclease-free water. RNA concentrations were determined by Nanodrop analysis and then stored at −70 °C.

RNA sequencing

RNA libraries were created using the Qiagen FastSelect RNA kit, and the libraries were quantified by Qubit and Qiaxcel RNA size selection. Libraries were then sequenced on Illumina NextSeq 2000 using 2×50 P3 chip, and 20 million reads per sample were generated. Each fastq was subjected to quality control and correction using FASTQC and FASTX trimmer. The trimmed fastq was aligned to the S288 reference genome using RNA-STAR. The SAM files were then mined for the alignment of primary counts using feature Counts. The counts are normalized using TMM normalization and Log2 transformed. The transformed data matrix was loaded into R to gather statistics and determine differential expression. The mean, variance, standard deviation, and standard error of means were calculated for each gene across each condition. A principal component analysis was performed to determine different clusters in the samples. Limma with a linear regression model was used to identify genes with log2 fold change >= 0.58 and p value < 0.05. P values were then adjusted for multiplicity using the Benjamini-Hochberg method. Only genes with an adjusted p-value < 0.05 were considered differentially expressed. These genes were used to create a heatmap with unsupervised hierarchical clustering. The gene list was loaded into String DB for pathway and functional analysis. Sequencing data have been deposited at the National Institutes of Health Gene Expression Omnibus (GEO) repository with accession number GSE312461.

cDNA synthesis and real-time quantitative PCR (RT-qPCR)

The first-strand cDNA was synthesized using the ImProm-II^™ Reverse Transcription System from Promega. Briefly, 2 μg of template RNA was reverse transcribed using a modified manufacturer’s protocol. First, the template RNA was mixed with 1 μL of 500 μg/mL oligo(dT)₁₅ primer and nuclease-free water up to 5 μL and placed on ice. Samples were then preheated in a thermal cycler at 70 °C for five minutes before being transferred immediately to ice for five minutes. Subsequently, 15 μL of RT reaction mix (containing 4 μL of 5X reaction buffer, 4 μL of 25 mM MgCl2, 1 μL of 10 mM dNTP, 1 μL of 40 U/μL Ribonuclease Inhibitor, 1 μL of 160 U/μL Reverse Transcriptase, and 6 μL of nuclease-free water) was added to each sample. The total 20 μL reaction mix was incubated at 25 °C for five minutes, shifted to 42 °C for 60 minutes, and then inactivate the reaction by incubating at 70°C for 15 minutes. The resulting cDNA was diluted 1:10 before use. The RT-qPCR was performed using PowerUp^™ SYBR^™ Green Master Mix according to the manufacturer’s protocol. Briefly, the total reaction volume of 10 μL was prepared with 1 μL of cDNA, 0.75 μL of 5 μM forward and reverse primer mix, 5 μL of SYBR Green, and 4.25 μL of nuclease-free water. Amplification was performed using the StepOnePlus PCR system (Applied Biosystems). The thermal cycling conditions involved UDG activation at 50 °C for two minutes, followed by activation of dual-lock DNA polymerase at 95 °C for two minutes, and then 40 cycles of denaturation at 95 °C for three seconds and annealing/extension at 60 °C for 30 seconds. The cycle threshold (Ct) at which the fluorescence intensity crossed the threshold was used to compare samples. Target genes were normalized to the signal of the endogenous control ACT1 (Supplementary Table S5 for primer details). Results were expressed as fold changes in relative expression. All statistical analyses and graphs were generated using GraphPad Prism 10.

