Proteomic analysis of dietary restriction in yeast reveals a role for Hsp26 in replicative lifespan extension

Dietary restriction (DR) has been shown to increase lifespan in organisms ranging from yeast to mammals. This suggests that the underlying mechanisms may be evolutionarily conserved. Indeed, upstream signalling pathways, such as TOR, are strongly linked to DR-induced longevity in various organisms. However, the downstream effector proteins that ultimately mediate lifespan extension are less clear. To shed light on this, we used a proteomic approach on budding yeast. Our reasoning was that analysis of proteome-wide changes in response to DR might enable the identification of proteins that mediate its physiological effects, including replicative lifespan extension. Of over 2500 proteins we identified by liquid chromatography–mass spectrometry, 183 were significantly altered in expression by at least 3-fold in response to DR. Most of these proteins were mitochondrial and/or had clear links to respiration and metabolism. Indeed, direct analysis of oxygen consumption confirmed that mitochondrial respiration was increased several-fold in response to DR. In addition, several key proteins involved in mating, including Ste2 and Ste6, were down-regulated by DR. Consistent with this, shmoo formation in response to α-factor pheromone was reduced by DR, thus confirming the inhibitory effect of DR on yeast mating. Finally, we found that Hsp26, a member of the conserved small heat shock protein (sHSP) family, was up-regulated by DR and that overexpression of Hsp26 extended yeast replicative lifespan. As overexpression of sHSPs in Caenorhabditis elegans and Drosophila has previously been shown to extend lifespan, our data on yeast Hsp26 suggest that sHSPs may be universally conserved effectors of longevity.


Summary
Dietary restriction (DR) has been shown to increase lifespan in organisms ranging from yeast to mammals. This suggests that the underlying mechanisms may be evolutionarily conserved. Indeed, upstream signalling pathways, such as TOR, are strongly linked to DR-induced longevity in various organisms. However, the downstream effector proteins that ultimately mediate lifespan extension are less clear. To shed light on this, we used a proteomic approach on budding yeast. Our reasoning was that analysis of proteome-wide changes in response to DR might enable the identification of proteins that mediate its physiological effects, including replicative lifespan extension. Of over 2500 proteins we identified by liquid chromatography-mass spectrometry, 183 were significantly altered in expression by at least 3-fold in response to DR. Most of these proteins were mitochondrial and/or had clear links to respiration and metabolism. Indeed, direct analysis of oxygen consumption confirmed that mitochondrial respiration was increased several-fold in response to DR. In addition, several key proteins involved in mating, including Ste2 and Ste6, were downregulated by DR. Consistent with this, shmoo formation in response to α-factor pheromone was reduced by DR, thus confirming the inhibitory effect of DR on yeast mating. Finally, we found that Hsp26, a member of the conserved small heat shock protein (sHSP) family, was upregulated by DR and that overexpression of Hsp26 extended yeast replicative lifespan. As overexpression of sHSPs in Caenorhabditis elegans and Drosophila has previously been shown to extend lifespan, our data on yeast Hsp26 suggest that sHSPs may be universally conserved effectors of longevity.

Introduction
Dietary restriction (DR) is the most robust form of environmental manipulation known to increase longevity in a range of organisms [1]. First discovered to extend the lifespan of laboratory rats [2], DR has since been shown to increase the healthy lifespan of many organisms, including yeast, nematodes and flies [3]. DR has been studied extensively in the budding yeast Saccharomyces cerevisiae for both chronological and replicative ageing models [4,5]. The most common method of performing DR in yeast is through the reduction of glucose from the standard 2% concentration to either 0.5% or 0.05%, with the latter resulting in the largest replicative replicative lifespan extension [6]. Such studies in budding yeast have led to the identification of conserved genetic pathways linked to DR's longevity and healthspan effects, including Ras/PKA, TOR/Sch9 and sirtuins [1,5,7]. Indeed, direct [8] or indirect [9] activation of the yeast sirtuin, Sir2, by DR-induced changes in NAD/NADH ratios has been proposed as a longevity-promoting mechanism.
However, this idea is controversial, as it has been shown that DR does not increase Sir2 silencing activity [10,11] and that DR still increases replicative lifespan in sir2 deletion mutants [12]. Hence, although significant progress has been made using various model organisms [1], the key downstream effector proteins that ultimately mediate longevity extension by DR remain unclear.
To begin to address this issue, we set out to determine how DR alters cells at the global proteome level using S. cerevisiae as a model organism. Yeast are well suited for this, as they are unicellular and hence the confounding issues of tissuespecific gene/protein expression inherent with multicellular organisms are avoided. In addition, the small size of the budding yeast genome (~12 Mbp) and proteome (~6000 proteins), coupled with the ease of genetic manipulation facilitates Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210432/922044/bcj-2021-0432.pdf by guest on 22 October 2021 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20210432 6 identification and experimental testing of candidate effectors of longevity. Here we describe our use of liquid chromatography-mass spectrometry (LC-MS) to identify differentially expressed proteins under standard (2% glucose) and DR (0.05% glucose) conditions. This confirmed earlier reports that DR causes a major physiological shift towards increased mitochondrial respiration [13], but also revealed that DR inhibits mating by two distinct mechanisms. Importantly, we identify Hsp26 as a protein that is induced by DR and show here that Hsp26 overexpression increases replicative lifespan.

