The mammalian longevity associated acetylome

The significant increase in human longevity and the growing proportion of elderly in society over the past century are accompanied by an exponential rise in age-related diseases¹. This led to an urgent need for new approaches for extending human healthspan. However, to this end, a deeper understanding of the underlying mechanisms of healthy ageing is required. Armed with such knowledge, it will be possible to develop interventions that will help alleviate the negative impact of ageing and convert the elderly population from dependent to contributing individuals. One intriguing option to explore how healthy ageing may be achieved is by analyzing nature’s largest ongoing biological experiment, namely, evolution, and in particular, the development of long-lived animals.

To date, few studies have sought to understand the mechanisms underlying the increased lifespans of long-lived organisms. These studies suggested that such species develop and enhance specific longevity-favoring characteristics such as body size, brain development, sociality, increased DNA repair, and protection against tumor formation². Specifically, a number of proteins/pathways shown to control lifespan in short-lived model organisms may have life-extending functions in long-lived ones. For example, the expression of the well-known tumor suppressor protein p53, which was shown to control the lifespan of short-lived organisms³, is significantly increased in elephants. This finding was suggested to explain the low incidence of cancer and increased survival in these long-lived mammals^4,5.

Various attempts were also performed at the -omics level to identify the regulation of lifespan in long-lived animals. Meta-analysis of age-related gene expression profiles in mice, rats, and humans identified common signatures of ageing, including inflammation, immune response, and energy metabolism-related pathways⁶. Wider meta RNA-seq analyses across databases of 26 or 41 mammalian species also identified transcriptional signatures of known longevity-related pathways^7,8. In addition to the above-mentioned pathways, these analyses identified DNA repair, IGF1 expression, and mitochondrial translation. Interestingly, the expression of these genes was modulated by pro-longevity interventions, including mTOR inhibition and caloric restriction (CR)^9,10. In contrast to transcriptome analyses, most attempts to identify longevity-associated proteomic signatures were performed in humans. For example, a recent study identified 754 human plasma proteins that are associated with chronological age and are mostly related to inflammatory response, organismal injury, cell and organismal survival, and cell death pathways¹¹. In addition, Coenen et al. identified 273 plasma proteins significantly associated with ageing¹². Pathway and network analyses of these proteins highlighted several pathways, including those that were shown to regulate longevity, such as IGF signaling, sulfur binding, TNFα (inflammation), and metabolic diseases. Likewise, among the pathways of ageing-related proteins identified by Wyss-Coray and his colleagues were sulfur binding and IGF1-related hormonal signaling¹³. Besides transcriptome and proteome analyses, several studies attempted to characterize the human age-related metabolome and found several known pathways that were shown to regulate longevity, such as kynurenine NAD⁺ biosynthesis, one carbon/transsulfuration, inflammation, oxidative stress, and lipid metabolism pathways^8,14.

Reversible post-translational modifications (PTMs) refer to the post-translational addition of any of a plethora of modifying molecules, such as acetyl, phosphoryl, and methyl groups, to one or more specific amino acid residues of a target protein (PTM sites)¹⁵.

Surprisingly, despite the extensive studies on various -omics datasets, to the best of our knowledge, no in-depth study has been performed to explore the role of reversible protein PTMs in longevity regulation. Yet, some PTMs were found to be associated with age-related diseases and to accumulate with age¹⁶. The attachment of specific moieties to a protein can affect a wide range of its characteristics, including stability, enzymatic activity, localization, and interactions¹⁷. Thus, PTMs regulate various biological processes, such as signal transduction, gene expression, DNA repair, and metabolism. As a result, in addition to the fundamental cellular roles, PTMs are a key factor in regulating systemic processes, including adaptation to environmental changes encountered by the organism over a lifetime. The plasticity of such reversible PTMs can maintain homeostasis to preserve a healthy lifespan and thus provides a specific rationale for their regulation of longevity. Moreover, it can be hypothesized that over the course of evolution, specific PTM sites evolved or mutated to support the lifespan extension occurring in long-lived organisms. Accordingly, exploring the regulation of lifespan by PTMs could offer major new insight into the ageing process.

Protein acetylation on lysine ε-amino group is one of the most extensively studied PTMs in eukaryotes. Lysine acetyltransferases/deacetylases (KATs/KDACs) control the acetylation status of thousands of proteins¹⁸. Still, although acetylation has been connected to the regulation of lifespan¹⁹, it remains largely unknown which of these proteins/pathways directly regulate ageing. Significant results^{19,20,21,22,23} have demonstrated that KAT/KDAC activities directly control yeast, nematode, fly, and mammalian²⁴ lifespan. For example, a transgenic (TG) mouse model overexpressing the SIRT6 deacetylase lives longer and with improved health compared to wild-type (WT) littermates²⁵. Yet, acetylation sites that control a healthy lifespan remain to be specifically explored. Traditionally, a point mutation of acetylated lysine (K) to glutamine (Q) is used in molecular biology to mimic constant acetylation, while the exchange from K to arginine (R) is used to mimic the fixation of the non-acetylated state of the protein²⁶. Our hypothesis is that during the evolution of long-lived organisms, a limited set of acetylation sites in short-lived organisms were specifically replaced by amino acids that mimic the fixed state of such PTMs, potentially resulting in lifespan extension. Likewise, the acquisition of new acetylation sites in long-lived organisms could also support the longer lifespans of such animals.

Here, to explore the relationship between acetylation and lifespan regulation in mammals, we created a computational tool, which we named Post-translational modificAtions Regulator Of Healthspan (PHARAOH). PHARAOH identifies sites significantly associated with lifespan extension by comparing the conservation of acetylation sites in a set of 107 mammalian proteomes and their correlation to longevity. Pathway analyses of these sites identified various ageing-related pathways, such as fatty acid metabolism, PPAR signaling, TCA cycle, translation, one-carbon cycle /transsulfuration pathway (TSP), and DNA repair. Validation of specific acetylation sites on cystathionine beta synthase (CBS) from the TSP and Ubiquitin carboxyl-terminal hydrolase 10 (USP10) of the DNA repair pathway addresses the mechanism underlying their positive effect on longevity.

Identifying longevity-associated acetylation sites

To identify PTMs that potentially regulate the extension of lifespan in long-lived organisms, we employed the PHARAOH computational tool. PHARAOH compares the conservation of PTM sites and whether each site was replaced by another amino acid in a set of all identified mammalian orthologous proteins. Next, PHARAOH uses sequence and lifespan data to examine whether any PTM or specific amino acid (AA) replacement is associated with a longer lifespan. Thus, we utilized PHARAOH to search for specific lysine acetylation sites that regulate longevity. Global mouse and human acetylomes (Supplementary Data 1 and 2) were created based on identified acetylation data found in the PhosphoSite²⁷ database together with a mouse acetylome generated by our lab, and consisted of 25,959 and 22,849 mouse and human acetylation sites, respectively (Fig. 1a left panel). For this study, we used both high and low throughput PhosphoSite analyses. To identify the acetylation sites that are significantly associated with longevity, three additional datasets were created. An orthologous protein dataset and phylogenetic tree were created using the OrthoFinder tool²⁸ based on 107 mammalian proteomes from Uniprot²⁹ (Fig. 1a, middle panel). In addition, a maximum lifespan dataset for these animals was generated, based on The Animal Ageing and Longevity Database (AnAge)³⁰ (Fig. 1a, right panel). Specifically, during evolution, an acetylated/deacetylated lysine (K) can be converted into arginine (K-to-R) or glutamine (K-to-Q), mimicking a permanently deacetylated or acetylated lysine, respectively²⁶ (Supplementary Fig. 1a). Interestingly, acetylated lysines tend to be more conserved than unacetylated lysines (χ², P = 0.08, Supplementary Fig. 1b) in comparison between mouse and human. Thus, the PHARAOH tool calculates the significance of the correlation between conservation of a given acetylated site versus conversion to R or Q and maximal lifespan (Fig. 1b).

PHARAOH was designed to assess and identify correlations between amino acid (AA) changes and a longer lifespan. a The tool uses three types of data sets, mouse/human acetylomes (left box), a collection of mammalian orthologue proteins, and the phylogenetic tree of each species based on its orthologs set (middle box), and mammalian maximal lifespans (right box). b The tool analyses changes in amino acids at each acetylation site using pairwise sequence alignment between the acetylated peptide and the orthologous protein of each mammal (upper panel), generating a replacement matrix (middle panel). Subsequently, statistical analysis and validation against the phylogenetic tree are employed to examine the correlation between amino acid replacements and a longer lifespan (lower panel). c An example of a significant acetylation site on Bphl K188 from K-to-R analysis – the central portion of the figure showcases the OrthoFinder phylogenetic tree. Surrounding the tree, the inner circle represents the lifespan of each mammal, while the outer circle displays the amino acid found at that site in each mammal. Lysine (K) in pink arginine (R) in purple, other AA in turquoise, and in green sites with no ortholog found. Color coding is shown on the right. Created in BioRender. Cohen (2025) https://BioRender.com/y84l114.

