The C-terminus of SIRT1 can be cleaved by cathepsin B at amino acid H533 to generate a lower-functioning, N-terminally intact 75 kDa polypeptide (75SIRT1) that might be involved in age-related pathologies. However, the mechanisms underlying cathepsin B docking to and cleavage of SIRT1 are unclear. Here, we first identified several 75SIRT1 variants that are augmented with aging correlatively with increased cathepsin B levels in various mouse tissues, highlighting the possible role of this cleavage event in age-related pathologies. Then, based on H533 point mutation and structural modeling, we generated a functionally intact ΔSIRT1 mutant, lacking the internal amino acids 528–543 (a predicted C-terminus loop domain), which exhibits resistance to cathepsin B cleavage in vitro and in cell cultures. Finally, we showed that cells expressing ΔSIRT1 under pro-inflammatory stress are more likely to undergo caspase 9- dependent apoptosis than those expressing 75SIRT1. Thus, our data suggest that the 15-amino acid predicted loop motif embedded in the C-terminus of SIRT1 is susceptible to proteolytic cleavage by cathepsin B, leading to the formation of several N-terminally intact SIRT1 truncated variants in various aging mouse tissues.
Tissue aging is often accompanied by chronic, low-dose inflammation, which is usually correlated with the development of age-associated diseases, in a process referred to as inflamm-aging (Franceschi et al., 2000). It is well established that inflamm-aging involves a systemic increase in pro-inflammatory biomarkers [e.g. interleukin 6 (IL6)], which contributes to the development of chronic diseases, in part, due to progressive cell senescence (Franceschi et al., 2000; Franceschi and Campisi, 2014; Maggio et al., 2006). Senescing cells of aging tissues have been established as a source for a senescence-associated secretory phenotype (SASP), which contributes to an increase in C-reactive proteins, IL6 and proteolytic enzymes, eventually causing tissue damage and loss of function with age (Tchkonia et al., 2013; Campisi and Robert, 2014; Grabowska et al., 2017).
One of the most studied enzymes in the process of aging and age-related pathologies is SIRT1, a ubiquitous nicotinamide adenine dinucleotide (NAD+)-dependent protein deacetylase, which regulates the histone and nonhistone acetylation status of numerous targets within the cell (McBurney et al., 2013; Haigis and Sinclair, 2010; Anderson et al., 2003). The regulation of SIRT1 activity is of crucial importance for numerous physiological processes (Haigis and Sinclair, 2010), and loss of SIRT1 activity can considerably disrupt cellular homeostasis and facilitate various metabolic and age-related diseases (‘early aging’) (Wang et al., 2012; Houtkooper et al., 2012; Dvir-Ginzberg et al., 2008; Herskovits and Guarente, 2014). One of the mechanisms by which SIRT1 might regulate early aging is through its ability to deacetylate RELA (Yeung et al., 2004), a key component of the inflammatory mediator NFκB. Significantly, the activity of both SIRT1 and NFκB is highly linked with the ability of the cells to secrete SASP components (Hayakawa et al., 2015; Freund et al., 2010). Thus, maintaining the normal deacetylase activity of SIRT1 might inhibit SASP secretion to prevent age-related tissue damage and dysfunction.
Why the enzymatic activity of SIRT1 is disrupted during aging is, as yet, unclear, although several (not mutually exclusive) factors have been hypothesized to affect this activity; most notably, the availability of NAD in the cell and various translational and post-translational modifications, such as alternative splicing, SIRT1 phosphorylation, ubiquitination or SUMOylation (Lynch et al., 2010; Sasaki et al., 2008; Peng et al., 2015; Yang et al., 2007). We previously found that, in human cartilage samples and mouse models of osteoarthritis (OA; an age-associated disease), inflammation induces reduced SIRT1 activity via its truncation by the lysosomal cysteine protease cathepsin B (Dvir-Ginzberg et al., 2011; Elayyan et al., 2017). Although cathepsins are typically bound to the acidic lumen of the lysosome (where they participate in protein recycling), exposure to pro-inflammatory mediators, such as TNFα, can drive lysosomal permeability (Guicciardi et al., 2000) and the subsequent release of cathepsins to the cytoplasm and nucleus, where they target cellular proteins and alter various cellular circuits, including epigenetic reprogramming and apoptosis (Dvir-Ginzberg et al., 2011; Duncan et al., 2008; Droga-Mazovec et al., 2008; Oppenheimer et al., 2012; Cirman et al., 2004).