Chemicals and reagents.
Materials for yeast culture were obtained from Sigma-Aldrich (Poole, UK) and Formedium (Norwich, UK). PCR primers were supplied by Sigma Genosys (Havenhill, UK), genomic DNA isolation kits were from Invitrogen (Paisley, UK); and PCR enzymes/reagents were from Promega (Southampton, UK). The custom-made Hsp26 antiserum was made by inoculating rabbits with a synthetic peptide (CVKKIEVSSQESWGN) corresponding to the C-terminal 14 amino acids of the protein preceded by a cysteine for conjugation, and was supplied by Genosphere Biotechnologies (Paris, France). All other materials were obtained from Sigma-Aldrich.

Yeast strains.
BY4741 (MATa his3Δ1 leu2Δ0 met15 Δ0 ura3Δ0) was used as the background strain [14]. GFP-labelled strains in this background were used to validate mass spectrometry identifications and were obtained from ThermoFisher (Paisley, UK).
The isogenic BY4742 strain (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) was used as a control for shmoo assays. The HSP26 overexpression plasmid was made by PCR cloning the HSP26 open reading frame downstream of the ADH1 promoter in the centromeric p415 vector [15]. BY4741 wild type strains transformed with this plasmid or an empty vector were grown on synthetic dropout media lacking leucine (SD-leu) to select for the LEU2 marker.

Proteomics
The sample preparation and proteomic workflows described below were based on our previously published method [16], with modifications. Four independent biological replicate samples for each condition were used for the analysis.

Yeast culture
Yeast liquid cultures of 10 mL were grown overnight at 30 ˚C in sterile containers.
Cultures were then diluted 1/20 in YPD containing either 2% or 0.05% glucose, in a 1 L conical flask and left to grow at 30 ˚C shaking until an OD₆₀₀ of 0.6 (mid-log) had been reached, which took around 4 hours in 2% glucose and around 6 hours in 0.05% glucoseCultures were spun down at 4000 g for 5 minutes and pellets recovered, washed in sterile deionised water, and frozen at -80 ˚C. Yeast pellets were collected until ~1.5x10⁹ cells had been frozen for each condition.

Sample preparation
Cell samples for a given condition and replicate were recombined into 1 mL bead beating buffer (50 mM ammonium bicarbonate and EDTA-free cOmplete protease inhibitor (Roche)), and centrifuged for 10 minutes at 4000 g at 4 o C. The supernatant was removed and 250 μL bead beating buffer added. Cell lysis was achieved by automated glass bead-beating using a MINILYS® homogenizer (Precellys, UK), applying 15 × 30 sec cycles at 4°C with a 1-min break between each cycle, when lysates were cooled on ice. Lysates were centrifuged for 10 minutes at 13,000 g at fine needle. Flow-through was then collected via centrifugation for 10 minutes at 4000 g at 4 o C and combined with the supernatant. The total volume for the combined fractions was recorded to estimate the number of cells/mL for each condition. A protein assay was then performed to determine the amount of protein within each sample.

Label-free protein quantification
The raw LC-MS files were analysed in Progenesis QI for Proteomics, label-free analysis software which aligns the files and then peak picks for quantification by peptide abundance. The Progenesis QI workflow creates a virtual aggregate run comprising all the data from the individual samples which allows features to be cross identified from other samples, overcoming stochastic sampling limitations of DDA.
The software first aligned the LC-MS files and peak picked the aligned peptides. An aggregate file was generated that contained all the peaks from all runs in the experiment so that there are no missing values. Normalisation was performed using the "normalize against all proteins" option. The software assumed that most proteins are not changing in abundance and normalisation factors are used to adjust peptide intensities. The peptide list was exported into MASCOT and searched against the UniYeastS288c protein database (with carbamidomethyl cysteine as a fixed modification and methionine oxidation as a variable modification, a precursor mass

High resolution respirometry
Yeast cell respiration was determined at 30°C using an Oxygraph-2 k system (Oroboros, Innsbruck, Austria) equipped with two chambers. Yeast cells (2 mL) at a concentration of 3.5 × 10 6 /ml, in YP media were added to each chamber and assays conducted in biological triplicate. The chambers were closed and routine respiration was recorded. LEAK respiration was determined by the addition of 150 μM TET, an ATP synthase inhibitor. Uncoupled respiration was then determined by the addition of the ionophore FCCP (12 μM). The addition of 2 μM Antimycin A accounted for non-mitochondrial oxygen consumption. Data were analysed using DatLab software.

Shmoo assay
Cells were grown in 2% or 0.05% glucose media to an OD 600 of 0.6 and then 400 μl aliquots were centrifuged for 3 min at 3000 g. These were then resuspended in 400 μl of fresh media (2% or 0.05% glucose), spun down and placed in fresh media once more. At this point, 10 μl of 2 mg/ml α-factor or ethanol (vehicle control) was added to the cells, before being placed in an incubator at 30˚C. At set intervals, 20 μl aliquots were taken and placed on to a microscope slide and 100 single cells were checked at random to determine the proportion of cells that had started to form a mating shmoo.

Size-exclusion chromatography
Yeast colonies were grown overnight in 10 mL of YPD broth containing either 2% or

Replicative lifespan analysis.
This was performed essentially as described previously [17]. Briefly, yeast strains were streaked onto appropriate media and individual virgin cells moved to identifiable grid positions on the agar plate using an MSM micromanipulator (Singer Instruments, Somerset, UK). The number of daughter cells produced by each mother cell was then recorded. The plates were incubated at 30°C during working hours, and moved to 4°C overnight. Replicative lifespan was defined as number of daughter cells removed from the mother cell. Statistical analysis of replicative lifespan data was carried out using the online application OASIS 2 [18].  Table 2). The distribution of all identified proteins by expression change and p-value is shown as a volcano plot in Figure 1A . (NMR). All three mitochondrial functional capacity categories were increased several-fold in DR conditions ( Figure 2). Therefore, the DR-induced increase in mitochondrial protein expression evident in our proteomic analysis is mirrored by increased mitochondrial respiratory activity.
After mitochondrial proteins, the next largest group of proteins whose expression was increased by DR was those involved in carbohydrate metabolism ( Figure 1B). To validate our proteomics data, we grew selected GFP-tagged strains under standard and DR conditions and then immunoblotted for GFP ( Figure 3A). For  We also used the GFP-tagging approach to validate proteins that were downregulated by DR. For some proteins, such as Tod6, we failed to detect a GFP signal, but for Ste6-GFP there was a clear DR-induced reduction in the level of a GFP-immunoreactive band of around 27 kDa ( Figure 3A). As this is the molecular mass of free GFP, it likely reflects the previously reported recycling of Ste6-GFP to Intriguingly, other proteins involved in mating were also observed by LC-MS to be downregulated by DR, including the α-factor pheromone receptor, Ste2. When the MATa-cell-specific Ste2 protein binds to α-factor, this causes a characteristic plasma membrane deformation, known as a shmoo, to form in an attempt to fuse with the cell of the opposite mating type that released the α-factor [23,24]. To test if mating capabilities during DR are decreased due to the downregulation of Ste2, BY4741 MATa cells were exposed to purified α-factor and the percentage of cells forming shmoos assessed by microscopy. This revealed a significant decrease in the proportion of cells forming shmoos under DR compared to standard conditions ( Figure 3B), with a greater than 3-fold reduction in shmoos being observed at 2 hours. At later time points, the difference between standard and DR conditions diminished, presumably due to the α-factor being progressively degraded by secreted extracellular Bar1 protease [25]. No shmoo formation in response to αfactor was observed using BY4742 MATα cells under any conditions, demonstrating the specificity for MATa cells ( Figure 3B). Taken together, these data suggest that DR reduces mating by two separate mechanisms: decreased a-factor secretion via downregulation of its transporter, Ste6; and decreased α-factor binding via downregulation of its receptor, Ste2. Although most DR-regulated proteins were linked to respiration and metabolism, two heat shock proteins were strongly induced by DR: Hsp12 and Hsp26. Hsp12 is required for replicative lifespan extension by DR in yeast [26], but is not conserved in metazoans. Hsp26, in contrast, is a member of the evolutionarily conserved, α-crystallin domain-containing classical small heat shock protein (sHSP) family [27]. Intriguingly, sHSPs have been shown to increase longevity in C. elegans and Drosophila [28], suggesting that upregulation of Hsp26 by DR may be causally linked to lifespan extension. To validate the proteomic results, cell lysates were separated first by non-denaturing size-exclusion chromatography and subsequently by denaturing SDS-PAGE. As Hsp26 is known to form 24-subunit oligomers with a native molecular weight of ~600 kDa, while having a small monomeric mass of 26 kDa [27], we reasoned that it should be possible to detect and distinguish the forms of Hsp26 using this approach.
SDS-PAGE gels of lysates from standard and DR conditions exhibited a broadly similar protein expression pattern ( Figure 4A,B). However, one band appeared in 0.05% glucose that was not detected in 2% glucose (arrow in Figure   4B). This band displayed a molecular mass of just under 28 kDa but migrated on size exclusion chromatography at around 600-700 kDa, suggesting that it may be Hsp26. To verify this, western blotting was carried out using a custom-made antibody to the C-terminus of Hsp26 ( Figure 4C,D)

Discussion
In this work, we employed label-free quantitative proteomics to characterise the effects of DR on global protein levels and validated these findings using orthogonal functional assays. These studies have revealed the coordinated changes in protein expression that underlie DR-induced physiological reprogramming. The picture that emerges is that yeast cells respond to DR by making better use of scarce energy sources, reducing mating activity and increasing replicative lifespan ( Figure 6). In each of these cases, it is evident that multiple proteins and distinct mechanisms are the large increase in respiration, given that iron/sulphur clusters are assembled in mitochondria and are required for oxidative phosphorylation [29].
The DR-induced switch from fermentation to respiration we observed here was first documented by the Guarente laboratory, who reported a 2-fold increase in respiration in cells grown in 0.5% glucose compared to standard 2% glucose [13].
The higher 4-fold increase in respiration that we observed likely reflects the stronger effect of 0.05% glucose, which provokes a larger lifespan increase than 0.5% glucose [6]. It is well established that a similar switch from fermentation to respiration also occurs during the diauxic shift, resulting in an upregulation of the gloxylate cycle and gluconeogenesis [30]. Indeed, we found that DR increased the expression of enzymes involved in the glyoxylate cycle (5-fold increases in Mdh1 (malate dehydrogenase) and Icl1 (isocitrate lyase)) and gluconeogenesis (11-fold increase in available to yeast cells in standard YPD media comes from the 2% glucose supplied; reducing this concentration 40-fold to 0.05% in DR therefore greatly limits the availability of caloric substrates. This contrasts with the situation in the classical diauxic shift, where after consumption of available glucose, an abundance of caloric substrates remain in the form of ethanol produced from the earlier fermentation of glucose. Another intervention that has been shown to increase respiration and upregulation of mitochondrial gene expression is inhibition of TOR signaling [32]. Intriguingly, BY4741 tor1 deletion mutants in 2% glucose have the same (extended) replicative lifespan as wild type cells in 0.05% glucose; and 0.05% glucose cannot further extend replicative lifespan in tor1 mutants [33]. This indicates that the same form of DR in the same background strain that we used is TOR-dependent. As inhibition of TOR activity has been strongly implicated in DR-mediated longevity in various organisms [5], this may be of relevance beyond the yeast model.
Although there are no published proteomic data of actively growing cells in 0.05% glucose (mimicking the situation during replicative lifespan), there has been a previous microarray analysis of mRNA expression in cells growing in 0.5% glucose.
That study identified only 124 differentially expressed genes at a threshold of 1.5-fold [13], compared to 782 proteins meeting this cut-off in our proteomic study. This may reflect the different glucose concentrations used, but may also reflect differences between indirect transcriptomic approaches versus direct proteomic analysis.
Nevertheless, some differentially expressed genes/proteins related to metabolism and mitochondrial respiration were common to both datasets, including Fit2, Hxk1, Ald4, and mitochondrial ribosomal subunits [13]. A more recent study of yeast cells grown in a non-dividing state under extreme glucose-limiting conditions found that over 40% of the differentially expressed proteins were involved in mitochondrial and/or respiratory functions [34]. Evidently, increased respiration is a consistent feature of growth in low glucose. However, it is clearly not required for replicative lifespan extension by DR, as the longevity increase in 0.05% glucose is unaffected by mutations that abolish mitochondrial respiration [6].
One of the hallmarks of DR in metazoans, ranging from worms through flies to rodents, is reduced reproduction [35]. This is widely thought to represent an evolutionary strategy to prevent limited resources being invested in offspring with little chance of survival in a nutrient-poor environment. However, in the yeast replicative lifespan model studied here, individual mother cells actually produce more offspring under DR than under standard conditions and so would appear to be an exception to this general rule. However, we show here that DR actively inhibits mating of MATa cells with cells of the opposite mating type (MATα) by two distinct mechanisms. The first, downregulating expression of the α-factor pheromone receptor, Ste2, has previously been shown to occur in cells grown in low glucose [36,37]. However, the second mechanism, reducing secretion of the a-factor pheromone via downregulation of its transporter, Ste6, has not previously been reported. DR therefore inhibits mating both by releasing less pheromone to signal mating competence to other cells and by simultaneously decreasing the ability to respond to pheromone released by cells of the opposite mating type. Hence, the universal effect of DR on reproduction can indeed be applied to yeast, provided that this is specific to sexual reproduction via mating, as opposed to asexual reproduction via mitosis. It is tempting to speculate that this may reflect a more selfish universal underlying A major finding of our work is that DR upregulates Hsp26 expression and that overexpressing Hsp26 in standard conditions increases replicative lifespan.
Intriguingly, Hsp26 is one of only five validated long-lived asymmetrically retained proteins (LARPs), which accumulate in yeast mother cells during replicative ageing [38]; and Hsp26 has also been shown to increase in expression during chronological yeast ageing [39]. Furthermore, recent proteomic analyses showed that Hsp26 is one of the most highly upregulated proteins in response to sub-lethal heat shock [16], an environmental manipulation known to extend yeast replicative lifespan [40].
Hsp26 is widely accepted to function as a "holdase" molecular chaperone, which initially acts to prevent aggregation of client proteins and subsequently assists in the protein refolding stage [41]. However, its broad substrate specificity means that the key physiological client proteins remain unclear; and GFP-tagging studies show Hsp26 to be have a general cytoplasmic distribution, with no obvious localization to subcellular compartments [21]. Nevertheless, given that most proteins upregulated by DR are mitochondrial, it is possible that a proportion of Hsp26 may (transiently) localize to mitochondria in order to help maintain mitochondrial protein folding.
Although we demonstrated that replicative lifespan can be extended simply by raising Hsp26 levels, DR was nevertheless still able to increase longevity in hsp26 deletion mutants. This may be due to compensatory mechanisms that can substitute for Hsp26 function in its absence. Hsp26 is a member of the conserved sHSP family, which are ATP-independent "holdase" chaperones that help maintain cellular proteostasis by controlling protein misfolding and aggregation [42]. Hence, it may be

Competing interests statement
The authors declare that they have no competing interests.

Data availability statement
The mass spectrometry proteomics data have been deposited to the   BY4741 cells were grown to mid-log phase in either 2% or 0.05% glucose, and injected into sealed chambers of a high-resolution Oxygraph-2k respirometer.
Respiration was then measured through oxygen consumption in these chambers.