Full size image

A replacement matrix was built based on pairwise sequence alignment between each acetylated peptide and the orthologous mammalian protein. Using the replacement matrix, maximal lifespan data, and the phylogenetic tree, PHARAOH calculates the statistical significance of the correlation between an acetylation/replacement site and longevity (Fig. 1b and Supplementary Fig. 2, see “Methods” section for detailed description). The OrthoFinder phylogenetic tree of 107 mammals was validated with the recently published Zoonomia project³¹ phylogenetic tree of 62 overlapping mammals (Supplementary Fig. 3a). Importantly, using a correction based on the phylogenetic tree enabled us to eliminate the influence of evolutionary distances between different mammals. This ensures that any bias introduced by other evolutionary factors is neutralized. A representative output of the analysis is shown in Fig. 1c.

R/Q conversion of acetylation sites in longevity

Comparison of the acetylation sites contained in the mouse and human acetylomes with the orthologues’ dataset revealed 321 lysine to arginine (K-to-R) and 161 lysine to glutamine (K-to-Q) substitutions that were significantly associated with longer lifespan (Mann–Whitney FDR < 0.1; phylogenetic tree statistical correction FDR < 0.05) (Supplementary Fig. 3b, c). Importantly, the same analysis with lysine to leucine (K-to-L), as a random control, identified only 8 longevity-associated substitutions (Supplementary Data 3). These sites are only 3% of the total K-to-L replacements between mouse and human, demonstrating the insignificance of this process in comparison to K-to-R/Q, which consists of ~40% of such substitutes. In addition, in comparison to the global acetylome, K-to-R/Q replacements did not show significant enrichment in any specific protein domains or non-domain regions across the 9251 known domains in the mouse proteome (based on UniProt, Supplementary Data 4). Similar to the global acetylome, the largest group was mapped to disordered regions, suggesting that acetylation may play a role in stabilizing undefined structures, as previously reported^26,32. Interestingly, as seen in Supplementary Fig. 4, with the increase in lifespan during evolution, the conversion from K to either R or Q happened gradually rather than after a specific lifespan threshold. Pathway analyses and protein-protein interaction (PPI) of the K-to-R sites assigned these changes to proteins associated with protein translation and folding, as well as many metabolic pathways, including those of fatty acid metabolism, PPAR signaling, the TCA cycle, amino acid biosynthesis, and the one-carbon/TSP cycle. (Fig. 2a, b). Pathway analyses of the K-to-Q sites identified cytochrome p450, peroxisome, mitochondrial translation, taurine metabolism, fatty acid β-oxidation, and others (Fig. 2c, d). Importantly, pathway analyses of the human orthologs of these mouse proteins revealed a similar set of pathways, demonstrating that their role in longevity is potentially conserved (Supplementary Fig. 5a, b).

a Pathway enrichment analysis of K-to-R replacement sites (hypergeometric test and Benjamini–Hochberg p-value correction by Metascape), and b their corresponding MCODE clusters identified from PPI networks. c, d Similar analysis for K-to-Q replacements. e CC analysis of K-to-R (green) and K-to-Q (blue) replacements. pLOGO of 7 amino acids flanking the mouse total acetylome (f), the longevity-associated acetylation sites replaced to R (g), or Q (h) sites in the mammalian ortholog set. K-to-R = lysine to arginine replacement, K-to-Q = lysine to glutamine replacement. For (f–h), significantly enriched AA are seen between the red lines.

Full size image

While maximal and average lifespan are correlated, their regulatory pathways may differ. We analyzed K-to-R/Q replacements using an average lifespan dataset of 93 mammals, based on Animal Diversity³³, Max Planck Longevity Records³⁴, and other datasets (Supplementary Data 5). Pathways unique to the average lifespan included beta-oxidation of long fatty acids, NAD/NADH metabolism, and glycolysis/gluconeogenesis (Supplementary Data 5 and 6). For maximal lifespan, the Alzheimer’s disease pathway was prominent. Despite these differences, over 80% of the pathways were common, suggesting that analyses based on maximal lifespan provide insights into both median and maximal lifespan. Interestingly, major members of these pathways were previously found to be associated with longevity. For example, the rate-limiting enzyme of the TSP, CBS, is a main producer of hydrogen sulfide (H₂S), which was found to mediate the effects of a CR diet on longevity³⁵. Cellular component (CC) analyses, based on Gene Ontology (GO) annotations, of the identified sites revealed that these proteins are localized mainly to the cytosol and the mitochondria (Fig. 2e). No K-to-R sites were found in the nuclear proteins, nor K-to-Q sites in proteins of the plasma membrane and the ER.

In order to identify the specific acetylation consensus sequence, pLogo analysis was performed on the whole mouse acetylome, and no consensus sequence was found, as previously published³⁶. Yet, an enrichment for lysine residues flanking the acetylated K was found, probably due to other acetylation sites nearby (Fig. 2f). Importantly, these positive K’s and R were eliminated on positions −1,−2 and were replaced with D/E on these positions. F/Y were enriched on position +1. These findings might have implications for the enzymatic activity of KATs/KDACs. pLogo analyses of main CC as cytosol, mitochondria, and nucleus showed that D/E at positions −1,−2 are mostly from mitochondrial and cytosolic acetylated proteins. Whereas the G/A at position −1 originated from nuclear proteins (Supplementary Fig. 6a–c). However, pLogo analysis on K-to-R sites identified overrepresented D/EAV at positions −3, −2, −1 and an enrichment for K downstream to the acetylation site. Additionally, R/YK/S were underrepresented at positions −2, −1, and T on position +1 (Fig. 2g). Interestingly, tyrosine, serine, and threonine (Y, S, and T, respectively) can potentially be phosphorylated, suggesting that additional PTMs flanking the acetylation sites may also be associated with lifespan. For K-to-Q sites, pLogo analysis identified hydrophobic amino acids VVL at positions −3, −2, −1 and DGK at positions +1+2+3 (Fig. 2h). This finding suggests that specific KATs/KDACs may regulate the acetylation status of longevity-associated acetylation sites.

Since the effect of these conversions is expressed in long-lived animals, we searched for transcription factors that are known to regulate the enriched pathways in humans. Such factors can provide an additional regulatory layer on ageing. SP1, PPARG, PPARA, SREBF1, ATF2, RelA, VDR and NFKB1 regulators were found within K-to-R longevity-associated sites (Supplementary Fig. 7a). Importantly, these regulators are significantly associated with ageing mechanisms such as inflammation (RelA of NFKB1), DNA repair (ATF2 and SP1), and metabolism (PPARG, PPARA and VDR). The same analysis identified SP1 and NFE2L2 /NRF2 transcription factors in the K-to-Q set (Supplementary Fig. 7b). Similarly, NFE2L2 /NRF2 is involved in ageing-related pathways, such as protection against oxidative stress, and likely mediates CR protection against carcinogenesis³⁷. In addition, gene-disease association (GO_DisGeNET) analysis was performed in order to examine which diseases are associated with longevity-related K-to-R or K-to-Q sites in humans. Interestingly, the majority of the most highly significant results were of metabolic diseases, particularly liver-related, and other diseases such as diabetes and other metabolic-related pathologies such as myopathy and lethargy, all known to be ageing-related (Supplementary Fig. 7c, d).