We previously found that the truncation of SIRT1 [i.e. truncated SIRT1 (1–533 amino acids) versus the full-length SIRT1 (1–747 amino acids)] by cathepsin B occurs at amino acid H533 (at the C-terminus of SIRT1) and yields 75SIRT1, a 75-kDa polypeptide with a reduced chromatin binding ability in human chondrocytes (Dvir-Ginzberg et al., 2011; Oppenheimer et al., 2012). The truncation of SIRT1 by cathepsin B was later demonstrated in other models of age-related diseases that have an inflammatory component, including endothelial senescence cells (Chen et al., 2012), ocular autoimmune disease (Gardner et al., 2015) and nonalcoholic steatohepatitis (de Mingo et al., 2016). These findings suggest that the truncation of SIRT1 by cathepsin B to yield the inactive 75SIRT1 variant could be relevant to numerous tissues and pathologies by fine-tuning SIRT1 activity in the cell; however, the structural setting and mechanism by which this truncation occurs are, as yet, unclear. Although cathepsins (unlike caspases) are unlikely to cleave their substrates based on the recognition of particular amino acid sequences (Turk and Stoka, 2007; Biniossek et al., 2011), they might prefer a target based on its structural traits, which could permit their docking onto potential sites of cleavage. Thus, to date, it is unclear whether cathepsin B truncates SIRT1 based on the structural characteristics of SIRT1 or on sequence specificity.
The purpose of this study was twofold: first, we aimed to assess whether SIRT1 truncation occurs in multiple aging tissues in mouse models; and second, we aimed to determine, using in silico structural prediction and biochemical analysis, which structural features of SIRT1 enable the docking of cathepsin B onto SIRT1 and its subsequent cleavage at amino acid H533, and whether H533 is critical for this process. Elucidating the mechanism and dynamics of these docking and cleavage events will contribute to our understanding of various age-associated diseases linked to reduced SIRT1 activity, and might facilitate the design of novel polypeptide inhibitors to rescue SIRT1 from cleavage, thus preserving its normal activity and, accordingly, maintaining cell homeostasis.
Tissue-specific variations in truncated SIRT1 in aged mice
To determine whether the C-terminal truncation of SIRT1 is a widespread phenomenon during aging [rather than a cartilage-specific feature (Dvir-Ginzberg et al., 2008, 2011; Oppenheimer et al., 2012)], we examined crude tissue extracts from various organs derived from either young or aged mice (3 or 16 months old, respectively). The levels of full-length SIRT1 (flSIRT1) in aged mice, as compared with those in young mice, were significantly higher in the liver, significantly lower in the lung and spleen, and not significantly different in the pancreas, spleen or fat tissues. In contrast, the levels of 75SIRT1, which possesses only amino acids 1–533 of SIRT1, were significantly higher in aged mice than in young mice in the lung, liver and brain (Fig. 1A). Importantly, 75SIRT1 in aged mice showed some molecular weight variation between tissues (Fig. 1A). In addition, as compared with young mice, the levels of active cathepsin B were significantly higher in aged mice in tissues extracted from the lung, liver and brain (Fig. 1B). Thus, in these mouse strains, a concomitant age-dependent increase in both 75SIRT1 and cathepsin B was found in the lung, liver and brain.
The H533 amino acid of SIRT1 is not essential for cathepsin B cleavage
To test whether the cleavage of SIRT1 by cathepsin B is sequence-specific or structure-specific, and because the cleaved 75SIRT1 variant bears a histidine (H533) in its P1 site (h.SIRT1; Fig. 2B), we initially assessed whether H533 is essential for SIRT1 cleavage by cathepsin B. To this end, we point-mutated this histidine to alanine, thus generating an H533A flSIRT1 mutant. Because previous proteomic analyses indicated that alanine (and glycine) are preferable for cathepsin B cleavage at this position (Biniossek et al., 2011), we expected that, if the cleavage is sequence-specific, the H533A mutant will be more susceptible to cathepsin B cleavage than the wild-type (WT) flSIRT1. To test this expectation, we transfected HEK293 cells with either the WT flSIRT1 or the H533A flSIRT1 mutant and subjected them to TNFα treatment. Under these conditions, the levels of 75SIRT1 (75 kDa) and flSIRT1 (120 kDa) were found to be similar between cells transfected with the H533A flSIRT1 mutant and those transfected with the WT flSIRT1 (Fig. 2A), indicating that the H533A mutation did not affect SIRT1 cleavage by cathepsin B, as compared with the WT flSIRT1. These data suggest that, at least for the H533 site of SIRT1, cleavage by cathepsin B is structure-specific rather than sequence-specific. To further explore this possibility, we modeled the structures of flSIRT1 and 75SIRT1 (Fig. 2C; SD1–SD4 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973) and employed a directed docking analysis for cathepsin B on the flSIRT1 (Fig. 2C; SD5 and SD6 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973). In this analysis, cathepsin B docked in an unstructured loop motif at amino acids 522–550 in the C-terminal domain of flSIRT1 (Fig. 2D).
