Tra1 controls the transcriptional landscape of the aging cell

Abstract Gene expression undergoes considerable changes during the aging process. The mechanisms regulating the transcriptional response to cellular aging remain poorly understood. Here, we employ the budding yeast Saccharomyces cerevisiae to better understand how organisms adapt their transcriptome to promote longevity. Chronological lifespan assays in yeast measure the survival of nondividing cells at stationary phase over time, providing insights into the aging process of postmitotic cells. Tra1 is an essential component of both the yeast Spt-Ada-Gcn5 acetyltransferase/Spt-Ada-Gcn5 acetyltransferase-like and nucleosome acetyltransferase of H4 complexes, where it recruits these complexes to acetylate histones at targeted promoters. Importantly, Tra1 regulates the transcriptional response to multiple stresses. To evaluate the role of Tra1 in chronological aging, we took advantage of a previously characterized mutant allele that carries mutations in the TRA1 PI3K domain (tra1Q3). We found that loss of functions associated with tra1Q3 sensitizes cells to growth media acidification and shortens lifespan. Transcriptional profiling reveals that genes differentially regulated by Tra1 during the aging process are enriched for components of the response to stress. Notably, expression of catalases (CTA1, CTT1) involved in hydrogen peroxide detoxification decreases in chronologically aged tra1Q3 cells. Consequently, they display increased sensitivity to oxidative stress. tra1Q3 cells are unable to grow on glycerol indicating a defect in mitochondria function. Aged tra1Q3 cells also display reduced expression of peroxisomal genes, exhibit decreased numbers of peroxisomes, and cannot grow on media containing oleate. Thus, Tra1 emerges as an important regulator of longevity in yeast via multiple mechanisms.