Immunoblot (IB) analysis

The IB analysis for the proteins of interest was performed by collecting total protein from mid-log phase cells (OD₆₀₀=0.8 – 1.0). Briefly, 50 mL of cultured cells were pelleted, washed with water, and then lysed by adding glass beads and bead beating in 120 μL of lysis buffer (10% glycerol, 10 mM Tris (pH 8.0), 0.1% NP40, 300 mM NaCl) with freshly added protease inhibitors (2 μg/mL leupeptin, 2 μg/mL pepstatin, 2 μg/mL aprotinin, 1 mM Dithiothreitol (DTT), 1x protease inhibitor cocktail and 1 mM PMSF). Lysed extracts were clarified at 4°C for 15 minutes, and the supernatant was then collected. The protein concentration was then quantified by Bradford assay, and protein extracts were stored at −70 °C. Samples (40 μg) were prepared in 2X sample buffer, denatured by boiling for five minutes, and then resolved by either 10% or 12% SDS-PAGE. Resolved proteins were transferred to a 0.45 μm PVDF membrane, and then the PVDF membrane was washed with 1X TBST buffer (Tris-buffered saline (TBS) and 1% Tween-20) before blocking with 5% milk/TBST for one hour at room temperature. Blots were then incubated with primary antibody to the following proteins and/or modified proteins: H3 (cat# 39763), H4 (cat# 39269), H2A (cat# 39235), H3K36me3 (cat# 61101), and H3K23Ac (cat#38131) from Active Motif; G6PDH (cat# A9521) and FLAG-M2 (cat# F1804) from Sigma-Aldrich; RPS6/S6 (cat# ab40820) and Tubulin (cat# ab184970) from Abcam; phosphoS6 (Ser235/236) (cat# 22115) from Cell Signaling; HA (cat# SC805) from Santa Cruz; H3K36me2 (gifted from Dr. Strahl)⁴⁴. Primary antibodies were diluted in 5% milk/TBST buffer and incubated at 4 °C overnight with gentle shaking. The next day, the blot was washed three times with 1X TBST, incubated with HRP-conjugated secondary antibody (Mouse Anti-Rabbit peroxidase-conjugated IgG (cat# 211-032-171) and Goat Anti-Mouse Peroxidase-conjugated IgG (cat# 115-035-174) from Jackson Immuno Research) in blocking buffer at room temperature for one hour, and then the immunoblots were developed using the Immobilon^™ Western Chemiluminescent HRP Substrate system. IBs were analyzed and quantified with ImageJ, and the band intensity data were used to generate graphical data using GraphPad Prism.

Cell death analysis

The cell death analysis was performed using PI and YO as described previously with minor modifications²¹. Briefly, cells were grown in SC media overnight to an OD₆₀₀=0.6–0.8 and then either mock-treated or treated with 50 nM rapamycin for six hours before harvesting. An OD₆₀₀=1.0 cells were pelleted, washed with cold PBS, and then collected by centrifugation. Cell pellets were resuspended in 1 mL cold PBS, and then 5 μg/mL of PI and 5 μM of YO were added, mixed well, and incubated at room temperature for 20 minutes. Subsequently, 100,000 events of each stained cell sample were recorded on a Bio-Rad ZE5 flow cytometer using 488 nm excitation, and then data were collected from either a 530/30 bandpass nM green fluorescence emission for YO, or 610/20 bandpass nM red fluorescence emission for PI. Gating on cells was performed to exclude debris using single-color stained controls with standard compensation. The cell populations were categorized into three groups: live cells with low green fluorescence, apoptotic cells with higher green fluorescence, and dead cells with both red and green fluorescence. The cell count percentages of each population were used to generate graphical data in GraphPad Prism.

Cell cycle analysis

Cells were grown in SC medium overnight to an OD₆₀₀=0.8–1.0, and cell cycle analysis was performed as previously described with minor modifications^79,80. In brief, 1.5 mL culture was transferred into a 15 mL centrifuge tube and pelleted at 3000xg for five minutes. Cell pellets were washed once with 1 mL cold MilliQ water and then pelleted again at 3000xg for five minutes. Pellets were resuspended in 1.5 mL cold MilliQ water before adding 3.5 mL of ice-cold ethanol (100%) dropwise while vortexing to prevent clumping, and then the samples were either incubated at room temperature for one hour or overnight at 4 °C. Cells were pelleted at 3000xg for five minutes, and the ethanol was discarded. Cell pellets were washed twice with 1 mL ice-cold sodium citrate buffer (50 mM sodium citrate, pH 7.2 adjusted with citric acid) by centrifuging at 3000xg for five minutes. The pellets were resuspended in 300 μL of RNase A solution (2 mg/mL of RNase A in 50 mM Tris (pH 8) and 15 mM NaCl) and then transferred to a 1.5 mL centrifuge tube before incubating at 37 °C for four hours with periodic mixing. After incubation, cells were collected by centrifuging at 3000xg for five minutes and then resuspended directly in 500 μL of freshly prepared protease solution (5 mg/mL pepsin in water with 4.5 μL/mL concentrated HCl added). The cell suspension was incubated at 37 °C for 20 minutes and then the cells were pelleted by centrifuging at 3000xg for five minutes. Cells then resuspended in 500 μL of ice-cold sodium citrate buffer and either kept on ice or at 4 °C until staining. For staining, 50 μL of a freshly prepared 1 mL SYTOX solution (1 μM SYTOX in sodium citrate buffer) was added to the cell suspension and then incubated at room temperature for 10 minutes. The tube containing the cells was transferred to ice and then sonicated for 15 seconds at low power (20% amplitude) with the tube kept on ice throughout the process. The sample was recorded as 20,000 events in the Bio-Rad ZE5 flow cytometer and analyzed using FlowJo software as previously described⁸⁰.