The predicted unstructured C-terminal loop motif of SIRT1 is essential for cleavage by cathepsin B
Next, we aimed to use HeLa cell lines to test our SIRT1 mutants. To this end, we first wanted to confirm that SIRT1 cleavage in these cell lines is indeed mediated by cathepsin B, rather than by other cathepsins. Thus, we co-incubated the cells either with a specific cathepsin B inhibitor (CA-074me) or with a pan-cathepsin inhibitor (GB111-NH2), and found that, under TNFα stimulation, both inhibitors decreased the levels of 75SIRT1 to a similar degree, as compared with untreated flSIRT1 controls (Fig. 3A).
To examine whether the unstructured 522–550 loop of flSIRT1 is essential for cathepsin B cleavage, we generated ΔSIRT1, a recombinant SIRT1 mutant lacking amino acids 528–543. Then, we transfected ΔSIRT1, flSIRT1 or 75SIRT1 into HeLa cells, either with or without TNFα stimulation, and compared their degree of cleavage by cathepsin B. Although the stimulated flSIRT1-transfected cells demonstrated increased levels of the 75SIRT1 variant, as compared with those of untreated cells, the stimulated ΔSIRT1-transfected cells did not show such an increase, indicating that the mutant is resistant to cleavage under TNFα stimulation (Fig. 3B). Additionally, the levels of ectopically transfected ΔSIRT1, flSIRT1 or 75SIRT1 were not affected by the stimulation (Fig. 3B). These findings were confirmed in HEK293 cell lines, which were treated with TNFα or GB111-NH2 and TNFα to simulate untreated cells. HEK293 cells similarly showed reduced cleavage of ΔSIRT1 following TNFα stimulation (Fig. 3C).
Next, we co-incubated the purified flSRIT1 or ΔSIRT1 with cathepsin B and measured the time course of cleavage. Cathepsin B cleaved flSIRT1 to produce 75SIRT1 15 min post-incubation (Fig. 4A), whereas ΔSIRT1 cleavage was not observed at 30 min (Fig. 4B). In agreement with these findings, examining longer incubation periods revealed that the levels of 75SIRT1 were significantly higher in flSIRT1 than in ΔSIRT1 after 60 min (Fig. 4C). The data were also quantified as the ratio between 75SIRT1 and the 110 kDa SIRT1 (which could be either flSIRT1 or ΔSIRT1) (Fig. 4D).
Deacetylase activity is maintained in the ΔSIRT1 mutant
To assess whether the ablation of amino acids 528–543 of SIRT1 affects its deacetylase activity, possibly triggering aberrant changes in the acetylation of SIRT1 histone and nonhistone targets, we tested the in vitro deacetylase activity of flSIRT1, 75SIRT1, ΔSIRT1, a commercial recombinant SIRT1 (rSIRT1) positive control and a no-enzyme negative control. This analysis revealed that the activity of 75SIRT1 is similar to that of the no-enzyme negative control, whereas that of flSIRT1 and ΔSIRT1 is higher than that of the negative control and similar to that of the positive control, rSIRT1 (Fig. 5A). Next, we analyzed the deacetylase activity of flSIRT1, 75SIRT1 and ΔSIRT1 in HEK293 cells under TNFα stimulation by quantifying H3K9/14 and p65 (also known as RELA) acetylation levels (Yeung et al., 2004; Blander and Guarente, 2004) (input presented in Fig. 5B). This analysis indicated that both flSIRT1 and ΔSIRT1 are enzymatically active to a similar degree under these conditions (Fig. 5C,D). However, in contrast to cells overexpressing flSIRT1 or ΔSIRT1, cells overexpressing 75SIRT1 demonstrated a 20% increase in H3 acetylation and a twofold increase in p65 acetylation, indicating that 75SIRT1 is enzymatically inactive, as previously reported for chondrocytes and cartilage (Elayyan et al., 2017; Oppenheimer et al., 2012). Of note, pcDNA (expression vector) controls possess similar substrate acetylation states to those presented for flSIRT1 and ΔSIRT1, which might be related to possible saturation of the deacetylation activity beyond substrate availability. On the other hand, the reduced deacetylase activity of 75SIRT1-overexpressing cells versus pcDNA controls could be a result of the capacity of 75SIRT1 to interrupt the overall SIRT1 homotrimer activity (Vaquero et al., 2004), or alter the complex's nuclear localization (Tanno et al., 2007; Flick and Lüscher, 2012), as previously reported in human chondrocytes (Oppenheimer et al., 2012).