CBS K386R enhances its pro-longevity activity

Next, we further explored the role of a longevity-associated acetylation site on CBS, a mediator of the CR response. CBS K386 was acetylated in the mouse reference dataset (both in our lab and PhosphoSite data) and exchanged with R in long-lived mammals. The average lifespan of mammals with K-to-R conversion, such as humans, was significantly higher than that of animals with conserved K at this site (Mann–Whitney FDR < 0.1, q-value = 0.00014; tree statistical correction FDR < 0.001) (Fig. 3a, b and Supplementary Fig. 8). Thus, we next followed the effect of K386R replacement on CBS H₂S production activity. Human embryonic kidney (HEK) 293T cells overexpressing either WT, K386Q, or K386R mouse CBS were examined for their H₂S production capacity using the lead acetate method as previously described³⁸. As seen in Fig. 3c, in comparison to cells overexpressing the WT protein, overexpression of K386R CBS resulted in significantly higher H₂S production capacity (p < 0.05). Cells expressing K386Q CBS had a similar H₂S production capacity as cells expressing the WT protein. In the reverse experiment, the H₂S production activity of immunoprecipitated WT, R389K, or R389Q human CBS was tested. While activity of the R389K mutant was similar to the activity of the WT protein, the acetylation-mimicking mutant R389Q exhibited a significantly reduced H₂S production capacity (p < 0.001, Fig. 3d). The non-significant trend of lower H₂S production of R389K compared to the WT CBS is most likely due to low percentages of acetylation on the mutated lysine. All together, these findings show that constitutive deacetylation of the CBS K386 residue, as found in long-lived animals, promotes H₂S production and potentially contributes to their longer lifespan.

a Maximal lifespan of animals with conserved K vs. animals with K-to-R conversion at cystathionine beta synthase (CBS) lysine 386 (K386) (FDR < 0.1). Dot color represents the maximal lifespan of a given species. Schematic representation of the various maximal lifespans of species with K vs. R is shown on the left. Species are ordered by lifespan along the Y-axis. Elephant’s silhouette was contributed by Leonardo255. Created in BioRender. Cohen (2025) https://BioRender.com/l52n801. b Sequence alignment of CBS K386 in mouse and R389 in human using Clustal Omega. c Hydrogen sulfide (H₂S) production capacity in HEK293T cells lysates overexpressing flag-tagged mouse wild-type (WT), K386R, K386Q CBS or GFP control, using lead acetate strips. Tubulin was used as a loading control. d H₂S production capacity of flag-tagged human WT, R389K or R389Q CBS immunoprecipitated from HEK293T cells. ImageJ quantification of (c and d) is shown on the right. Number of biological replicates: n = 3 (c) and 4 (d) independent experiments. H₂S production was corrected to CBS expression levels. For (c) and (d), one-way ANOVA with Bonferroni post hoc was used. For all graphs, bars represent mean ± SEM. *p < 0.05, ***p < 0.001. Source data are provided as a Source Data file.

Full size image

The long-lived acquired acetylome

During evolution, the specific generation of new acetylation sites would be a complementary set of events to the above-mentioned replacements. Specifically, this would involve the replacement of R and Q residues present in short-lived animals with acetylated K residues in long-lived mammals (R-to-K and Q-to-K, respectively). Therefore, using PHARAOH, we examined the conservation of human acetylated K sites among all mammalian orthologs. Particularly, the data was searched for R-to-K and Q-to-K sites within the 22,849 human acetylation sites. We identified 495 R-to-K and 150 Q-to-K sites that were significantly associated with longer lifespan (Mann–Whitney FDR < 0.1, tree statistical correction FDR < 0.05). In comparison to the global human acetylome, R-to-K replacements did not show significant enrichment in any specific protein domains or non-domain regions across the 9767 known domains in the human proteome (based on UniProt, Supplementary Data 4). However, a comparison of the total number of acetylation sites mapped to domains between the global human acetylome and R-to-K replacements revealed a significant association, which was not observed for Q-to-K replacements. Similar to mouse data, the largest group was mapped to disordered regions. Enrichment analyses of the R-to-K sites assigned these changes to various pathways, including chromosome organization, amino acid metabolism, and cell cycle (Fig. 4a). Strikingly, PPI analysis identified many ageing-related pathways in this dataset, including cellular respiration, ribosomal biogenesis, regulation of protein translation, response to stress, and mismatch DNA repair and diseases of DNA repair pathways (Fig. 4b and Supplementary Fig. 9a). Analyses of Q-to-K events also identified many ageing-related pathways, such as response to stress, the TCA cycle, sulfur compound metabolic process (including proteins from the TSP related pathway, one-carbon pool by folate cycle), and response to starvation (Fig. 4c). PPI analysis also identified ribosomal biogenesis and DNA repair pathways (Supplementary Fig. 9b). Interestingly, CC analyses of all identified R/Q-to-K sites revealed that these proteins localized to most cell components. However, no R-to-K sites were found in proteins expressed in the endoplasmic reticulum (ER), and no Q-to-K sites were found in the extracellular space or cytoplasmic proteins (Fig. 4d).

a Pathway enrichment analysis of R-to-K replacement sites (hypergeometric test and Benjamini–Hochberg p-value correction by Metascape) and b their DNA repair corresponding MCODE clusters identified from PPI networks. c Pathway enrichment analysis of Q-to-K replacement sites using Metascape. d CC analysis of R-to-K (green) and Q-to-K (blue), based on Gene Ontology (GO) annotations. e Pathway enrichment analysis of R-to-K DNA metabolic processes and f their corresponding DNA repair MCODE clusters identified from PPI networks.

Full size image

pLogo analysis of the global human acetylome revealed an enrichment for lysine and an under-representation of C in the flanking sequence. As suggested above for the mouse acetylome, these lysines are the adjacent acetylated lysines (Supplementary Fig. 9c). Interestingly, there was an overrepresentation for G/A/D in position −1, suggesting that in humans, there is a G/A/D _(-1)K enrichment. Further pLogo analyses of CC showed that G/A/D in position −1 stemmed from the mitochondrial and cytosolic acetylated proteins, whereas nuclear protein showed only G at −1 (Supplementary Fig. 9d–f). Likewise, a pLogo analysis was performed using the R-to-K sites (Supplementary Fig. 9g) and showed an enrichment for K downstream of the acetylated site, E on position −7 and SS/AK on positions −1 and −2. For Q-to-K sites, the pLogo analysis did not identify a specific logo besides a tendency for F on position −2 and a higher probability for A between −7 and +1 (Supplementary Fig. 9h). Further search for regulators of the enriched pathways in identified R-to-K proteins showed that the vast majority of regulators have a negative/positive role in tumorigenesis, such as STAT1/3, P53, BRCA1, PARP1, MYCN, and E2F1/4 (Supplementary Fig. 9i). Likewise, out of the seven identified regulators of Q-to-K proteins, TP53, STAT1, BRCA1, and SPI1 have a direct enhancing/preventing role in tumorigenesis, whereas manipulation of the others was also suggested to affect cancer (Supplementary Fig. 9j). In addition, GO_DisGeNET analyses of R-to-K and Q-to-K revealed several human ageing-associated diseases, including myocardial ischemia, inflammatory disorders, multiple types of cancer, ataxia telangiectasia, and Werner syndrome (Supplementary Data 7 and 8).

Remarkably, as seen in Supplementary Fig. 10, the initial conversions from R/Q-to-K occurred early in evolution and more sites accumulated as it progressed. Interestingly, as seen in Fig. 4e, f, pathway enrichment analyses of the long-lived associated acetylome also highlighted various DNA repair pathways. These include non-homologous end joining (NHEJ), homology-directed repair of DNA double-strand breaks, and DNA mismatch repair. Importantly, substantial evidence suggests DNA repair as a key mechanism of longevity, mostly via protection against the two major threats to long survival: cancer and cognitive decline. Thus, acquiring acetylation on DNA repair proteins during evolution can support healthy longevity. This suggests that accumulating new acetylation sites helps address the gradual increase in body size associated with longevity.