Introduction
Saccharomyces cerevisiae has been extensively studied as a eukaryotic model for lifespan regulation since many of the underlying molecular mechanisms are conserved from yeast to mammals (Mortimer and Johnston 1959;Fontana et al. 2010;Kaeberlein 2010;Mirisola and Longo 2022). Chronological lifespan (CLS, defined as the extent of time nondividing cells survive in a nutrientdeprived environment; MacLean et al. 2001;Laun et al. 2006;Murakami et al. 2008) and replicative lifespan (the number of cell divisions before replicative senescence; Steinkraus et al. 2008) are 2 distinct experimental aging models (Laun et al. 2006). CLS experiments are carried out by culturing cells for a substantial period at the stationary phase. During chronological aging, yeast cells go through distinct growth phases: the mid-log phase, diauxic shift, and stationary phase. Each growth phase is illustrated by divergent metabolic activities and gene expression profiles, which are analogous to crucial characteristics of mammalian aging cells including cell cycle arrest, increased respiratory activity, and lipid and protein homeostasis (Tissenbaum and Guarente 2002;Beach et al. 2013Beach et al. , 2015Medkour and Titorenko 2016;Chadwick and Lajoie 2019;Chadwick et al. 2020;Mohammad et al. 2020). Studies of yeast chronological aging have enabled researchers to identify several key pathways that regulate longevity (Longo and Fabrizio 2012) such as Ras/Pka (Longo 1999) and Sch9/Tor (Fabrizio et al. 2001) and the role of sirtuins (Kaeberlein et al. 1999), not only in yeast but also in other organisms.
How cells modulate their gene expression in response to stresses including aging involves all components of the transcriptional machinery. The recruitment of RNA polymerase II to promoters is tightly regulated by general transcription factors, gene-specific activators, and coactivators. Coactivators are multisubunit protein complexes that interface between general transcription factors and gene-specific activators (Kingston et al. 1996;Eberharter and Becker 2002;Helmlinger et al. 2011) and/or regulate modification of histone proteins and nucleosome remodeling Rando and Winston 2012). Thus, coactivators play a vital role in global gene regulation. Spt-Ada-Gcn5 acetyltransferase (SAGA) and nucleosome acetyltransferase of H4 (NuA4) are prototypical multisubunit coactivator protein complexes that are conserved among eukaryotic organisms. SAGA and NuA4 contain lysine acetyltransferases Gcn5 and Esa1, respectively, which acetylate both histone and nonhistone proteins (Grant et al. 1997;Clarke et al. 1999;Steunou et al. 2014;Downey 2021). Transcriptome analyses for yeast strains lacking the SAGA components Gcn5 or Spt3 revealed that 10% of the stress-related genome is controlled by the SAGA complex (Huisinga and Pugh 2004). SAGA and NuA4 also participate in other non-transcriptional activities, such as telomere maintenance and DNA repair (Bird et al. 2002;Downs et al. 2004;Lin et al. 2008;Atanassov et al. 2009;Cheng et al. 2018Cheng et al. , 2021. In addition, SAGA contains a deubiquitinating module (DUB) that includes the ubiquitin protease Ubp8 that cleaves ubiquitin from histone H2B (Henry et al. 2003;Morgan et al. 2016) and other targets. The additional targets for deubiquitination by the mammalian homolog of Ubp8, USP22, are particularly noteworthy for their role in cancer progression (Prokakis et al. 2021;Ning et al. 2022;De Luca et al. 2022).
Tra1 is uniquely found in both the SAGA and NuA4 complexes (Saleh et al. 1998;Grant et al. 1998;Allard et al. 1999). Tra1 belongs to the phosphoinositide-3-kinase-related kinase (PIKK) family, which includes ataxia telangiectasia mutated (ATM; Tel1 in S. cerevisiae), ataxia telangiectasia, and Rad3 related (ATR; Mec1 in S. cerevisiae), the DNA-dependent protein kinase catalytic subunit (DNA-PKc), mammalian target of rapamycin (Tor1 and Tor2 in S. cerevisiae), and SMG-1 (suppressor with morphological effect on genitalia family member) (Smith and Jackson 2003;Hill and Lee 2010;Shiloh and Ziv 2013). PIKK proteins have 4 common principal domains: an N-terminal Huntingtin, elongation factor 3 (EF3), protein phosphatase 2A (PP2A), and TOR1 (HEAT), followed by FRAP-ATM-TRRAP (FAT), phosphatidylinositol 3-kinase (PI3K), and FRAP-ATM-TRRAP C-terminus (FATC) domains (Keith and Schreiber 1995;Bosotti et al. 2000;Mordes et al. 2008). Approximately, half of the Tra1 molecule consists of helical HEAT repeats and this section interacts with activator proteins to initiate transcription (Brown et al. 2001;Bhaumik et al. 2004;Knutson and Hahn 2011;Lin et al. 2012). A helical FAT domain located C-terminal to the HEAT repeats wraps the N-terminal section of the PI3K domain (Díaz-Santín et al. 2017;Sharov et al. 2017). At the C-terminus, PIKK family members share a highly conserved phosphoinositide-3-kinase (PI3K) domain. The PI3K domain consists of N-and C-terminal subdomains with a cleft between these 2 lobes, where ATP binds in the catalytically active PIKK family members (Huse and Kuriyan 2002;Taylor and Kornev 2011;Pavletich and Yang 2013). In all PIKK family members except Tra1/TRRAP, the PI3K domain regulates cell signaling by phosphorylating downstream target proteins. The PI3K domain of Tra1 lacks residues essential for ATP binding and phosphate transfer and therefore has no demonstrable kinase activity (Saleh et al. 1998;Nelson et al. 2006;Mutiu et al. 2007). Interestingly, however, mutation of conserved arginine residues to glutamine (termed tra1 Q3 ) in the kinase cleft domain results in slow growth, increased stress sensitivity, and decreased transcription of SAGA-regulated genes supporting a role for this pseudokinase domain (Berg et al. 2018).
In light of the importance of epigenetic changes in aging (Booth and Brunet 2016;Kane and Sinclair 2019), the involvement of SAGA and NuA4 in stress response processes, and the unique role of Tra1 in both SAGA and NuA4 (Helmlinger et al. 2011;Cheung and Díaz-Santín 2019), we hypothesized that Tra1 contributes to longevity by regulating the transcriptional responses to stress associated with aging. Indeed, using a loss-of-function mutant, we find that Tra1 significantly alters the transcriptional landscape of the aging cell. Tra1 therefore emerges as a new regulator of the chronological aging process in yeast.

Drugs
Oleic and myristic acid and diamide were purchased from Milipore-Sigma. Propidium iodide was from Thermo Fisher Scientific. 2-(N-morpholino)ethanesulfonic acid (MES) buffer solution was from Alfa Aesar.

Strains and plasmids
Yeast strains were constructed in the W303A background or BY4742 and are described in Supplementary Table 1. Strains expressing either wild-type FLAG-tagged TRA1 or tra1 Q3 were created by integrating an SphI-SacI fragment from pCB2527 or pCB2537, respectively, into W303A and selecting HISþ cells expressing either wild-type URA3-FLAG-tagged TRA1 or tra1 Q3 PI3K domain and YCplac111-DED1pr-YHR100 as previously described (Berg et al. 2018). yemRFP-SKL was generated by cloning yemRFP-SKL into the SpeI/SalI sites of pRS416-GDP (Mumberg et al. 1995).