The binding of cytochrome C to 75SIRT1 is more frequent than its binding to ΔSIRT1
To determine cytochrome C (CytoC) binding to flSIRT1 and 75SIRT1, we used a hypothetic structural analysis via PyMOL software. Despite their identical sequences, the superimposition of the structures of flSIRT1 and 75SIRT1 revealed a root-mean-square deviation (RMSD) of 19Å in their N-terminal region (Fig. 6A), which might accommodate a structural rearrangement favoring distinct binding partners, such as CytoC (Oppenheimer et al., 2012). Indeed, as shown in Fig. 6B (top) (and in SD9 and SD10 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973), the docking of flSIRT1 to CytoC presented fewer potential hydrogen bonds (namely, amino acids C17, Y48, Y67, N71, K73, K74, K80 and I82 of CytoC are predicted to form hydrogen bonds with amino acids D124, L147, G149, E151, T154 and E214 of flSIRT1) than the docking of 75SIRT1 to CytoC (Fig. 6B, bottom; SD7 and SD8 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973) (namely, amino acids Q17, K28, T28, K73, K80, I82, and K87 of CytoC are predicted to form hydrogen bonds with amino acids in positions D127, E132, E133, D150, I153, T154, N155, S159, E161 and E165 of 75SIRT1). In addition, amino acid F82 of CytoC is predicted to form a van der Waals interaction with amino acid F157 of 75SIRT1.
Previous reports demonstrated an enhanced binding capacity of 75SIRT1 to CytoC in chondrocytes (Oppenheimer et al., 2012). Here, we opted to quantify this binding using in-gel mass spectroscopy and to compare the frequency of the binding of CytoC to 75SIRT1 to its binding to the ΔSIRT1 mutant, which is resistant to cathepsin B cleavage. To this end, HEK293 cells were transfected with either 75SIRT1 or ΔSIRT1 and treated with TNFα for 24 h. Next, the cytoplasmic protein extract was isolated and immunoprecipitated for Flag-tag, which was fused to the two ectopically transfected proteins. The Flag immunoprecipitants were analyzed using in-gel mass spectroscopy, and the peptide ratio of SIRT1 and CytoC was calculated between 75SIRT1- and ΔSIRT1-transfected cells (Table 1; SD11 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973). The findings of these analyses confirmed that CytoC and SIRT1 were more abundant in the cytoplasmic extract of 75SIRT1-transfected and TNFα-stimulated cells, as compared with that of ΔSIRT1-transfected and TNFα-stimulated cells, which is in line with our previous report in human chondrocytes (Oppenheimer et al., 2012).
ΔSIRT1 expression enhances caspase 9 activity and apoptosis
We previously showed that 75SIRT1 binds CytoC to ultimately prevent cell apoptosis (Oppenheimer et al., 2012). In that study, we speculated that this 75SIRT1–CytoC binding prevents apoptosis by sequestering CytoC from assembling an active apoptosome (which contains CytoC, APAF1 and procaspase 9) and hindering the subsequent activation of caspase 9. Here, to test this speculation, we examined apoptosis in HeLa cells overexpressing either flSIRT1, ΔSIRT1 or 75SIRT1, and stimulated with TNFα, as compared with cells expressing a pcDNA, as a control, under the same conditions. A fluorescence-activated cell sorting (FACS) analysis revealed that the number of annexin-positive populations was significantly lower in cells overexpressing 75SIRT1 than in cells overexpressing ΔSIRT1, but was similar to that in control cells expressing pcDNA (Fig. 7A). In addition, after treating all cell cultures with a cathepsin B inhibitor to eliminate any residual cathepsin activity under the TNFα stimulation, the levels of apoptosis were similar in cells overexpressing ΔSIRT1 and in those overexpressing flSIRT1 (Fig. 7A).
Because the HEK293 cell lines showed similar SIRT1 cleavage and cathepsin B activity to that of HeLa cells, we next overexpressed Flag-tagged 75SIRT1, flSIRT1 or a pcDNA control in HEK293 cells and immunoprecipitated CytoC to analyze its ability to bind either 75SIRT1 or flSIRT1 (Fig. 7B). This analysis revealed an increased Flag/CytoC ratio in 75SIRT1-overexpressing cells (Fig. 7B, bottom graph), indicating increased 75SIRT1–CytoC binding in 75SIRT1-overexpressing cells compared with flSIRT1-overexpressing cells. Notably, APAF/CytoC binding was increased in flSIRT1-overexpressing cells, as compared with 75SIRT1-overexpressing cells, indicating augmented apoptosome assembly in the former. As increased CytoC binding to 75SIRT1 might mitigate the activation of caspase 9, we also assessed whether the ability of 75SIRT1 to bind CytoC (Fig. 7B) results in aberrant caspase 9 activity. As expected, 75SIRT1-overexpressing cells demonstrated reduced caspase 9 activity compared with flSIRT1- or ΔSIRT1-overexpressing cells (Fig. 7C). Although 75SIRT1-overexpressing cells presented the lowest caspase 9 activity rates, these rates were not significantly different from those in the pcDNA controls. A possible explanation for this observation is that the pcDNA controls also underwent endogenous SIRT1 cleavage and, therefore, expressed a certain degree of 75SIRT1, which might bind CytoC and reduce caspase 9 activation. In addition, as the rate-limiting factor is cytoplasmic CytoC, the overexpression of 75SIRT1 might not necessarily enhance survival, as compared with that of pcDNA controls. Overall, our data indicate that 75SIRT1 plays a crucial role in preventing cell death by impairing the activity of caspase 9 under pro-inflammatory stress conditions.