USP10 K714 Acetylation controls PCNA stability

Next, we aimed to further elucidate the effect of acquired acetylation sites in longevity. To this end, the effect of K714 acetylation of a DNA repair pathway protein (Fig. 4b) ubiquitin-specific peptidase 10 (USP10) (Mann–Whitney FDR < 0.05; tree statistical correction FDR < 0.05) was examined (Fig. 5a). In short-lived mammals, there is a R instead of the K714 residue of human USP10. USP10 catalyzes a hydrolase activity, which removes conjugated ubiquitin from its target proteins, such as Proliferating Cell Nuclear Antigen (PCNA), thereby stabilizing them³⁹. Increased PCNA levels are associated with poor prognosis of various tumors⁴⁰. Indeed, a correlation test showed a significant positive correlation between PCNA and USP10 levels in lung, glioblastoma, and breast cancers (P = 0.001, 0.000001, and 0.002, respectively) (Fig. 5b), supporting the function of USP10 in stabilizing PCNA. Since long-lived animals tend to be physically larger, one of the major challenges facing such animals is the higher probability of cancer development with age^41,42. Thus, the acquisition of USP10 acetylation in long-lived animals might contribute to addressing this challenge. PCNA protein levels were examined in human colorectal carcinoma (HCT116) cells overexpressing either WT, K714Q or K714R human USP10. The translation inhibitor cycloheximide (CHX) was used to specifically examine the effect on protein stabilization. As seen in Fig. 5c, in comparison to cells overexpressing the WT protein or the mutant K714R, overexpression of K714Q resulted in significantly higher PCNA protein levels with or without CHX treatment. Besides PCNA, USP10 has a spectrum of targets that might affect tumorigenesis as well⁴³. Thus, we followed the role of acetylated USP10 on its global deubiquitylation activity. As seen in Supplementary Fig. 11, in comparison to WT or K714Q, the expression of constitutively deacetylated USP10 mutant K714R results in significantly lower global ubiquitylation levels. Therefore, acetylation/deacetylation controls USP10 activity, and it would be of interest to identify additional USP10 targets that affect longevity. This result shows a potential role of the acquired acetylation on USP10 in reducing cancer incidence, and hence its role in promoting longevity. Altogether, the fixation of the acetylation status or the acquisition of new acetylation sites during evolution enables long-lived animals to enhance pro-longevity mechanisms. Specifically, our study demonstrated that acetylations play a crucial role in promoting H₂S production and DNA repair, two key factors that play a significant role in enhancing a healthy and extended lifespan.

a Significant acetylation site on Ubiquitin-Specific Peptidase 10 (USP10) lysine 714 (K714) from R-to-K analysis. The central portion of the left panel shows the OrthoFinder phylogenetic tree. Surrounding the tree, the inner circle represents the lifespan of each mammal, while the outer circle displays the amino acid found at the site in each mammal (K in pink, R in blue, and other AA in green). Color code is shown on the right (left panel). Maximal lifespan of animals with conserved K vs. animals with R/Q/other AA at USP10 K714 (FDR < 0.1). Dot color represents the maximal lifespan of a given species. b Pairwise Pearson correlation between USP10 and PCNA protein expression levels in lung (left), glioblastoma (central), and breast (right) cancers. Each dot represents a single patient, color code indicating tumor stage is shown on the right. c Biological replicates of PCNA levels in HCT116 cells overexpressing GFP, WT, K714R, and K714Q USP10 with or without cycloheximide (CHX) treatment for 8 h. Each condition was tested using biological triplicates. n = 3 independent experiments. Tubulin was used as a loading control. ImageJ quantification is shown below. One-way ANOVA with Bonferroni post hoc was used. For all graphs, bars represent mean ± SEM. **p < 0.01, ***p < 0.001, ****p < 0.0001. Source data are provided as a Source Data file.

Full size image

Tens of thousands of reversible acetylation sites allow the organism to respond to changes in internal and external conditions and thereby prolong survival of long-lived animals. To identify those acetylation sites that promote longevity, we developed the PHARAOH computational tool. The conversion of 482 acetylation sites found in short-lived mammals to either R or Q residues in long-lived mammals was significantly associated with extended lifespan. These proteins, although localized to almost all CC, are significantly enriched in specific cellular pathways, particularly ageing-related pathways. These include translation, fatty acid metabolism, PPAR signaling, TCA cycle, amino acid biosynthesis, and the one-carbon/TSP, mitochondrial translation, taurine metabolism, and others. Further activity validation of a specific longevity-associated acetylation site, K386 of CBS, showed that the conversion to a permanently deacetylated state in long-lived mammals increases the production of the pro-longevity gasotransmitter, H₂S. In addition, 695 acetylated lysines were significantly acquired in long-lived mammals. These sites are also distributed to all CC; however, a larger fraction was found in the nucleus and extracellular compartments. The acquired acetylated lysine sites were also enriched in various pathways, including ageing-related pathways such as translation, TCA cycle, and amino acid biosynthesis. Likewise, regulation analyses of the significantly changed acetylated proteins identified transcription factors associated with many ageing-related pathways such as oxidative stress, inflammation, and DNA repair. Uniquely, many identified acetylated proteins are enriched in cell cycle and DNA repair pathways. Further activity analysis of USP10, a member of these pathways, showed that the acquisition of acetylated sites in long-lived mammals restrains the oncogenic activity of this protein. Altogether, these findings identified the longevity-associated acetylome and provide two mechanisms, via H₂S production and anti-cancer, for its positive impact on lifespan.

Pathway analyses of longevity-associated fixed or new acetylation sites (K-to-Q/R or Q/R-to-K, respectively) identified numerous ageing-associated pathways. These include pathways that were also identified by previous transcriptomic and metabolomic analyses, such as one carbon/TSP, translation, mitochondria-related pathways, energy, and fatty acid metabolism. In addition to demonstrating the strength of the PHARAOH tool, these results also emphasize the importance of various pathways identified by PHARAOH in regulating longevity. Moreover, the identified acetylation sites revealed a new layer of regulation of these fundamental longevity pathways.

The enzymatic activities of two members of the one carbon/TSP pathway, cystathionine gamma-lyase (CGL) and cystathionine beta synthase (CBS), produce the gasotransmitter H₂S. Enhanced H₂S production was found to mediate the positive effect of CR³⁹. PHARAOH analysis found a significant correlation between CBS K386R conversion and longevity. Accordingly, CBS K386R adaptation significantly increased H₂S production (Fig. 3). This suggests that the fixation of deacetylated CBS might mediate the increased lifespan of long-lived mammals. Indeed, it would be of great interest to follow the effect of such substitution, as others found by PHARAOH, on mice lifespan in future research. However, continuous exposure to excessive levels of H₂S is highly toxic⁴⁴. Therefore, it is noteworthy that the observed increase in H₂S production did not reach the toxic concentration range, and it is likely that as yet uncharacterized mechanisms maintain H₂S levels within the beneficial range.

Interestingly, whereas fixed or new acetylation sites share enriched pathways, we also identified unique pathways associated with acetylation sites that were gained in long-lived animals. Specifically, these unique pathways include DNA repair and the cell cycle. Longevity is positively associated with increased body size. As a result, long-lived animals have 3–6 orders of magnitude more cells than short-lived mammals. Given their lifespan, which is up to 30 times longer, one would expect that long-lived animals would show much higher cancer incidence. Yet, as established by Peto’s Paradox, no statistically significant relationship is found between body size and cancer incidence⁴⁵, indicating that long-lived animals have developed mechanisms to avoid cancer. Indeed, previous studies showed that elephants have lower cancer incidence due to 20 copies of the p53 tumor suppressor gene^4,5. Hence, we suggest that one mechanism underlying the longer lifespan of large animals is via the conversion of R or Q in DNA repair and cell cycle proteins to acetylated K, to protect against cancer. In addition, in the acquisition of a new acetylation site on DNA repair and cell cycle proteins, the reversible acetylation could allow the cell to control the decision between repair and continued division, senescence, or apoptosis. In support of this, our findings of a significant positive correlation between USP10 and PNCA levels (Fig. 5b), along with an acquired acetylation site on the human USP10 (K714), which decreases PCNA stabilization by USP10, demonstrate the need for acetylated K with reduced activity in humans.

The accumulation of longevity-associated acetylation sites (R/Q-to-K, Supplementary Fig. 10) occurred more rapidly in evolution in comparison to K-to-Q/R along with the increase in lifespan (Supplementary Fig. 4). Cancer-related pathways such as DNA repair and cell cycle were enriched in R/Q-to-K proteins. This suggests that relatively early during evolution, mammals had to acquire these new acetylation sites in order to cope with the increase in body mass and the potential increase in cancer risk.

In comparison to R/Q-to-K, CC analysis of K-to-R/Q showed a significant enrichment of mitochondrial proteins (Figs. 2e and 4d). Mitochondrial protein acetylation occurs mostly through a non-enzymatic pathway, mediated by acetyl-CoA⁴⁵ and to date, only mitochondrial KDACs but not KATs have been identified. Thus, acetylation of proteins within the mitochondria is only partially reversible. Therefore, it is possible that during the evolution of longevity, these mitochondrial longevity-associated acetylation sites were fixed to avoid dependence on fluctuating acetyl-CoA levels. Importantly, the partial dependency in enzymatic activities can lead to a large number of acetylated mitochondrial proteins, and the finding of many mitochondrial pathways here might be inevitable. However, in comparison to the number of all mitochondrial acetylation sites within the global acetylome, the amount of longevity-associated mitochondrial sites within all PHARAOH K-to-Q/R results is statistically significant (Chi-square test, P < 0.00001). Thus, the high representation of mitochondrial pathways identified by PHARAOH’s significant fixed sites is not arbitrary. In addition, fixation of these sites enables the acetyl-CoA to be used in other cellular compartments or for other purposes, such as a source of energy by the TCA cycle. Indeed, recent findings showed that increased longevity is associated with maintaining energy production in old age^25,46,47. Moreover, fixation of acetylation might reduce the dependency of KDACs on external stimuli for releasing Acetyl-CoA for energy.