MNase assay

The MNase assay was performed as described previously with minor modifications^39,81. Briefly, cells were grown in 100 mL SC media overnight and then collected at an OD₆₀₀=0.8–1.0 by centrifuging at 3000xg for five minutes. Cell pellets were washed with 1.5 mL of cold water and then pelleted at 3000xg for five minutes before either continuing or freezing samples at −80 °C. To perform the MNase assay, cell pellets were washed in cold PSB buffer (20 mM Tris-HCl (pH 7.4), 100 mM NaCl, and 10 mM 2-β-mercaptoethanol (2-βME)), followed by cold ST buffer (1 M sorbitol and 20 mM Tris-HCl (pH 7.4)). Cell pellets were resuspended in 1.5 mL of cold ST buffer and measure the initial OD at OD₆₀₀ by diluting 30 μL of cells in 1 mL of water, and then 10 units/mL of Zymolyase 20T in 1 mM DTT was added before incubating at 30 °C for approximately 30 minutes. Again, measure the OD₆₀₀ by diluting 30 μL of cells in 1 mL of water. Once the OD₆₀₀ was less than 80% of the initial starting OD₆₀₀, a fresh ST buffer was added to the top of the tube, and spheroplasts were pelleted by centrifuging at 7500 rpm for five minutes at 4 °C. The supernatant was discarded, and the spheroplasts were then washed with 1.5 mL of cold ST buffer. Spheroplasts were pelleted again at 7500 rpm for five minutes at 4 °C, and then they were lysed in 800 μL of F-buffer, pH 6.5 (18% Ficoll-400, 40 mM K-phosphate monobasic (pH 7.0), 1 mM MgCl₂), to which 5 mM 2-βME and protease inhibitors (10 μg/mL leupeptin, 1 μg/mL pepstatin, 1 μg/mL aprotinin, 1x protease inhibitor cocktail and 1 mM PMSF) were freshly added and then samples were incubated on ice for five minutes. Separately, a new centrifuge tube was prepared containing 600 μL FG-buffer pH 6.5 (7% Ficoll-400, 20 % glycerol, 40 mM K-phosphate monobasic (pH 7.0), 1 mM MgCl_2, and freshly added 5 mM 2-βME and protease inhibitors. Spheroplast lysate was layered on top of the FG-buffer and then centrifuged at 12,500 rpm for 20 minutes at 4 °C. The supernatant was then discarded and the nuclear pellet then placed on ice. Nuclei were resuspended in 500 μL of MNase digestion buffer (10 mM HEPES (pH 7.5), 35 mM NaCl, 0.5 mM MgCl₂, and 0.5 mM CaCl₂) and divide into 5 aliquots before adding the following MNase concentrations: 0 U (undigested), 2.5 U, 5 U, 7.5 U, and 10 U. Samples were incubated at 37 °C for 10 minutes and then 12.5 μL of MNase stop solution (65 mM EDTA, 7% SDS) was added to stop the reactions. To each sample, 20 μL of 2 mg/mL RNase A was then added and incubated at 37 °C for one hour. After incubation, 20 μL of 10% SDS and ~1 M of potassium acetate was added, and then the aqueous phase was isolated by phenol/chloroform extraction and maximum centrifugation for 10 minutes at room temperature. The purified DNA was ethanol precipitated, washed with 70% ethanol, and then resuspended in 20 μL of water. The results were visualized by resolving on a 2% agarose gel prestained with Midori green safe stain.

Reactive oxygen species (ROS) quantification

ROS levels were quantified by DHE staining as previously described with minor modifications⁸². Cells were grown in SC media overnight before mock treating or treating with 50 nM rapamycin for 24 hours at an OD₆₀₀ =0.8–1.0. After treatment, an equivalent of OD₆₀₀= 1.0 number of cells was centrifuged, washed twice with PBS, and then resuspended in 200 μL of staining buffer (0.1% glucose and 1 mM EDTA, pH 8.0, in PBS). Cells were transferred to a 96-well black flat-bottom microplate, and then 30 μM of DHE was added to each well and incubated at 30 °C for 15 minutes in the dark. ROS levels were measured using a fluorescence plate reader (BioTek Cytation 5) with an excitation wavelength of 520 nm and an emission wavelength of 600 nm^83,84. For experimental positive control, cells were treated with 10 mM hydrogen peroxide to increase ROS levels, while 50 mM NAC was used to confirm ROS quenching. The fluorescence readings of each group are used to generate graphical data with GraphPad Prism.

Software and statistical analysis

All experiments were conducted with n = 3 or more independent biological replicates, as indicated in the figure legends. GraphPad Prism was used to calculate statistical significance. For comparing multiple groups, a one-way ANOVA with Bonferroni's multiple comparisons test was performed. For comparing two groups, an unpaired t-test was used. Results are presented as the mean ± standard deviation (SD) with a 95% confidence interval, and a p-value of <0.05 was considered statistically significant. To represent the significance, we used (*) for a one-way AVONA and (^#) for the t-test. Hypergeometric testing was used to test significance of transcriptome overlap for the data represented in Supplementary Fig. S4E using the online software (https://flaski.age.mpg.de/venndiagram/).

Supplementary Material

Supplementary Figure 1. (A) The H3WT and H3K37A cells were transformed with either a control vector or a vector expressing all four core histones, and then 5-fold serially diluted and spotted to the indicated plates for 3–8 days. (B) Whole cell extracts from cells in A were analyzed by IB for histone expression. (C-E) Quantification of histones H3, H4, and H2A normalized to α-tubulin. n=3–4; *p<0.05, ****p<0.0001. (F-I) RT-qPCR analyses were performed for each of the indicated histone genes. n=3–4; ns – no significance.

Supplementary Figure 2. (A-B) H3WT and H3K37A were grown to mid-exponential phase, stained with SYTOX Green, and analyzed by flow cytometry. The histogram displays the SYTOX green fluorescence intensity (DNA content) on the x-axis and cell count on the y-axis. Data are representative of the cell cycle results represented in Figure 2A. (C) H3WT and H3K37A were treated with 2.5U, 5U, 7.5U, and 10U of MNase for 10 minutes at 37 °C before resolving samples on a 2% agarose gel. Lane 1 – 100 base pair (bp) DNA ladder, Lane 2 – 6 H3WT, Lane 7 empty, Lane 8 – 12 H3K37A. UD – Undigested.

Supplementary Figure 3. (A) The dual fluorescence scatter plot flow cytometry analysis of H3WT, H3K37A, and their set2Δ derivatives cells that were mock or 50 nM Rap-treated for 6 hours. YO PRO-1 (YO) on the X-axis and propidium iodide (PI) on the Y-axis. Data are representative for the cell death results in Figure 3B. (B) H3WT and H3WT set2Δ were transformed with the control vector, or with the vector expressing either wild-type Set2 or the indicated Set2 catalytic mutants (in H3WT set2Δ), and H3K36me2, H3K36me3, and total H3 were analyzed by IB. (C) IB analysis of histone H3K23Ac and total H3 in H3WT, H3K37A, and their set2Δ derivatives. (D) Quantification of C. n=3; *p<0.05, **p<0.01 and ^#p<0.05.

Supplementary Figure 4. Bubble blot analysis of the cellular component gene ontology (GO) categories for the (A) H3WT (mock vs rapamycin), (B) H3WT vs H3K37A (mock), (C) H3WT mock vs H3K37A rapamycin, (D) H3WT rapamycin vs H3K37A rapamycin. GO analysis was performed through the STRING database. FDR – False discovery rate. (E) Venn analysis of differentially expressed genes (DEGs) from H3WT vs H3K37A (current study), WT vs set2Δ (McDaniel et al., 2017), and WT vs set2Δ (Li et al., 2023). Hypergeometric testing was used to analyze significance using a population size of 6275 total genes.

Supplementary Figure 5. (A-F) RT-qPCR analysis for the indicated histone genes. n=6; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, and ^#p<0.05, ^##p<0.01, ns – no significance. (G) The H3WT, H3K37A, and the hho1Δ derivatives of each were serially diluted 5-fold and spotted to the indicated plates for four days.

Supplementary Figure 6. (A-D) RT-qPCR analysis of the indicated overexpressed mitochondrial genes from the H3K37A+rapamycin sample as identified by RNA-seq. n=3; *p<0.05, **p<0.01, ***p<0.001, ns – no significance. (E-F) The graphical data of log2-normalized counts for CIT1 and CIT2 from the RNA-seq analysis. n=3 – 4; *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001, ns – no significance. (G) H3WT was either mock-treated or treated with 20 nM rapamycin, 50 mM NAC, or both inhibitors for one hour before analyzing mTORC1 activity by S6ph/S6 IB.

Funding Statement

This work was supported by the National Institutes of Health [R01GM138393 to R.N.L. and R35GM126900 to B.D.S].