The deacetylase SIRT1 has been established as a highly networked enzyme that participates in numerous critical regulatory circuits, including energy expenditure, chromatin integrity and gene regulation (McBurney et al., 2013). The capacity of cells to fine-tune SIRT1 activity is, therefore, crucial for maintaining a functional tissue, especially because SIRT1 activity is often reduced in various age-related diseases (Bordone et al., 2007; Hall et al., 2013). Indeed, our analysis of aging organs proves that SIRT1 cleavage to 75SIRT1 is enhanced with age in many tissues, but these tissues also presented variations in the molecular weight of the cleaved SIRT1. Although these aberrant 75 kDa N-terminally intact SIRT1 variants could indicate that various other cathepsins cleave SIRT1 in a tissue-specific manner, they might arise as a result of the inability of cathepsin B to identify a specific amino acid sequence prior to cleavage. Thus, we hypothesized that the cleavage of SIRT1 is not exclusively dependent on amino acid recognition by cathepsin B; rather, based on in silico modeling, we found a C-terminal unstructured motif, which we show is essential for cathepsin B docking onto and cleavage of SIRT1.
By removing a sequence of 15 amino acids (i.e. amino acids 528–543) from the predicted C-terminal loop domain (spanning amino acids 522–550), we prevented the cleavage of SIRT1 (i.e. ΔSIRT1 mutant). The ΔSIRT1 mutant maintained the amino acid region of 631–655 (in proximity to the C-terminal domain), which had previously been reported to be essential for SIRT1 activity (Kang et al., 2011). Therefore, we expected this ΔSIRT1 mutant to maintain its deacetylase activity at levels comparable to those of flSIRT1, as we indeed validated in this report. On the other hand, a study by Ghisays and colleagues (Ghisays et al., 2015) showed that the N-terminal domain of SIRT1 is also crucial for its deacetylation activity. However, in our study, the N-terminally intact 75SIRT1, which was overexpressed in vivo, showed reduced deacetylase activity, collectively indicating that both the N-terminus and the C-terminus are essential for SIRT1 activity.
The ΔSIRT1 mutant was relatively resistant to cathepsin B cleavage, suggesting that cathepsins are likely to target accessible structural motifs rather than cleaving in a sequence-dependent manner. This structural preference of cathepsin B when targeting SIRT1 for cleavage can span the range of amino acids 522–550 (a predicted unstructured loop domain), thereby generating multiple N-terminally intact polypeptides of SIRT1 within the 75 kDa range. Further support for this protein domain being unstructured is the lack of structural (diffraction) data for this domain in existing crystal structures. For example, in the crystal structures of SIRT1 that were solved without bound substrates [e.g. Protein Data Bank (PDB) 4IG9 and 4KXQ, amino acids 234–510], the structural data for the 522–550 amino acids region is, to date, absent (based on a PDB search). However, because our study is based on modeling rather than on experimental structural analysis, we cannot conclude in full confidence that this region is indeed an unstructured loop motif.
On a functional level, ΔSIRT1 maintained its deacetylase activity similar to flSIRT1 and demonstrated reduced CytoC binding compared with 75SIRT1. These results are similar to those from previous reports that compared the binding of flSIRT1 and 75SIRT1 to CytoC (Oppenheimer et al., 2012). Additional support for the functional similarity of ΔSIRT1 to flSIRT1 was obtained when analyzing cell death as a result of TNFα stimulation, which was higher in cells expressing flSIRT1 and ΔSIRT1 than in those expressing 75SIRT1. Further analysis showed that cells expressing either flSIRT1 or ΔSIRT1 have higher caspase 9 activity than those expressing 75SIRT1, in support of the notion that the cleavage of SIRT1 interferes with the activity of the apoptosome under pro-inflammatory stimuli. Given these data and those of aging tissues, we speculate that an age-associated accumulation of 75SIRT1 might contribute to cell senescent phenotypes by prolonging survival, but it might also promote a pro-inflammatory response facilitated by hyperacetylated RELA, as we show for the first time in 75SIRT1-expressing cells.
Taken together, the results obtained in this study not only shed light on the biochemical properties of 75SIRT1, they also establish that its generation is not a tissue-specific event and that variations in the cleaved SIRT1 are likely caused by the capacity of cathepsin B to dock within the C-terminal loop site of SIRT1. Based on our observations, pharmacological targeting of this predicted loop site might yield SIRT1 proteins that are more stable to cathepsin B cleavage without interfering with SIRT1 deacetylase activity.
MATERIALS AND METHODS
All experimental procedures involving mice were conducted according to the National Institutes of Health Animal Research Advisory Committee (ARAC) guidelines, based on the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC). The Institutional Animal Care and Use Committee of the Hebrew University of Jerusalem approved the study protocol (MD-12-13383-4), in which 129-SIRT1tm1Mcby mice (RIKEN BioResource Center) were bred with CD1 mice, and SIRT1 heterozygote mice were then inbred to generate WT SIRT1 mice. The mice were allowed to reach the age of either 3 or 16 months under a 12-h light/dark cycle and with food and water provided ad libitum. Both male and female mice were killed by cervical dislocation following ketamine or xylazine anesthesia (200 mg/kg).
Cell cultures and transfections
HEK293 and HeLa cells were cultured in Dulbecco's modified Eagle medium (Sigma-Aldrich) containing 10% fetal calf serum (FCS) and 2% penicillin-streptomycin (Biological Industries Beit-HaEmek). A Polyjet transfection reagent (SignaGen Laboratories) was used according to the manufacturer's instructions. To induce pro-inflammatory stimuli, the growth medium was replaced and incubated for 24 h with a defined Bio-MPM-1 medium (05-060-1A, Kibbutz Beit-HaEmek) supplemented with TNFα (50 ng/ml; PeproTech Asia). To inhibit cathepsin activity, cells were pretreated with the pan-cathepsin inhibitor GB111-NH2 (2 µM for 24 h), a kind gift from Dr Galia Blum (School of Pharmacy, Hebrew University of Jerusalem, Jerusalem, Israel). Alternatively, to selectively inhibit cathepsin B, cells were treated with a final concentration of 2 µM CA-074me for 24 h (A2S Technologies). Cell extracts were analyzed for caspase activity using the colorimetric Caspase 9 Assay Kit (ab65608, Abcam), following the manufacturer's instructions.
Antibodies and reagents
The following primary antibodies were used for immunoblotting (IB) and immunoprecipitation (IP): anti-SIRT1 (07-131, Millipore; IB 1:1000), anti-Flag (sc-807, Santa Cruz Biotechnology; IB: 1:2000, IP 1:1000)), anti-cathepsin B (ab58802, Abcam; IB 1:1000), anti-β-tubulin (T5168, Sigma-Aldrich; IB 1:3000), anti-actin (sc-477718, Santa Cruz Biotechnology; IB 1:3000), anti-GAPDH (CST-2118, Cell Signaling Technology; IB 1:2000), anti-histone H3 (06-755, Millipore; IB 1:1000), anti-acetyl-histone H3 (07-360, Millipore; IB 1:1000), anti-p65 (ab16502, Abcam; IB 1:1000) and anti-p65_AcK310 (ab52175, Abcam; IB 1:1000). The following secondary antibodies were used (1:2500 dilution): anti-mouse-alkaline phosphatase (AP) conjugated (A3562, Sigma-Aldrich), anti-rabbit-AP conjugated (A3687, Sigma-Aldrich) and anti-goat-AP conjugated (A2168, Sigma-Aldrich). Anti-SIRT1 and anti-Flag showed a linear response for diluted purified protein substrates (data not shown).
Generating SIRT1 mutant constructs
pcDNA-Flag-His-SIRT1 (i.e. flSIRT1, under a CMV promoter) was a kind gift from Prof. Danny Rienberg, University of New York, New York, USA. Based on this construct, we designed 75SIRT1 (containing the 1–1602 bp of the SIRT1 gene) by using the forward primer (FP) 5′-GCAGAGCTCTCCCTATCAGTGATAGAGATC-3′ from the plasmid backboned and the reverse primer (RP) 5′-CCACTTGAACTCAGAAAGATCTGTACTTCTCCACACCC-3′, with a point mutation in the RP to generate the XbaI restriction site at the 1602 bp position and a stop codon at the 1604 bp position. This construct expressed 75SIRT1 corresponding to amino acids 1–534 of the translated SIRT1 protein. A deleted SIRT1 construct (ΔSIRT1) was designed by eliminating a portion of the predicted unstructured loop site of the SIRT1 protein domain (amino acids 528–543). The ΔSIRT1 clone was generated in two steps. In the first step, we generated a 1–1581 bp SIRT1 construct, possessing the SacI restriction site at the 1575 bp position, by using the FP 5′-GCAGAGCTCTCCCTATCAGTGATAGAGATC-3′ from the plasmid backbone and the RP 5′-AGGTGTGGGTGGGAGCTCTGACAAATAA-3′. In the second step, we ligated the 1–1581 bp SIRT1 construct to a synthetic construct from 1615 bp to 2241 bp (SIRT1; Syntezza Bioscience), possessing the SacI restriction site at the 1609 bp position, and XbaI at the 2242 bp position of the synthetic construct. Following ligation selection and sequence validation, a ΔSIRT1 lacking amino acids 528–538 was generated and further extended to amino acids 528–543 using a site-directed mutagenesis kit (Q5 Site-Directed Mutagenesis Kit, NEB) 5′-ACTTCACCACCAGATTCTTC-3′ and RP 5′-ACTGAGCTCTGACAAATAAG-3′. The resulting mutant, ΔSIRT1, lacked the amino acid sequence PPTPLHVSEDSSSPE (528–543 in the flSIRT1 WT protein) and harbored an R-to-S mutation in amino acid 528.
In silico structural modeling
Protein sequence information for the human flSIRT1 (NP_036370; PubMed protein database) was used to generate the flSIRT1 structure by homology modeling (SD1 and SD2 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973) employing the Robetta server (Baker Laboratory; robetta.bakerlab.org) and PDB structures 1J8F (human SIRT2 histone deacetylase), 3K5W (carbohydrate kinase, Helicobacter pylori) and 2LCI-P (loop NTPASE fold, synthetic construct) as a template. The final flSIRT1 structure was selected based on Ramachandran plot criteria. Similarly, a homology model of the human 75SIRT1 structure (amino acids 1–534; SD3 and SD4 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973) was constructed using the structure of Saccharomyces cerevisiae SIRT2 deacetylase (PDB 2HJH) as a template.
The docking of cathepsin B (PDB 2FRQ) to flSIRT1 was modeled, after eliminating water and ligand molecules from the cathepsin B structure, using the ClusPro 2.0 server (http://cluspro.bu.edu/). Spatial constraints (advanced mode) were applied to direct docking of amino acids 530–540 of flSIRT1 to the active site of cathepsin B. This binding mode was suggested based on our previous study (Dvir-Ginzberg et al., 2011). The ten top-ranked (based on the ClusPro energetic scoring function) docked complexes were manually analyzed using PyMOL (http://www.pymol.org/) and the docking mode, in which H533 residue of cathepsin B was part of the flSIRT1 binding interface, was selected (SD5 and SD6 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973).
Unbiased docking of CytoC to flSIRT1 or 75SIRT1 was performed using the structural coordinates of CytoC (PDB 3ZCF; note that the first methionine is missing from this file), as previously described for cathepsin B (SD7 and SD8 in https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973).
Immunoblot and immunoprecipitation
Harvested cell pellets were suspended in RIPA buffer. To prevent protease activity, all lysates were supplemented with a protein inhibitor cocktail (cOmplete, Roche), 1 mM phenylmethylsulfonyl fluoride (Sigma-Aldrich), and 1 µg/ml ALLN (Sigma-Aldrich). To inhibit protein deacetylation, the cell lysates were additionally supplemented with 5 mM sodium butyrate (Millipore), 1 µM Trichostatin A (Sigma-Aldrich) and 7 mM nicotinamide (Sigma-Aldrich). Immunoprecipitation experiments were conducted using protein extracts diluted in a low-stringency buffer (10 mM HEPES, 150 mM NaCl, 5% glycerol, pH 7.4) supplemented with the above-indicated inhibitors. Cell lysates were incubated overnight with 2 μg of antibodies and 50 μl of protein-A/G beads (Santa Cruz Biotechnology). Precipitated immunocomplexes were eluted with a sample buffer (161-0791, Bio-Rad) and subjected to immunoblotting. For immunoblotting, protein extracts were run on standard SDS-PAGE gels (8%, 10% or 15%), followed by transfer onto a polyvinylidene fluoride membrane (162-0177, Bio-Rad). The membranes were blocked in 3.5% nonfat milk in PBS and incubated with primary antibodies, as suggested by the manufacturer. Following three consecutive washes with PBS-Tween 20 (0.1%), the membranes were incubated with a secondary antibody, as indicated above. ImageJ (https://imagej.nih.gov/ij/) was used for semiquantification of band intensity in all immunoblots, and the acquired values were normalized to those of loading controls.
For on-bead digestion, samples were subjected to in-solution, on-bead, tryptic digestion. The proteins were first reduced by incubation with dithiothreitol (5 mM; Sigma-Aldrich) for 30 min at 60°C and alkylated with 10 mM iodoacetamide (Sigma-Aldrich) in the dark for 30 min at 21°C. The proteins were then subjected to digestion with trypsin (Promega) for 16 h at 37°C. The digestions were stopped by trifluroacetic acid (1%). Following digestion, peptides were desalted using solid-phase extraction columns (Oasis HLB). Data from the on-bead samples were imported into Expressionist software (Genedata) for retention time alignment and peak detection of precursor peptides. A master peak list was generated from all MS/MS events and sent for database searching using Mascot v2.5 (Matrix Sciences). The data were searched against the human protein sequences from UniprotKB (http://www.uniprot.org), appended with 125 common laboratory contaminant proteins. Proteomic experiments and analyses were performed in the Israel National Center for Personalized Medicine (INCPM), Weizmann Institute of Science, Rehovot, Israel.
Protein purification and in vitro cleavage assays
For the in vitro cleavage study, flSIRT1 and ΔSIRT1 were overexpressed in HEK293 cells for 36 h. The cells were lysed in a buffer containing 20 mM Tris-HCl, 250 mM NaCl, 10 mM imidazole (pH 8.0) and a cocktail of protease inhibitors, and then subjected to three freeze-thaw cycles and dounce homogenization. The clear supernatant, containing His-tag proteins, was loaded onto the Ni-NTA column, and the bound proteins were eluted with a gradient of imidazole in an elution buffer (20 mM Tris-HCl, 250 mM NaCl, 250 mM imidazole, 10% glycerol, pH 7.2). The purity of the eluted fractions was examined by immunoblotting on a polyacrylamide gel. Pure protein fractions were pooled together and concentrated using a 30 kDa cutoff amicon filter (EMD Millipore). The purified SIRT1 protein (2 µg) was incubated with 0.25 units (219362, Calbiochem) or 0.3 µg/ml (557704, BioLegend) of recombinant human cathepsin B in a 0.1 M Tris-HCl buffer (pH 7.2) at 37° C for various time points, as indicated in Fig. 4. To examine SIRT1 activity, 1.6 µg of the purified SIRT1 protein (flSIRT1, ΔSIRT1 or 75SIRT1) was incubated with a fluoro-substrate peptide (0.2 mM) and fluorescence was measured (excitation 355 nm, emission 460 nm) at predetermined time points, according to manufacturer guidelines (CycLex SIRT1/SIRT2 Deacetylase-Fluorometric Assay Kit, MBL International).
FACS analyses were conducted using an MEBCYTO Apoptosis kit (MBL International), according to the manufacturer's instructions. Briefly, following treatment, the supernatant and adherent cells were collected, centrifuged (7600 g, 10 min), and washed twice with PBS. The collected cells were adjusted to 0.1×106/ml in the FACS buffer supplied in the kit. The cell suspension was then stained with 5 µl fluorescein isothiocyanate (FITC)-labeled annexin V. Cells subject to 5 µg/ml etoposide for 24 h served as a positive control for the FACS analysis (BD Accuri C6 Plus, BD Biosciences). The data were collected and analyzed using FACScan software (Becton Dickinson).
All experiments were repeated (n>3) and tested for statistical significance amongst paired treatments using the Mann–Whitney test, assuming a confidence level of 95% (P<0.05). Differences between multiple experimental groups were tested using the Kruskal–Wallis test, assuming a confidence level of 95% (P<0.05) to be statistically significant. Error bars in Figs 4D, 5A and 7A–C indicate the standard deviation of the mean value of a given data point. In plots showing individual data points (Figs 1A,B, 2A, 3A–C and 5C,D), black bars indicate the mean value of the individual points.
We thank the INCPM proteomics core. GB111-NH2 was a kind gift from Dr Galia Blum (School of Pharmacy, Hebrew University of Jerusalem). pcDNA-hSIRT1 plasmid was a kind gift from Prof. Danny Reinberg, NYU.
Conceptualization: A.K., Y.D., L.B.-A., O.Q., E.R., G.B., Y.H.M., M.D.-G.; Methodology: A.K., Y.D., L.B.-A., O.Q., J.E., E.R., G.B., Y.H.M., S.E., M.D.-G.; Software: A.K., L.B.-A., G.B., Y.H.M., S.E.; Validation: A.K., Y.D., L.B.-A., O.Q., J.E., E.R., G.B., Y.H.M., S.E., M.D.-G.; Formal analysis: A.K., Y.D., L.B.-A., O.Q., J.E., E.R., G.B., Y.H.M., S.E., M.D.-G.; Investigation: A.K., Y.D., L.B.-A., O.Q., J.E., E.R., G.B., Y.H.M.; Resources: A.K., M.D.-G.; Data curation: A.K., Y.D., L.B.-A., O.Q., J.E., E.R., G.B., S.E., M.D.-G.; Writing - original draft: A.K., L.B.-A., O.Q., E.R., Y.H.M., S.E., M.D.-G.; Writing - review & editing: A.K., Y.D., O.Q., E.R., Y.H.M., S.E., M.D.-G.; Visualization: A.K., Y.D., O.Q., M.D.-G.; Supervision: A.K., J.E., E.R., S.E., M.D.-G.; Project administration: E.R., M.D.-G.; Funding acquisition: M.D.-G.
This work was supported by the Israel Science Foundation [121/12 and 370/17 to M.D.-G.], Rosetrees Trust [M447 to M.D.-G.], Planning and Budgeting Committee of the Council for Higher Education of Israel [Postdoctoral Scholarship to A.K.] and Rosa Luxemburg Stiftung [Scholarship to Y.D.].
Detailed modeling, docking, validation and mass spectrometry data supporting this work are available at https://figshare.com/articles/Kumar_et_al_2018_-_JCS/6730973 (SD1–SD11).
The authors declare no competing or financial interests.