The conversion of 482 acetylation sites of short-lived mammals to R or Q in long-lived mammals was significantly associated with longevity. Yet, the impact of individual replacements on lifespan extension and whether some have redundant effects is not known. In line with this, as seen in Supplementary Fig. 4, K-to-R/Q transitions occurred gradually during the evolution of longevity. It would be interesting to use machine learning tools to explore which combination of converted sites has the highest positive effect on longevity. Nevertheless, acetylation is only one modification out of ~400 known PTMs. Further research is required to explore the impact of different combinations of lysine acetylation and other modifications on longevity.

Do all pathways contribute to longevity? To address this, we performed pathway analysis using non-significant sites that fit the selection criteria applied by PHARAOH. A comparison of the top 20 identified pathways with all longevity-associated pathways revealed three that were not overlapping: mitochondrial protein degradation, cytoskeleton in muscle cells, and carboxylic acid metabolic processes. Since these pathways involve up to hundreds of proteins, some of which are shared with significant pathways, it’s challenging to draw evolutionary conclusions about lifespan extension. This complexity is compounded by the absence of methods to analyze how specific protein modifications affect the broader activity of these proteins within pathways.

Most animals die before reaching old age due to various causes, leaving natural selection with a limited window to influence lifespan. Thus, several theories, such as mutation accumulation, antagonistic pleiotropy, and the disposable soma theory, have been proposed to explain the evolution of aging and lifespan⁴⁸. The antagonistic pleiotropy theory suggests that alleles or pathways that have a beneficial effect early in life, and therefore tend to accumulate, also have a detrimental effect later in life, thereby promoting aging. In the context of longevity-associated acetylations, one could propose that the acquisition of new acetylation sites (R/Q-to-K replacements) allows the organism to maintain early-life benefits, while deacetylation or acetylation can mitigate the negative effects in later life. An example of this might be USP10, which promotes growth via PCNA/cell cycle regulation during youth, while potentially its acetylation in later life slows these processes, reducing the risk of age-related cancer. The alternative disposable soma theory suggests that aging arises from an imbalance in resource allocation, prioritizing reproduction over somatic maintenance, given the likelihood of early death from external threats. However, improvements in metabolism could enable organisms to invest more in somatic maintenance and repair. Indeed, replacements leading to constant acetylation or deacetylation (K-to-R/Q replacements) are often linked to metabolic pathways, such as beta-oxidation and glucose metabolism, which fits the disposable soma theory by enhancing energy production. These improvements in repair, particularly those related to genome stability, would also benefit from R/Q-to-K replacements that affect DNA repair, promoting the cellular flexibility required for fitness. They will also allow complexity, particularly of the brain, that is associated with increased lifespan⁴⁹.

One striking finding of this study is that in contrast to global acetylation sites, the flanking polypeptide sequences of the longevity-associated acetylation sites are enriched or depleted in specific amino acids (Figs. 2g, h and Supplementary Fig. 9g, h). This finding suggests that the activity of one or more specific KDAC/s or KAT/s regulates longevity. In yeast, genetic manipulation of KDACs, such as hda1, hda2, or rpd3 deletion, or Sir2 overexpression significantly extends lifespan^22,50,51. In addition, deletion of the KAT SAS2 significantly extends yeast lifespan²⁰. Likewise, in worms and flies, mutation or overexpression of various KDACs/KATs was found to extend their lifespan^50,52,53. In mammals, only the manipulation of sirtuins was found to extend mouse lifespan. Overexpression of hypothalamic-specific SIRT1 extends mouse lifespan by 10%⁵⁴ and whole body SIRT6 overexpression extends mouse lifespan by up to almost 30%⁴⁶. Thus, it would be of great interest to explore which of the identified longevity-associated acetylation loci is regulated by SIRT6. Moreover, to date, no manipulation in KAT activity or levels has been found to extend mammalian lifespan. Accordingly, it would be important to elucidate if a mammalian homolog of KAT that was found to extend yeast/worm/fly lifespan, or another as yet unidentified KAT homolog, regulates mammalian lifespan. Identification of such a KAT, together with SIRT6, would provide potential targets for developing drugs supporting healthspan extension.

In this study, we characterized the longevity-related acetylome and its associated pathways. These findings shed light on the previously unknown role of PTMs in general, particularly acetylation, in regulating longevity. In addition, this set of acetylation sites and, as yet unidentified, key regulators of these changes during evolution can suggest future therapeutic targets to promote longevity.

The research conducted in this study complies with all relevant ethical regulations. All procedures involving mice and experimental protocol (32-06-20) were approved by the Institutional Animal Care and Use Committee of Bar Ilan University and by the Ministry of Health of Israel.

Animal experimentation and acetylome generation

Mice were housed on a 12 h light / dark cycle at room temperature (22–24 °C) with a relative humidity within 45–65% in the Bar Ilan animal facilities. Mice were maintained on a standard rodent chow diet with ad libitum access to food and water. All mice used in these experiments were in good health. Mice were kept under specific pathogen-free conditions in IVC cages that were routinely screened and found negative for viral serology and both endo and ectoparasites. Male littermates were used for all experiments.

LC-MS/MS-based acetylome

Lysine acetylome of mouse liver was performed as described⁵⁵ with minor modifications. Briefly, liver tissues from 18 male mice (C57BL/J6) were homogenized in denaturation buffer containing 9 M urea, 20 mM HEPES, 1 µM TSA, and 5 mM NAM. Samples were sonicated for 20 s with an amplitude of 40%, and then centrifuged. Next, 7 mg lysate was taken from each sample and added to 1 mM DTT. Samples were then incubated for 30 min at room temperature with shaking at 1000 rpm, added to 5 mM IAA, and incubated for an additional 30 min. Trypsin / LysC (1 µg / 1 mg protein) cleavage was performed at room temperature for 2-3 h. The samples were then diluted with 20 mM HEPES to a final urea concentration of 2 M. Trypsin was added at a ratio of 1 µg / 100 µg protein, and the samples were incubated overnight at room temperature. The samples were then added to 1% acetonitrile (ACN) and adjusted to pH <3 with TFA. The samples were purified on C18 columns (Waters WAT036800), and the eluate was taken for IP acetyl lysine. A mixture of two pan-acetyl lysine antibody beads (Immunochem and Cell Signaling) was used at a 1:1 ratio⁵⁵. The process was then continued following the Cell Signaling Acetylome kit protocol.

Mass spectrometry analysis

The tryptic peptides were desalted using C18 tips (Top tip, Glygen), dried, and re-suspended in 0.1% Formic acid. The peptides were resolved by reverse-phase chromatography on 0.075 × 180-mm fused silica capillaries (J&W) packed with Reprosil reversed phase material (Dr Maisch GmbH, Germany). The peptides were eluted with different concentration of Acetonitrile with 0.1% of formic acid: a linear 180 min gradient of 5 to 28% acetonitrile followed by a 15 min gradient of 28–95% and 25 min at 95% acetonitrile with 0.1% formic acid in water at flow rates of 0.15 μl/min. Mass spectrometry was performed by Q Executive HFX mass spectrometer (Thermo) in a positive mode (m/z 300–1800, resolution 120,000 for MS1 and 15,000 for MS2) using repetitively full MS scan followed by collision induces dissociation (HCD, at 27 normalized collision energy) of the 30 most dominant ions (>1 charges) selected from the first MS scan. The AGC settings were 3 × 106 for the full MS and 1 × 105 for the MS/MS scans. The intensity threshold for triggering MS/MS analysis was 1 × 104. A dynamic exclusion list was enabled with an exclusion duration of 20 s.

Data Analysis

The mass spectrometry data was analyzed using the MaxQuant software 1.5.2.8^56,57 for peak picking and identification using the Andromeda search engine, searching against the mouse proteome from the Uniprot database with mass tolerance of 6 ppm for the precursor masses and the fragment ions. Oxidation on Met, Dimethyl (KR), Methyl (KR), Trimethyl (K), Acetyl (K), and protein N-terminus acetylation were accepted as variable modifications, and carbamidomethyl on cysteine was accepted as static modifications. Minimal peptide length was set to six amino acids, and a maximum of two miscleavages was allowed. The data was quantified by label-free analysis using the same software. Peptide- and protein-level false discovery rates (FDRs) were filtered to 1% using the target-decoy strategy. Protein tables were filtered to eliminate the identifications from the reverse database and common contaminants and single peptide identifications.