Culture conditions and cell viability assay
Both the TRA1 and tra1 Q3 yeast cells (W303A derivatives) were grown to saturation overnight in synthetic complete medium (2% glucose with appropriate selection). An aliquot of the overnight cultures was diluted in fresh media and grown at 30 C in a rotating drum. Cell viability assays were conducted using propidium iodide as described previously (Chadwick et al. 2016). Briefly, cells were washed and resuspended in phosphate-buffered saline (PBS) containing 1 mM of propidium iodide. The positive control was prepared by boiling cells for 10 min before resuspending cells in propidium iodide staining solution. Unstained cells were used as a negative control. All samples were incubated for 10 min at room temperature in 96-well plates before imaging on a Gel doc system (Bio-Rad). The optical density (OD 600 ) was measured using a BioTek Epoch 2 microplate reader. Both the TRA1 and tra1 Q3 cells were cultivated for several days and the cell viability of the chronologically aged cells was quantified at various time points throughout the aging process. Alternatively, the CLS assay was performed using caloric restricted medium (0.1% glucose), medium buffered with 0.1 M MES buffer, or media containing 0.1% oleic acid, 0.1% myristic acid, or 2% glycerol. To assess sensitivity to acetic acid, cells were grown in synthetic complete medium for 4 days and treated with acetic acid for 200 min with concentrations up to 0.08 mM and the cell viability assessed using propidium iodide. Survival rates were computed using the mean gray value from images along with OD 600 using the ANALYSR program (Chadwick et al. 2016).

Growth assays
A single colony of yeast strain was inoculated and grown overnight in 2% synthetic medium without leucine at 30 C, 220 rpm. Overnight cultures were serially diluted and OD 600 measured using a spectrophotometer. Cell cultures were normalized to OD 600 0.1-and 5-fold serial dilutions were spotted on solid media. Plates were incubated for 2 days at 30 C before taking images using a colony imager (S&P Robotics). Growth was quantified as previously described (Petropavlovskiy et al. 2020).

RNA isolation and sequencing
Both TRA1 and tra1 Q3 cells were grown to saturation overnight, diluted 10-fold in 2% synthetic minus leucine medium and grown for 4 h at 30 C in a rotating drum. An aliquot of the cells was spun down at day 0 and day 3 and stored at À80 C. Total RNA was isolated from cells using the RiboPure yeast kit (Thermo Fisher Scientific) according to the manufacturer's instructions. RNA samples were prepared with 3 replicates for each genotype and time point. Quality of the RNA samples were analyzed using a Bioanalyzer to ensure a RIN value of 8 or higher.
Total RNA sequencing analysis was conducted by Azenta Life Sciences. RNA from each sample was converted into single stranded Illumina TruSeq cDNA libraries with poly dT enrichment. Libraries were sequenced on an Illumina HiSeq and each sample yielded between 27.5 and 39.7 million 150-bp paired-end sequencing reads. The raw reads and RNA count data were deposited in NCBI's Gene Expression Omnibus (Edgar et al. 2002).
Quality control, trimming, read alignment, and differential gene expression analysis Read quality was analyzed using FastQC (www.bioinformatics. babraham.ac.uk/projects/fastqc/). Trimmomatic default settings were used to trim low quality bases and adapter sequences (Bolger et al. 2014). Next, reads were aligned with the reference genome of S. cerevisiae S288C sequence (R64-2-1; www.yeastge nome.org/) using STAR (Dobin et al. 2013). Only uniquely mapping reads were retained. featureCount was used to count the reads mapping to each gene (Liao et al. 2014). Differential expression analysis was performed with DESeq2 (Love et al. 2014). Differentially expressed genes with a Benjamini-Hochberg adjusted P-value cutoff of 0.05 were considered for further analysis.
To determine the enrichment of Rlm1 targets, promoter analysis of differentially expressed genes (log 2 fold change <2 and Pvalue cutoff of 0.05) was performed using Yeastract (www.yeas tract.com) (Monteiro et al. 2020).

Fluorescence microscopy
TRA1 and tra1 Q3 cells expressing yemRFP-SKL were grown to saturation overnight, diluted 10-fold in 2% synthetic minus leucine medium and grown for 4 h at 30 C in a rotating drum. An aliquot of the cells was then spun down, washed, resuspended in PBS, and transferred to a Nunc Lab-Tek chamber slide. Cells were imaged using a Zeiss Axiovert A1 wide-field fluorescence microscope with 63Â 1.4 NA oil objective using a Texas Red filter (586 nm excitation/603 nm emission) and an AxioCam ICm1 R1 CCD camera. ImageJ was used to analyze the images (Schneider et al. 2012).