Orthologous proteins and phylogenetic tree prediction

OrthoFinder 2.5.4²⁸ was applied to compute the orthologs of mouse or human proteins in 107 mammals. To reduce computation time, the proteomes of human/mouse were grouped with those of other mammals within separate directories, allowing the algorithm to be executed in parallel across these directories. Next, a phylogenetic tree comprising all 107 mammals was constructed based on the calculation of all versus all orthologs.

PHARAOH analyzed two phylogenetic trees, the Zoonomia project tree³¹, containing 62 animals overlapping with our dataset, and the OrthoFinder²⁸-based tree using our 107-mammals’ dataset (Supplementary Fig. 3a). Final outcomes were derived from the intersection of significant findings obtained from both analyses. The distribution of these results is presented in Supplementary Fig. 3b–e.

Phylogenies and comparative methods

Statistical validation

Since during evolution different organisms evolved from common ancestors, it introduces a level of dependence between animals. Thus, the standard assumption of variable independence cannot be used for the statistical analysis of the PHARAOH⁵⁸. To acknowledge the fact that species are not fully independent, a novel statistical test was devised that incorporates the evolutionary context. This allows the examination of the relationship between amino acid changes and longevity, regardless of phylogenetic distances. The statistical test consisted of two parts. First, the correlation between maximal lifespan and AA replacement was examined; then, a correction was introduced for the distances between the animals on the evolutionary tree. A pseudocode illustrating the algorithm of PHARAOH is shown in Supplementary Data 9.

Part I (Supplementary Fig. 2a) – correlation between amino acid exchange and longevity-

Data preparation

The acetylome datasets used for the replacement matrix are based on the 14 AA flanking the amino acid of interest. An entry in the matrix is denoted as: Mi,j = the amino acid found in mammal j, at the orthologous site to the i acetylation site in human/mouse.

For each row in the replacement matrix, the mammals were divided into two groups based on the amino acid found at the specific acetylated site: lysine site of reference (K) (includes either mouse or human), or arginine (R)/glutamine (Q)/other. Each group contains the maximal lifespan in captivity for each member.

Correlation test

To identify AA substitutions (K-to-R/Q) that significantly correlate with longevity, the Mann–Whitney test was employed to compare the lifespans between the two above-mentioned groups that were divided based on their amino acid found at the acetylation site. A minimum of three mammals was required in each group to ensure significant findings. False Discovery Rate (FDR) correction was applied to account for multiple testing, and a significance level of q-value ≤ 0.1 was adopted.

Part II (Supplementary Fig. 2b and c) – phylogenetic correction

The assumption that the lifespan difference between the two mammals in each comparison is derived from an exchange of the examined acetylated lysine, without contributions from phylogeny, was taken as the H₀ hypothesis.

To normalize the impact of the phylogenetic tree on the statistical validation process, each acetylation site (represented by a row in the replacement matrix) was further divided into a set of pairs derived from the database. The relationship between a pair of mammals is represented by three factors (Supplementary Fig. 2, table):

• Lifespan difference.

• Phylogenetic (tree) distance.

• The amino acid (AA) found in the examined site.

Importantly, this part is independent of Part I and aims to examine if the difference in lifespan is due to changes in amino acids or their relative position on the evolutionary tree.

Strengthening, weakening, and non-informative categories

The entries in the table shown in Supplementary Fig. 2 are categorized into three distinct groups based on their impact on the hypothesis: strengthening, weakening, or not informative.

• Claims that are not informative with regard to the hypothesis (colored in gray in the table):

  • Pairs that have a small difference in lifespan, a small distance in the phylogenetic tree, and have the same amino acid at the acetylation site (e.g., human and chimp sharing the same AA at the specific site).

  • Pairs that have a large difference in lifespan, a large distance in the phylogenetic tree, and have different amino acids at the acetylation site (e.g., mouse and human that have different AAs at a specific site).

• Claims that strengthen the hypothesis (colored in green in the table):

  • Pairs that have a small difference in lifespan, a large distance in the phylogenetic tree, and have the same amino acid at the acetylation site (e.g., human and whale sharing the same AA at a specific site).

  • Pairs that have a large difference in lifespan, a small distance in the phylogenetic tree, and have different amino acids at the acetylation site (e.g., mouse and naked mole rat, having different AAs at a specific site).

• Claims that weaken the hypothesis (colored in red in the table):

  • Pairs that have a small difference in lifespan, a small distance in the phylogenetic tree, and have different amino acids at the acetylation site (e.g., human and chimp having a different AA at a specific site).

  • Pairs that have a small difference in lifespan, a large distance in the phylogenetic tree, and have different amino acids at the acetylation site (e.g., human and whale having different AAs at a specific site).

  • Pairs that have a large difference in lifespan, a small distance in the phylogenetic tree, and have the same amino acid in the acetylation site (e.g., mouse and naked mole rat having the same AA at a specific site).

  • Pairs that have a large difference in lifespan, a large distance in the phylogenetic tree, and have the same amino acid at the acetylation site (e.g., mouse and human having the same AA at a specific site).

Weight calculation

Weights were assigned to each comparison according to the following equation:

Where P is the probability of the mammalian pair occupying a specific position in the table (Supplementary Fig. 2b, right column), regardless of the amino acid found at the acetylation site. For instance, the probability of a pair from the data falling into the first strengthening row is determined by their lifespan difference being smaller than average, AND their distance in the phylogenetic tree being larger than average. The strengthening or weakening category is assigned positive or negative weights, respectively (table in Supplementary Fig. 2).

Score calculation

Each acetylation site was addressed as a collection of mammalian pairs, with each pair assigned a position/category in the options table.

The overall score for an acetylation site was computed as shown in the following equation:

In which the score for each site is calculated as the sum of the values received from all two-mammal comparisons: weight multiplied by the difference in lifespans and the distance in the phylogenetic tree (both normalized to numbers between 0 and 1, see table in Supplementary Fig. 2).

Permutation Test

To determine a p-value for each site’s score, 1000 permutations of the non-zero weights were performed, effectively randomizing the positions where pairs are placed in the tree. FDR correction was applied to adjust for multiple testing.

Changed acetylation sites are considered longevity-associated only in case they are significant in both the Mann–Whitney test and the phylogenetic tree validation described here (Supplementary Fig. 2, bottom).

PHARAOH implementation

The PHARAOH tool was fully implemented in Python 3. The replacement matrix algorithm uses SQLite3 tables as the database. Pandas, more_itertools, and NumPy Python libraries were used throughout the creation of the matrix. In the statistical validation, Bio (biopython) was used for working with phylogenetic data (Phylo), as well as statsmodels and scipy.stats for the Mann-Whitney and FDR tests. Matplotlib was used for basic visualization of the outputs.

Lysine residue conservation

PHARAOH was employed to investigate the conservation of lysine residues. Matplotlib was used for basic visualization, with or without acetylation, between mouse and human, by comparing proteomics/acetylome data. The conservation patterns within the proteomes were independently assessed. Orthologous proteins were obtained through OrthoFinder, and the alignment of these orthologous protein sequences was conducted using the Linux version of bl2seq. To identify unacetylated lysines, the acetylome sites were subtracted from the proteome outcomes. Subsequently, the percentage of each amino acid replacement was calculated in relation to the total count of lysines in the proteome, or the acetylated lysines present within the acetylome.

Analysis of the acetylome results

Pathway enrichment analyses were performed via the Metascape site⁵⁹, which uses MCODE⁶⁰ and various interaction tools to create the protein-protein interaction network. Cytoscape⁶¹ was used to edit and visualize the PPI networks.

Gene ontology was used for CC analysis, and the results were manually divided into categories of extracellular, nucleus, cytoplasm, plasma membrane, mitochondrion, endoplasmic reticulum, and others.

pLogo⁶² (probability Logo generator) was used with the appropriate background (mouse background for the mouse acetylome and human for the human acetylome).

TRRUST⁶³ database was used for human transcription factor analysis. Gene-disease association analysis was performed using DisGeNET⁶⁴.

R was used via RStudio to enable visualization of the results. ggtree⁶⁵ was used for visualization of the phylogenetic tree, and gheatmap was used to append the heatmaps to the phylogenetic tree. A heatmap of the replacement matrix data was created by ComplexHeatmap⁶⁶, filtered only to the significant results and the relevant changes of AAs. Proteomics data from the NIH Cancer Institute⁶⁷ was used for Pearson correlation between USP10 and PCNA in cancer data, visualized and calculated using R. Venn diagrams for the method’s visualizations were also created in R using eulerr⁶⁸.

Cloning and mutagenesis

The human and mouse CBS and the human USP10 cDNA sequences were cloned into a pcDNA3.1+ vector using EcoRI and XhoI (CBS) or XhoI and XbaI (USP10) restriction enzymes. All genes were tagged with a flag tag at the C-terminal region. Site-directed mutagenesis was done with PfuUltra II Fusion HS DNA Polymerase kit (Agilent), following the manufacturer’s protocol. Primers used here are listed in Supplementary Table 1 in the Supplementary Information file.

Cell culture and treatments

HEK293T and HCT116 cell lines were purchased from ATCC (cat. CRL-3216 and CCL-247, respectively) and grown in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1% glutamine, and 1% penicillin/streptomycin in 5% CO₂. For PCNA ubiquitin-related degradation, HCT116 cells were treated with 100 µg/ml cycloheximide for 8 h, and then harvested with PBSx1 and lysed in urea buffer for western blot analysis.

Transfection

HEK293T cells were seeded in 10 cm plates as previously published²⁵. The cells were grown for 48 h before harvesting in cold PBS x1.

HCT116 cells were transfected using Lipofectamine 3000 (Thermo L3000008) following the manufacturer’s protocol.

Flag immunoprecipitation (IP-flag)

HEK293T transfected with CBS-flag were harvested in cold PBS x1 and lysed on ice with lysis buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1% Triton, 0.5% NP40, 10% glycerol) for 30 min. The CBS-flag was immunoprecipitated from 1 mg lysate as previously described²⁵ and used for the H₂S production assay.

H₂S production capacity assay

H₂S capacity was measured as previously described³⁸ using 200–500 µg cell lysate or 5 µl eluate after IP-flag of CBS, supplemented with 10 mM L-cysteine, 10 µM PLP, and 10 mM L-homocysteine. Lead acetate papers were purchased from Sigma (37104-1EA). Measurements were quantified using ImageJ.

Western blot and antibodies

Western blot analyses were done as previously published²⁵, using antibodies for PCNA (Cell Signaling 13110S) or USP10 (Cell Signaling CST-8501S) as primary antibody and the appropriated HRP-conjugated secondary antibodies (ENCO). HRP-conjugated primary antibodies were used against Ubiquitin (Santa Cruz 8017), Flag (Proteintech HRP-66008), and Tubulin (Proteintech HRP-66240). Uncropped images are shown in the Data Source and Supplementary Information files.

Statistical analysis for H₂S production and western blot analyses

For the analysis of H₂S production capacity assay and western blot measurements, statistical significance was tested using one-way ANOVA.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

1. Zuo, W., Jiang, S., Guo, Z., Feldman, M. W. & Tuljapurkar, S. Advancing front of old-age human survival. Proc. Natl Acad. Sci. USA 115, 11209–11214 (2018).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

2. Shilovsky, G. A., Putyatina, T. S. & Markov, A. V. Evolution of longevity as a species-specific trait in mammals. Biochemistry 87, 1579–1599 (2022).

   CAS  PubMed  Google Scholar 

3. Matheu, A. et al. Delayed ageing through damage protection by the Arf/p53 pathway. Nature 448, 375–379 (2007).

   Article  ADS  CAS  PubMed  Google Scholar 

4. Abegglen, L. M. et al. Potential mechanisms for cancer resistance in elephants and comparative cellular response to DNA damage in humans. JAMA 314, 1850–1860 (2015).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

5. Sulak, M. et al. TP53 copy number expansion is associated with the evolution of increased body size and an enhanced DNA damage response in elephants. eLife 5, e11994 (2016).

   Article  PubMed  PubMed Central  Google Scholar 

6. de Magalhães, J. P., Curado, J. & Church, G. M. Meta-analysis of age-related gene expression profiles identifies common signatures of aging. Bioinformatics 25, 875–881 (2009).

   Article  PubMed  PubMed Central  Google Scholar 

7. Lu, J. Y. et al. Comparative transcriptomics reveals circadian and pluripotency networks as two pillars of longevity regulation. Cell Metab. 34, 836 (2022).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

8. Tyshkovskiy, A. et al. Distinct longevity mechanisms across and within species and their association with aging. Cell 186, 2929–2949.e20 (2023).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

9. Balasubramanian, P., Howell, P. R. & Anderson, R. M. Aging and caloric restriction research: a biological perspective with translational potential. EBioMedicine 21, 37 (2017).

   Article  PubMed  PubMed Central  Google Scholar 

10. Papadopoli, D. et al. mTOR as a central regulator of lifespan and aging. F1000Res. 8, (2019).

11. Sathyan, S. et al. Plasma proteomic profile of age, health span, and all-cause mortality in older adults. Aging Cell 19, e13250 (2020).

12. Coenen, L., Lehallier, B., de Vries, H. E. & Middeldorp, J. Markers of aging: unsupervised integrated analyses of the human plasma proteome. Front. Aging 4, 1112109 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

13. Lehallier, B. et al. Undulating changes in human plasma proteome profiles across the lifespan. Nat. Med. 25, 1843–1850 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

14. Panyard, D. J., Yu, B. & Snyder, M. P. The metabolomics of human aging: advances, challenges, and opportunities. Sci Adv. 8, eadd6155 (2022).

15. Khoury, G. A., Baliban, R. C. & Floudas, C. A. Proteome-wide post-translational modification statistics: frequency analysis and curation of the swiss-prot database. Sci. Rep. 1, 90 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

16. Santos, A. L. & Lindner, A. B. Protein posttranslational modifications: roles in aging and age-related disease. Oxid. Med. Cell Longev. 2017, 5716409 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

17. Walsh, C. T., Garneau-Tsodikova, S. & Gatto, G. J. Protein posttranslational modifications: the chemistry of proteome diversifications. Angew. Chem. Int. Ed. 44, 7342–7372 (2005).

    Article  CAS  Google Scholar 

18. Sadoul, K., Boyault, C., Pabion, M. & Khochbin, S. Regulation of protein turnover by acetyltransferases and deacetylases. Biochimie 90, 306–312 (2008).

    Article  CAS  PubMed  Google Scholar 

19. Lu, J. Y., Lin, Y. Y., Zhu, H., Chuang, L. M. & Boeke, J. D. Protein acetylation and aging. Aging 2, 911–912 (2011).

    Article  Google Scholar 

20. Dang, W. et al. Histone H4 lysine 16 acetylation regulates cellular lifespan. Nature 459, 802–807 (2009).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

21. Nakayasu, E. S. et al. Ancient regulatory role of lysine acetylation in central metabolism. mBio 8, e01894-17 2017).

22. Kaeberlein, M., McVey, M. & Guarente, L. The SIR2/3/4 complex and SIR2 alone promote longevity in Saccharomyces cerevisiae by two different mechanisms. Genes Dev. 13, 2570–2580 (1999).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

23. Peleg, S. et al. Life span extension by targeting a link between metabolism and histone acetylation in Drosophila. EMBO Rep. 17, 455–469 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

24. Bradshaw, P. C. Acetyl-CoA metabolism and histone acetylation in the regulation of aging and lifespan. Antioxidants 10, 572 (2021).

25. Naiman, S. et al. SIRT6 Promotes Hepatic Beta-oxidation Via Activation of PPARα. Cell Rep. 29, 4127–4143.e8 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

26. Cohen, H. Y. et al. Acetylation of the C terminus of Ku70 by CBP and PCAF controls Bax-mediated apoptosis. Mol. Cell 13, 627–638 (2004).

    Article  CAS  PubMed  Google Scholar 

27. Hornbeck, P. V. et al. PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res. 43, D512–D520 (2015).

    Article  CAS  PubMed  Google Scholar 

28. Emms, D. M. & Kelly, S. OrthoFinder: phylogenetic orthology inference for comparative genomics. Genome Biol. 20, 1–14 (2019).

    Article  Google Scholar 

29. Bateman, A. et al. UniProt: the universal protein knowledgebase in 2023. Nucleic Acids Res. 51, D523–D531 (2023).

    Article  CAS  Google Scholar 

30. Tacutu, R. et al. Human ageing genomic resources: new and updated databases. Nucleic Acids Res. 46, D1083–D1090 (2018).