Functional Tra1 is required for chronological aging
Tra1 is essential in yeast (Saleh et al. 1998) with the exception of the fission yeast Schizosaccharomyce pombe, which expresses a second TRA2 isoform (Helmlinger et al. 2011;Elías-Villalobos, Fort, et al. 2019). To determine the role of Tra1 in the aging process, we took advantage of a loss-of-function mutant that we previously characterized termed tra1 Q3 (Berg et al. 2018). tra1 Q3 contains 3 glutamine substitutions of arginine residues 3,389, 3,390, and 3,456 proximal or within the putative ATP-binding cleft of the PI3K domain. tra1 Q3 results in slow growth, increased stress sensitivity, transcriptional defects and impaired SAGA and NuA4 complex assembly (Berg et al. 2018;Razzaq et al. 2021). CLS was assessed by labeling cells with the viability dye propidium iodide (PI) at different time intervals during aging (Fig. 1a). tra1 Q3 cells display significantly reduced lifespan when compared to their wild-type counterparts (Fig. 1, b-d). Viability of tra1 Q3 cells in exponentially growing culture is comparable to wild type. tra1 Q3 cells begin to decline in viability when cells reach the stationary phase indicating that a functional TRA1 allele is required to extend CLS.
Since caloric restriction efficiently increases longevity in yeast and other organisms (Jiang et al. 2000;Fontana et al. 2010;Choi et al. 2013;Arlia-Ciommo et al. 2014;Leonov et al. 2017;Campos et al. 2018), we then tested whether functional Tra1 is required for lifespan extension by caloric restriction. Caloric restriction is achieved by growing the cells in 0.1% glucose medium whereas complete medium contains 2% glucose. We found that when grown under caloric restriction conditions, tra1 Q3 cells display extended lifespan similar to wild-type TRA1 cells (Fig. 2), suggesting that Tra1 function is dispensable for lifespan extension by caloric restriction.
tra1 Q3 cells are more sensitive to acetic acid Culture media acidification is a predominant cell-extrinsic factor associated with cell death during chronological aging (Burtner et al. 2009;Burhans and Weinberger 2009;Hu et al. 2014;Deprez et al. 2018;Chaves et al. 2021). During chronological aging, acetic acid is produced following depletion of glucose from the growth media when ethanol is utilized as the main carbon source. Indeed, long-lived mutants, such as sch9D and ade4D, show increased resistance to acetic acid (Burtner et al. 2009;Matecic et al. 2010). Therefore, we sought to determine if increased sensitivity to acidification of the cell culture media by acetic acid explains, at least in part, the shorter lifespan of tra1 Q3 cells. To this end, TRA1 and tra1 Q3 cells were grown to stationary phase for 4 days and treated with various concentrations of acetic acid. Indeed, tra1 Q3 increases sensitivity to acetic acid compared with wildtype TRA1 (Fig. 3, a and b). Media acidification during chronological aging is alleviated by buffering the cultures to pH 6 with citrate phosphate or MES, resulting in increased longevity (Burtner et al. 2009;Burhans and Weinberger 2009). We found that buffering the aging media pH with MES significantly extended the lifespan of both TRA1 and tra1 Q3 cells (Fig. 3, c and d), suggesting that the inability to respond to acid stress plays an important role in the reduced longevity phenotype associated with the tra1 Q3 mutation.

Tra1 controls the transcriptional landscape of the aging cell
We previously found that Tra1 regulates the expression of genes associated with several stress responses including cell wall stress and the response to misfolded proteins (Berg et al. 2018;Jiang et al. 2019;Razzaq et al. 2021). Based on findings that tra1 Q3 cells have a shorter lifespan, we next sought to determine how the loss of Tra1 function impacts the transcriptional landscape of the aging cell. We performed a global transcriptome analysis using RNA sequencing for TRA1 and tra1 Q3 cells before chronological aging (day 0) and after 3 days at stationary phase (day 3) to identify genes subjected to different transcriptional regulation. Day 3 was selected because the cells have reached stationary phase and still show modest minimal cell death ( Supplementary Fig. 1). We found that $60% of the variance in the transcriptome data is linked to the aging process and $20% is linked to the tra1 Q3 mutation (Fig. 4a). The larger distance between TRA1 and tra1 Q3 cells after aging suggests that transcriptional differences between the 2 strains is greater in aged cells. At day 0, 394 genes were statistically differentially regulated (175 downregulated and 219 upregulated) when comparing TRA1 and tra1 Q3 cells (P < 0.05, log 2 fold change >2) (Supplementary Table 2). At day 3, there were 2,026 statistically differentially expressed genes (978 downregulated and 1,048 upregulated) (Supplementary Table 2). When comparing the age-genotype interaction scores (Smith and Kruglyak 2008; Sardi and Gasch 2018), we found 287 genes (161 negative and 126 positive) with a log 2 score >2.5 and P < 0.05, suggesting that these genes differentially respond to the aging process (Fig. 4b) in the 2 genotypes (TRA1 and tra1 Q3 ). Genotype (TRA1 vs tra1Q3) by environment (day 0 vs day 3) interaction scores allow us to decipher how different genotypes specifically modulate gene expression during the aging process. Gene Ontology (GO) analysis revealed significant enrichment in several biological processes, including RNA binding, transmembrane transporter activity, oxidoreductase activity, and unfolded protein binding (Fig. 4c). Genes associated with misfolded protein stress (SSA1, HSP26, HSP42, SSA4, MID1, KAR2, SSA2, CPR6, HSP60, HSC82, APJ1, HSP10, HSP82) are upregulated in aged tra1 Q3 cells compared to wild-type suggesting that these cells exhibit increased proteotoxic stress. Genes with a negative age-genotype interaction score were also enriched in Rlm1 target genes, consistent with our previous results showing that Tra1 controls cell wall integrity ( Supplementary Fig. 2) (Berg et al. 2018;Razzaq et al. 2021). Genes with the highest positive and negative interaction scores are presented in Fig. 4d. Genes upregulated in aged tra1 Q3 cells compared to wild type include RGI1 (involved in metabolism under respiratory conditions) and BTN2 (a v-snare binding protein). Interestingly, BTN2 expression is increased under severe ethanol stress (Yamauchi and Izawa 2016), reinforcing the idea that tra1 Q3 cells are subject to increased stress during the aging process. Genes with positive age-genotype interaction are upregulated in wild-type TRA1 compared to tra1 Q3 in aged cells. Examples of such genes are SIP18 and MLS1. SIP18 encodes a phospholipid binding hydrophilin involved in vacuolar membrane fusion and is upregulated during replicative lifespan (Ghavidel et al. 2018). MLS1 encodes a malate synthase involved in the utilization of non-fermentable carbon sources and is also upregulated in replicatively aged cells (Lesur and Campbell 2004).
We previously showed that cells expressing tra1 Q3 increase expression of TRA1 as a possible compensatory mechanism for the loss of protein function (Berg et al. 2018). Here, we observed this phenomenon as tra1 Q3 mRNA increased $1.7-fold compared to wild-type at day 0. This difference was exacerbated in aged cells as TRA1 mRNA increased $3.2-fold in tra1 Q3 cells after aging (Fig. 5a). We also analyzed the mRNA levels of other SAGA and NuA4 components (Fig. 5b). Most striking are the upregulation of ADA2 in aged tra1 Q3 cells and downregulation of components of the SAGA DUB module (UBP8, SUS1, SGF73, SGF11). Among the NuA4 components, ARP4 was upregulated in aged tra1 Q3 cells. Interestingly, mutation of ARP4 affects CLS in combination with deletion of the linker histone Hho1 (Vasileva et al. 2021).
GO analysis of genes that respond differentially to the aging processing in tra1 Q3 cells revealed an enrichment in genes associated with oxidoreductase activity (Fig. 6a). Among those genes were CTT1 and CTA1, which encode 2 versions of catalase in S. cerevisiae. Ctt1 is cytoplasmic (Seah and Kaplan 1973) while Cta1 localizes to both the mitochondria and peroxisomes (Petrova et al. 2004). Catalase activity is crucial for hydrogen peroxide detoxification and is an important regulator of oxidative stress resistance associated with various conditions, including aging in yeast (Petriv and Rachubinski 2004;Agarwal et al. 2005;Mesquita et al. 2010;Rona et al. 2015;Guaragnella et al. 2019). Both genes are substantially upregulated in aged wild-type TRA1 cells but not in tra1 Q3 cells (Fig. 6b). Consequently, tra1 Q3 cells were more sensitive to oxidative stress induced by diamide (Fig. 6c). Thus, these data indicate that tra1 Q3 cells have reduced capacity to cope with oxidative damage that is usually associated with the aging process .
Several differentially expressed genes were also linked to peroxisomal b-oxidation and fatty acid metabolism (POT1, POX1, MLS1, DCI1, ECI1, TES1) (Fig. 7a). The b-oxidation pathway and peroxisome proliferation are crucial for chronological aging (Lefevre et al. 2013). In yeast, b-oxidation solely occurs in peroxisomes (Hiltunen et al. 2003) and strains with defective peroxisomes fail to grow in presence of oleate due to their incapacity to use fatty acid as a carbon source (Lockshon et al. 2007). We found that tra1 Q3 cells were unable to grow on media containing oleate or myristate, suggesting that they are defective in b-oxidation (Fig. 7b). Because acetyl-CoA produced by b-oxidation serves as an energy source only in respiratory-competent strains, we assessed growth on glycerol. We found that tra1 Q3 cells are unable to grow on plates containing glycerol (Fig. 7b). This phenotype along with the reduced lifespan was also recapitulated in a different strain background ( Supplementary Fig. 3). Since b-oxidation in yeast is performed solely in peroxisomes, we also analyzed the expression of PEX34 and PEX21 in TRA1 and tra1 Q3 cells during the aging process. Pex34 regulates peroxisome biogenesis (Tower et al. 2011). Pex21 regulates import of protein into the peroxisomal matrix (Purdue et al. 1998). Aged tra1 Q3 cells show reduced expression of PEX34 and PEX21 compared to aged wildtype cells (Fig. 7c). tra1 Q3 cells consequently display a reduced number of peroxisomes when labeled with the fluorescent reporter yemRFP-SKL at both day 0 and day 3 (Fig. 7, d and e). Therefore, our data suggest that tra1 Q3 cells have defective peroxisome function that might contribute to their reduced CLS.

A role for Tra1 in chronological aging
Tra1 has been proposed to act as protein-protein hub where it interacts with transcription activators to recruit SAGA and NuA4 coactivator complexes to specific promoters (Elías-Villalobos, Fort, et al. 2019). In both SAGA and NuA4, Tra1 regulates multiple stress responses associated with the aging process such as cell wall integrity, ethanol sensitivity, protein misfolding, lipid synthesis, and the response to DNA damage. Thus, Tra1 regulates b) Normalized survival rates over the aging process and calculated survival integral are shown in graphs. Significance was assessed using a 1-way ANOVA followed by a Tukey's multiple comparison test. **P < 0.01 and ****P < 0.0001. various aspects of the transcriptional response to the aging process. Interestingly, the Tra1 homolog in Drosophila, Nipped-A, is required to maintain proliferative capacity of intestinal stem cells through aging, indicating that Tra1 may play a role in aging across species (Tauc et al. 2017).
Caloric restriction impacts multiple stress response pathways that are associated with the aging process, such as those involved in proteotoxic stress (Choi et al. 2013;Arlia-Ciommo et al. 2018). Here, we also show that caloric restriction alleviates the accelerated aging phenotype associated with the tra1 Q3 mutantion. Since our data indicate that the tra1 Q3 mutant is subjected to increased stress during aging, it is reasonable to postulate that caloric restriction acts at least in part by alleviating these stresses. Interestingly, while the tra1 Q3 cells do not grow on glycerol, their respiratory defect appears to be less than the threshold required for lifespan extension by caloric restriction (Ocampo et al. 2012).
This study and previous work using the tra1 Q3 mutant (Berg et al. 2018;Razzaq et al. 2021) support an important regulatory role for the Tra1 PI3K domain, despite the absence of key kinase motifs Helmlinger et al. 2011). Our previous work showed that Tra1 Q3 has decreased association with SAGA and NuA components (Berg et al. 2018). This included Spt20, which is the preferential Tra1 interactor in both fission and budding yeast (Liu et al. 2019;Elías-Villalobos, Toullec, et al. 2019;Wang et al. 2020). Moreover, Tra1 regulates the incorporation of DUB module components into SAGA (Elías-Villalobos, Toullec, et al. 2019). Interestingly, we found decreased expression of DUB components in aged tra1 Q3 cells. Whether this is a response to mis-assembly of the complex remains to be determined. Similarly, Leo et al. (2018) demonstrated that SAGA, through Gcn5, is essential to properly maintain Ubp8 levels under respiratory conditions. Aging also exacerbated the increase in TRA1 expression that we previously observed in tra1 Q3 cells Fig. 3. Increased sensitivity of tra1 Q3 cells to media acidification is linked to shortened lifespan. a) tra1 Q3 cells display increased sensitivity to acetic acid. TRA1 and tra1 Q3 cells were grown for 4 days and subsequently treated with the indicated concentrations of acetic acid for 200 min and stained with propidium iodide to measure cell survival. b) Normalized survival rates upon acetic acid treatment were calculated and survival integrals are shown. c) TRA1 and tra1 Q3 cells were grown for the indicated time in standard synthetic complete medium with or without 0.1 M MES and stained with propidium iodide to measure cell survival. Stained cells were imaged in a 96-well plate. For each time point, a negative control (unstained cells), a positive control (boiled cells) and 5 replicates were analyzed. d) Normalized survival rates over the aging process and calculated survival integrals are shown. Significance was assessed using a 1-way ANOVA followed by a Tukey's multiple comparison test. ***P < 0.005 and ****P < 0.0001. Fig. 4. The tra1 Q3 mutation alters the transcriptional landscape of aging cells. a) Principal component analysis of centered log ratio of normalized reads from TRA1 and tra1 Q3 cells at day 0 and day 3. Each point represents a single biological replicate (n ¼ 3). b) Volcano plot of genes that respond differently to the aging process in the tra1 Q3 cells compared to wild-type TRA1 (coloured for dark points represent P < 0.05, log 2 fold change >1). c) Significantly enriched GO biological processes were determined for genes with both positive and negative age-genotype interaction with log 2 interaction score >2 (P < 0.05) in tra1 Q3 cells compared to wild type. d) Examples of genes with positive (SIP18, MLS1) and negative (RGL1, BTN2) age-genotype interaction. Normalized RNA sequencing read counts are shown for TRA1 and tra1 Q3 cells at day 0 and day 3.
(Berg et al. 2018). The nature of this feedback mechanism remains unclear. Upregulation of TRA1 is not observed in response to deleting other components of SAGA and NuA4 (Berg et al. 2018). TRA1 expression is also upregulated upon protein misfolding stress (Jiang et al. 2019). Therefore, it is reasonable to postulate that increased proteotoxic stress associated with the aging tra1 Q3 allele plays a role in this phenotype.

HAT complexes and lifespan regulation
While we show here that compromising Tra1 function shortens CLS, the role of the different components of the SAGA and NuA4 in aging is complex. Deleting GCN5 and SPT20 shortens CLS in winemaking yeast (Orozco et al. 2012;Picazo et al. 2015). Conversely, deleting the SAGA component SGF11 extends CLS (Garay et al. 2014) indicating that members of HAT complexes can differentially impact the aging process. The SAGA DUB module also regulates replicative aging via its interaction with Sir2 (McCormick et al. 2014;Mason et al. 2017). Deletion of GCN5 and its pharmacological inhibition extends replicative lifespan (Huang et al. 2020). Therefore, different components of SAGA can differentially affect aging. Arp4 (actin-related protein 4), an essential component of the NuA4 complex, is necessary for chronological aging through its interaction with Hho1 (Miloshev et al. 2019). Thus, different components of SAGA and NuA4 can differentially affect aging.
Further studies will be required to determine the role of Tra1 and the impact of the tra1 Q3 mutation on the global acetylation of both histone and non-histone substrates. Interestingly, tra1 Q3 prevents upregulation of ACS1 during aging (Supplementary Table 2). ACS1 encodes a nucleocytoplasmic acetyl-CoA synthetase whose expression increases during chronological aging (Lesur and Campbell 2004;Wierman et al. 2017), presumably to Fig. 5. Aged tra1 Q3 cells display increased TRA1 mRNA abundance. a) Normalized RNA sequencing read counts are shown for TRA1 mRNA in wild-type TRA1 and tra1 Q3 cells at day 0 and day 3. b) Log 2 fold change (tra1 Q3 /TRA1) for the mRNA of genes encoding SAGA and NuA4 components at day 0 and day 3. maintain the pool of acetyl-CoA required for histone acetylation (Takahashi et al. 2006). ACS1 null cells consequently display reduced CLS (Marek and Korona 2013). This is also in agreement with previous observations that cells deleted for NuA4 and SAGA components (eaf7D and gcn5D) display decreased levels of acetyl-CoA (Rollins et al. 2017).
Our data also showed that tra1 Q3 cells have increased sensitivity to acetic acid (Fig. 3). This is consistent with studies showing the importance of transcriptional regulation in acetic acid tolerance (Dong et al. 2017;Vall-Llaura et al. 2019;Chaves et al. 2021). A direct link to SAGA and NuA4 was suggested by Dong et al. (2017) who revealed that acetic acid treatment leads to differential expression of multiple SAGA/NuA4 components. Similar to what we observed, increased levels of ADA2 and EPL1 increase sensitivity to acetic acid (Dong et al. 2017). Nonetheless, the impact of acid stress in histone acetylation is complex and involves multiple pathways. Vall-Llaura et al. (2019) found that deacetylation by the Sir2 histone deacetylase regulates acetic acid tolerance. In addition, defects in the Set3 histone deacetylase complex are associated with increased tolerance to acid stress, suggesting that histone acetylation plays a protective role (Sousa et al. 2013). Of note, the most upregulated genes in aged tra1 Q3 cells, BTN2 and RGI1, are strongly upregulated by acetic acid (Dong et al. 2017). This is consistent with these cells experiencing high acid stress and potentially provides an explanation as to why buffering the growth media's pH alleviates the aging phenotype associated with the tra1 Q3 mutation.

Tra1, mitochondria, peroxisomes, and the aging process
Here, we showed that several genes important for b-oxidation are decreased in tra1 Q3 cells after chronological aging. During chronological aging, yeast cells utilize internal fat stores (Goldberg et al. 2009). Ultimately, free fatty acids are the substrate for peroxisomal b-oxidation. This process allows cells to produce acetyl-CoA that is used to generate the ATP in the mitochondria required for survival at stationary phase. Thus, cells that lack mitochondrial respiration (i.e. incapable of growing on glycerol) cannot grow on oleate (Lockshon et al. 2007). In our case, the absence of growth of the tra1 Q3 strain on medium containing oleate as the carbon source ( Fig. 7b) could be linked to deficient mitochondrial respiration. In S. cerevisiae, Gcn5 and Ubp8 are also required for respiration (Canzonetta et al. 2016;Leo et al. 2018), a process essential for chronological aging Ocampo et al. 2012).
We previously found that a TRA1 mutant displays negative genetic interactions with genes associated with mitochondrial function (Hoke et al. 2008). This could reflect altered regulation of the retrograde response associated with aging (Kim et al. 2004;Jazwinski 2005;Friis et al. 2014;Pogoda et al. 2021). The mitochondrial retrograde pathway signals to the nucleus via the Rtg proteins (Rtg1, 2, and 3) to upregulate genes associated with mitochondrial stress (Liao and Butow 1993;Zdralevi c et al. 2015). The canonical retrograde target is CIT2, which encodes the peroxisomal isoform of citrate synthase. The retrograde pathway regulates expression of several other peroxisomal proteins (Chelstowska and Butow 1995). Retrograde signaling also upregulates other genes involved in the TCA cycle, such as mitochondrial citrate synthase (CIT1), aconitase (ACO1), and NADþdependent isocitrate dehydrogenase (IDH1/2). Interestingly, we found that these targets of the retrograde pathway are downregulated in aged tra1 Q3 (Supplementary Table 2). While there is debate concerning the incorporation of Rtg2 in the SAGA-like complex (Pray- Grant et al. 2002;Adamus et al. 2021), our data suggest a role for Tra1 in the regulation of retrograde signaling that could contribute to the changes in peroxisomal gene expression observed in the aged tra1 Q3 cells.
Cells carrying deletions in genes encoding other SAGA components display reduced levels of peroxisomal genes Fig. 6. tra1 Q3 cells are sensitive to oxidative stress. a) Log 2 fold change (TRA1/tra1 Q3 ) for mRNA of genes associated with oxidoreductase activity in wild-type TRA1 and tra1 Q3 cells at day 0 and day 3. b) Normalized RNA sequencing read counts are shown for CTT1 and CTA1 in TRA1 and tra1 Q3 cells at day 0 and day 3. c) tra1 Q3 cells are sensitive to diamide. TRA1 and tra1 Q3 cells were spotted onto agar plates without (untreated) or with 1 mM diamide. Fig. 7. tra1 Q3 results in defective peroxisomes. a) Log 2 fold change (TRA1/tra1 Q3 ) for beta-oxidation genes at day 0 and day 3. b) tra1 Q3 cells show decreased growth in the presence of oleic acid. TRA1 and tra1 Q3 cells were spotted on agar plates containing either glucose (YPD), oleate, myristate, or glycerol as the carbon source. Growth of tra1 Q3 cells relative to TRA1 cells was quantified and is shown in the bar graph. c) Normalized RNA sequencing read counts are shown for PEX34 and PEX21 in wild-type TRA1 and tra1 Q3 cells at day 0 and day 3. d) TRA1 and tra1 Q3 cells expressing yemRFP-SKL were imaged using fluorescence microscopy at day 0 and day 3 of the aging process. The number of peroxisomes/cell is shown in bar graphs. Bar: 10 mm. (Ratnakumar and Young 2010) suggesting a further link between HAT complexes and peroxisomal biogenesis/functions. Interestingly, cells incompetent for b-oxidation have a less severe aging phenotype than cells devoid of peroxisomes, indicating that other peroxisomal functions are important in regulating CLS (Lefevre et al. 2013). Peroxisome proliferation is also associated with the early stage of replicative aging (Deb et al. 2022). Free oxidative radicals have long been suggested to regulate the aging process (Harman 1972). Peroxisomes contain catalase that metabolizes hydrogen peroxide and maintains the cellular redox balance (Lismont et al. 2015). Efficient import of catalase into the peroxisomes improves longevity in human cells (Koepke et al. 2007). Inhibiting human peroxisomal catalase triggers increased mitochondrial reactive oxygen species (Walton and Pizzitelli 2012). In contrast, deletion and pharmacological inactivation of either form of the yeast catalase (CTT1, cytosolic; CTA1, peroxisomal), is associated with extended CLS (Mesquita et al. 2010). It was proposed that lack of catalase in young cells triggers a sublethal level of oxidative stress that allows hormetic adaptation to oxidative stress and, consequently, lifespan extension (Mesquita et al. 2010). However, overexpressing catalase also extends the CLS of cells lacking superoxide dismutase (Dsod1) showing that catalase levels are indeed important to alleviate oxidative damage associated with aging (Rona et al. 2015). Therefore, inability of tra1 Q3 cells to properly regulate catalase expression likely has an important role in chronological aging. This is supported by previous studies showing that SAGA plays a major role in the transcriptional response to oxidative stress (Huisinga and Pugh 2004;Sansó et al. 2011;Kim et al. 2019).

Perspectives and conclusions
In conclusion, our work offers insight into how Tra1 regulates gene expression associated with several stress pathways such as regulation of cell wall integrity, mitochondria respiration and peroxisomal function that are crucial for chronological aging. Given the growing evidence that chronologically aged yeast cells form a heterogeneous population of quiescent and nonquiescent cells (Allen et al. 2006;Aragon et al. 2008;Miles et al. 2013;Mohammad et al. 2020), it will be important to define how transcriptional regulators like Tra1 and other SAGA/NuA4 components regulate gene expression, histone acetylation and chromatin accessibility within the aging population.

Data availability
Strains and plasmids are available upon request. Gene expression data are available at GEO with the accession number: GSE206033.
Supplemental material is available at G3 online.