    Article  CAS  PubMed  Google Scholar 

31. Vignieri, S. Zoonomia. Science 380, 356–357 (2023).

    Article  ADS  CAS  PubMed  Google Scholar 

32. Gu, W. & Roeder, R. G. Activation of p53 sequence-specific DNA binding by acetylation of the p53 C-terminal domain. Cell 90, 595–606 (1997).

    Article  CAS  PubMed  Google Scholar 

33. ADW: Home. https://animaldiversity.org/ (2020).

34. Longevity records: life spans of mammals, birds, amphibians, reptiles, and fish. https://www.demogr.mpg.de/longevityrecords/0203.htm (2002).

35. Hine, C. & Mitchell, J. R. Calorie restriction and methionine restriction in control of endogenous hydrogen sulfide production by the transsulfuration pathway. Exp. Gerontol. 68, 26–32 (2015).

    Article  CAS  PubMed  Google Scholar 

36. Polevoda, B. & Sherman, F. The diversity of acetylated proteins. Genome Biol. 3, reviews0006 (2002).

37. Martín-Montalvo, A., Villalba, J. M., Navas, P. & De Cabo, R. NRF2, cancer and calorie restriction. Oncogene 30, 505–520 (2011).

    Article  PubMed  Google Scholar 

38. Hine, C. et al. Endogenous hydrogen sulfide production is essential for dietary restriction benefits. Cell 160, 132–144 (2015).

    Article  CAS  PubMed  Google Scholar 

39. Park, J. M. et al. Modification of PCNA by ISG15 plays a crucial role in termination of error-prone translesion DNA synthesis. Mol. Cell 54, 626–638 (2014).

    Article  CAS  PubMed  Google Scholar 

40. Zhou, H. et al. The prognostic value of proliferating cell nuclear antigen expression in colorectal cancer: a meta-analysis. Medicine 97, e13752 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

41. Tollis, M., Boddy, A. M. & Maley, C. C. Peto’s paradox: how has evolution solved the problem of cancer prevention? BMC Biol. 15, 1–5 (2017).

    Article  Google Scholar 

42. Tidwell, T. R., Søreide, K. & Hagland, H. R. Aging, metabolism, and cancer development: from Peto’s paradox to the Warburg effect. Aging Dis. 8, 662–676 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

43. Tao, L. et al. USP10 as a potential therapeutic target in human cancers. Genes 13, 831 (2022).

44. Paul, B. D., Snyder, S. H. & Kashfi, K. Effects of hydrogen sulfide on mitochondrial function and cellular bioenergetics. Redox Biol. 38, 01772 (2021).

45. Baeza, J., Smallegan, M. J. & Denu, J. M. Mechanisms and dynamics of protein acetylation in mitochondria. Trends Biochem. Sci. 41, 231–244 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

46. Roichman, A. et al. Restoration of energy homeostasis by SIRT6 extends healthy lifespan. Nat. Commun. 12, 1–18 (2021).

    Article  Google Scholar 

47. Zhou, Q., Yu, L., Cook, J. R., Qiang, L. & Sun, L. Deciphering the decline of metabolic elasticity in aging and obesity. Cell Metab. 35, 1661–1671.e6 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

48. Kirkwood, T. B. L. & Austad, S. N. Why do we age? Nature 408, 233–238 (2000).

    Article  ADS  CAS  PubMed  Google Scholar 

49. Sol, D. Revisiting the cognitive buffer hypothesis for the evolution of large brains. Biol. Lett. 5, 130–133 (2009).

    Article  PubMed  Google Scholar 

50. Yu, R. et al. Inactivating histone deacetylase HDA promotes longevity by mobilizing trehalose metabolism. Nat. Commun. 12, 1–16 (2021).

    ADS  Google Scholar 

51. Kim, S., Benguria, A., Lai, C. Y. & Jazwinski, S. M. Modulation of life-span by histone deacetylase genes in Saccharomyces cerevisiae. Mol. Biol. Cell 10, 3125 (1999).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

52. Ikeda, T., Uno, M., Honjoh, S. & Nishida, E. The MYST family histone acetyltransferase complex regulates stress resistance and longevity through transcriptional control of DAF-16/FOXO transcription factors. EMBO Rep. 18, 1716–1726 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

53. Zhao, Y. et al. Immunosuppression induced by brain-specific HDAC6 knockdown improves aging performance in Drosophila melanogaster. Phenomics 2, 194–200 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

54. Satoh, A. et al. Sirt1 extends life span and delays aging in mice through the regulation of Nk2 homeobox 1 in the DMH and LH. Cell Metab. 18, 416–430 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

55. Rardin, M. J. et al. Label-free quantitative proteomics of the lysine acetylome in mitochondria identifies substrates of SIRT3 in metabolic pathways. Proc. Natl Acad. Sci. USA 110, 6601–6606 (2013).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

56. Cox, J. et al. Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol. Cell Proteom. 13, 2513 (2014).

    Article  CAS  Google Scholar 

57. Tyanova, S. et al. The Perseus computational platform for comprehensive analysis of (prote)omics data. Nat. Methods 13, 731–740 (2016).

    Article  CAS  PubMed  Google Scholar 

58. Felsenstein, J. Phylogenies and the comparative method author. Am. Nat. 125, 1–15 (1985).

    Article  Google Scholar 

59. Zhou, Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat. Commun. https://doi.org/10.1038/s41467-019-09234-6 (2019).

60. Bader, G. D. & Hogue, C. W. V. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinform. 4, 1–27 (2003).

    Article  Google Scholar 

61. Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13, 2498–2504 (2003).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

62. O’Shea, J. P. et al. pLogo: a probabilistic approach to visualizing sequence motifs. Nature Methods 10, 1211–1212 (2013).

63. Han, H. et al. TRRUST: a reference database of human transcriptional regulatory interactions. Sci. Rep. 5, 1–11 (2015).

    CAS  Google Scholar 

64. Piñero, J., Saüch, J., Sanz, F. & Furlong, L. I. The DisGeNET cytoscape app: exploring and visualizing disease genomics data. Comput. Struct. Biotechnol. J. 19, 2960–2967 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

65. Yu, G., Lam, T. T. Y., Zhu, H. & Guan, Y. Two methods for mapping and visualizing associated data on phylogeny using Ggtree. Mol. Biol. Evol. 35, 3041–3043 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

66. Gu, Z. Complex heatmap visualization. iMeta 1, e43 (2022).

67. Thangudu, R. R. et al. Abstract LB−242: proteomic data commons: a resource for proteogenomic analysis. Cancer Res 80, LB−242 (2020).

    Article  Google Scholar 

68. Micallef, L. & Rodgers, P. eulerAPE: drawing area-proportional 3-Venn diagrams using ellipses. PLoS ONE 9, e101717 (2014).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

69. Perez-Riverol, Y. et al. The PRIDE database at 20 years: 2025 update. Nucleic Acids Res. 53, D543–D553 (2025).

    Article  PubMed  Google Scholar 

Download references

Author notes

1. These authors contributed equally: S. Feldman-Trabelsi, N. Touitou.

Authors and Affiliations

1. The Mina & Everard Goodman Faculty of Life Sciences, Bar-Ilan University, Ramat-Gan, Israel

   S. Feldman-Trabelsi, N. Touitou, R. Nagar, Z. Schwartz, A. Michelson, S. Shaki, M. Y. Avivi, B. Lerrer & H. Y. Cohen

2. The Sagol Healthy Human Longevity Center, Bar-Ilan University, Ramat-Gan, Israel

   S. Feldman-Trabelsi, N. Touitou, R. Nagar, Z. Schwartz, A. Michelson, S. Shaki, M. Y. Avivi, B. Lerrer & H. Y. Cohen

3. Department of Evolutionary and Environmental Biology, University of Haifa, Haifa, Israel

   S. Snir

Contributions

H.Y.C. conceived and supervised the project, analyzed the results, and wrote the manuscript, which was reviewed by all of the authors. S.F.T. created the computational tool, performed computational and bioinformatics analyses, and wrote the manuscript. N.T. planned and performed the in vitro experiments and analyzed the results. R.N. and Z.S. performed the USP10-related experiments and analyzed the results. A.M. and S.Sh. assisted with the computational and bioinformatics experiments. M.Y.A. created in silico and analyzed CBS structure. B.L. assisted with biological experiments and data interpretation. S.Sn. provided expertise with respect to creating a statistical model for the computational tool and bioinformatics experiments.

Corresponding author

Correspondence to H. Y. Cohen.

Competing interests

The authors declare no competing interests.

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions