Ex-527 inhibits Sirtuins by exploiting their unique NAD⁺-dependent deacetylation mechanism

Ex-527 Is a Selective Sirtuin Inhibitor and Requires NAD⁺ for Inhibition.

A first kinetic analysis of Sirt1 inhibition by Ex-527 (27, 28) was done with the Fluor-de-Lys (FdL) substrate, a peptide carrying a fluorophore that potentially causes artifacts (12, 30). To investigate the molecular inhibition mechanism, we first tested selectivity and kinetics using a continuous assay (31) and nonmodified peptides derived from physiological substrates for Sirt1 (p53), Sirt3 [acetyl-CoA synthetase 2 (ACS2)], and Sirt5 [carbamoyl phosphate synthethase 1 (CPS1)]. Because inhibition was proposed to be uncompetitive with NAD⁺ (27), we adjusted NAD⁺ concentrations according to the respective K_M values to allow comparisons. The IC₅₀ values are 0.09 ± 0.03 µM for Sirt1 and 22.4 ± 2.7 µM for Sirt3 (Fig. 1C), in agreement with the FdL values (0.1 and 49 µM, respectively) (27). Because Sirt1 crystals became available only recently, we included the bacterial homolog Sir2 from Thermotoga maritima (Sir2Tm) in our investigation. Sir2Tm was efficiently inhibited by Ex-527 (IC₅₀ 0.9 ± 0.3; Fig. 1C), and we thus used it as a representative of the potently inhibited Sirtuins for structural studies. Furthermore, Ex-527 had no pronounced effect on Sirt5-dependent deacetylation, consistent with FdL tests (28), and showed no inhibition of Sirt5-dependent desuccinylation (Fig. 1D), the dominant Sirt5 activity identified recently (32).

To identify the enzyme state recognized by Ex-527, and thus suitable ligands for cocrystallization, we analyzed inhibition kinetics. Activity assays in presence of varying Ex-527 concentrations showed that inhibition of Sir2Tm and Sirt3 by Ex-527 is noncompetitive with substrate peptide (Fig. 1 E and F; Fig. S1A; Sir2Tm K_i = 1.8 ± 0.4 µM, Sirt3 K_i = 33.4 ± 4.4 µM). Analogous assays for the cosubstrate NAD⁺ showed no competition (Fig. 1 G and H; Fig. S1A), but for both Sirtuins inhibition was uncompetitive (Fig. S1 B–G; Sir2Tm: K_i = 3.3 ± 0.4 µM, Sirt3 K_i = 31.3 ± 2.1 µM), indicating that NAD⁺ assists in inhibition. These results are consistent with FdL data on Sirt1 (27) and show that Ex-527 inhibits the potent targets Sirt1/Sir2Tm, as well as the less sensitive Sirt3, with the same, NAD⁺-dependent and apparently peptide-independent mechanism.

The S Enantiomer of Ex-527 Occupies the C-Pocket and Interacts with ADP Ribose.

The presence of peptide substrate appears optional for Ex-527 binding, but NAD⁺ acts as an essential coligand. Because no Sir2Tm crystals could thus far be obtained in the absence of peptide, we used the following strategy to obtain the high-affinity Sir2Tm/inhibitor complex: we soaked crystals of a Sir2Tm/acetylated peptide complex, obtained in the presence of 1.5 mM Ex-527, with 1 mM NAD⁺ to enable cosubstrate binding and subsequent inhibitor association. Stopping the soak by freezing crystals after 2 min yielded good diffraction data (1.9 Å resolution; Table 1) for a Sir2Tm complex that contained Ex-527 and NAD⁺ derivatives (see below). Human Sirt3, in contrast, readily crystallizes in different states (12, 13), and we thus added Ex-527 to the running reaction of Sirt3, NAD⁺, and acetylated substrate before crystallization. Well-diffracting crystals were obtained (2.0 Å resolution; Table 1) that revealed Ex-527 and a coligand in the NAD⁺ pocket (see below).

The overall structures of the Ex-527 complexes of Sir2Tm and Sirt3 (Fig. 2A) differ only in the cofactor binding loop conformations (see below) from previous structures of these proteins (11–13, 16). In the Sir2Tm/Ex-527 complex, cocrystallized peptide and the ADP ribose part of the soaked-in NAD⁺ are visible. The electron density reveals a mixture of ligands due to the dynamic processes started by NAD⁺ soaking. Density analysis and rerefinement with combinations of substrates, products, intermediate, and inhibitor (Fig. S2 A–H) revealed the following complexes: an Ex-527 molecule is clearly indicated by electron density, with a C-ring and carbamide group in the C-pocket (Fig. 2 B and C). Part of the pocket, however, is occupied by nicotinamide from NAD⁺ that has not reacted and thus has acetylated peptide as coligand (40% occupancy). Coligands for Ex-527 are the products deacetylated peptide and 2′-O-acetyl-ADP ribose (60% occupancy for all three ligands), which provides part of the binding site (Fig. 2C). To confirm that products and inhibitor can form a complex with Sir2Tm, we then crystallized Sir2Tm mixed with deacetylated peptide, 2′-O-acetyl-ADP ribose, and Ex-527. The obtained structure (3.3 Å resolution; Table 1) revealed Ex-527 and 2'-O-acetyl-ADP ribose (Fig. 2D) arranged as observed in the soaked complex and partial occupation (∼50%) for the deacetylated peptide. In addition to its interaction with the coproduct, Ex-527 binds with its cyclohexenylcarbamide in the Sir2Tm C-pocket (Fig. 2 B–E), mimicking nicotinamide binding (33). The density reveals exclusively the S enantiomer of the Ex-527 racemic mixture as ligand, consistent with the finding that only one stereoisomer, called Ex-243, acts as inhibitor and that the S-form is the active isomer of a related compound (27–29). Only the kinked C-ring of the S isomer allows simultaneous positioning of the carbamide in the C-site and the aromatic rings in an almost perpendicularly oriented pocket formed by Gln98, Ile159, Phe48, and His116. We propose the term extended C-site (ECS) inhibitors for ligands exploiting both pockets. The environment of the Ex-243 chlorine is mainly nonpolar (Phe48 and Ile159), but the Val160 backbone carboxyl acts as a polar interaction partner for the chlorine, possibly forming a halogen bond (distance 3.6 Å, 117° angle to the carbonyl).

In contrast to the Sir2Tm/Ex-243 complex, the Sirt3 complex comprises a complete NAD⁺ molecule in addition to the well-defined inhibitor, and no density for a peptide (Fig. 2F; Fig. S2 I and J). The density again identifies the Ex-527 S enantiomer, Ex-243, in the C-site and the neighboring hydrophobic pocket (Fig. 2F; Fig. S2K). In addition to the contact to NAD⁺-derived ribose, the Ex-243 A-ring interacts with the nicotinamide of NAD⁺, which is in a nonproductive, extended conformation placing the nicotinamide in the acetyl-Lys binding site (Fig. S2L), as recently described for a Sirt1/NAD⁺/Ex-527 complex (29). Because Ex-243 binds to a binary Sirt3/NAD⁺ complex, we cocrystallized these partners in the absence of peptide (2.0 Å resolution; Table 1). Under these crystallization conditions, Ex-243 bound to a complex of Sirt3 and ADP ribose (Fig. 2G), likely formed from NAD⁺ through nonenzymatic hydrolysis, showing that even the ribose moiety of ADP ribose is sufficient for generating an Ex-243 binding site. We conclude that effective Ex-243 binding requires the cosubstrate binding pocket to be occupied, but our structures indicate two possibilities for the Ex-243 coligand under physiological conditions: NAD⁺ and 2′-O-acetyl-ADP ribose.

Ex-243 Binds Preferentially After Intermediate Formation and Allows Product Formation.

To identify the major Ex-243 coligand in solution, we performed binding experiments with Sirt3 and Sir2Tm using microscale thermophoresis (Fig. 3 A and B). Ex-527 showed negligible affinities (Sir2Tm: K_d > 180 μM; Sirt3 K_d > 330 µM) to apo-enzymes and to the Sirtuins saturated with 1 mM peptide substrate (Sir2Tm: K_d > 170 μM; Sirt3: K_d > 180 µM). Adding saturating NAD⁺ concentrations (5 mM), however, improved Ex-527 binding, with K_d values of 6.0 ± 0.4 µM for Sir2Tm and 16.5 ± 2.9 µM for Sirt3. Strikingly, addition of both substrates resulted in even higher affinities, with K_d values of 4.9 ± 0.5 µM (Sir2Tm) and 10.0 ± 1.4 µM (Sirt3). Thus, peptide substrate alone has no influence on Ex-243 binding, but it promotes tighter binding in presence of the cosubstrate, which enables alkylimidate formation. These results and the inhibition kinetics (see above) show that the arrangement in the Sirt3/NAD⁺/Ex-243 complex is not relevant for inhibition, because the inhibitor forces nicotinamide in the acetyl-Lys pocket, which would lead to competition between inhibitor and both substrates (Fig. S2L). Instead, these data suggest that Ex-243 binding after intermediate formation dominates inhibition, which is supported by the observation that nicotinamide release foregoes Sirt1 inhibition by Ex-243 (27).

Our Sir2Tm structures suggest that Ex-243 inhibits by stabilizing a Sirtuin complex with coproduct. To confirm the relevance of this coproduct complex for inhibition in solution, we analyzed product formation at saturating Ex-243 concentrations. Varying enzyme amounts were tested, and deacetylated peptide was quantified by MS. We find that for Sirt1, Sir2Tm, and Sirt3, product is indeed formed, but only up to one product molecule per Sirtuin molecule (Fig. 3 C–E; Fig. S3A). A control experiment with Sirt3 and the adenosine analog and potent Sirtuin inhibitor Ro-31-8220 [IC₅₀ Sirt1 3.5 ± 0.4 µM (34); IC₅₀ Sirt3 3.7 ± 0.7 µM, this study] yielded no product. We thus conclude that Ex-243 inhibition, in contrast to more common inhibition mechanisms, involves binding after intermediate formation and allows generation of the coproduct 2′-O-acetyl-ADP ribose.

Crystal Structure of a Native Intermediate Complex and Incompatibility with Ex-243 Binding.

Ex-243 might either target Sirtuin complexes with alkylimidates, with subsequent product formation, or it could directly bind to the coproduct complex. As a first test for an intermediate/Ex-243 interaction, we generated Sirt3 complexes with a more stable thio-analog of the alkylimidate (13, 35). Conformation and position of the thio-alkylimidate in the absence (2.05 Å resolution; Table 1; Fig. S3B) and presence of Ex-243 (2.7 Å resolution; Table 1; Fig. S3C) were indistinguishable (Fig. S3D). Also in presence of Ex-243, the inhibitor was not bound, and the C-pocket instead occupied by water molecules. To exclude the possibility that the artificial thio-modification causes the incompatibility of intermediate and Ex-243, we next generated a Sirt3 complex with native alkylimidate through soaking of Sirt3/acetyl-peptide crystals with a mix of NAD⁺ and Ex-243. The obtained Sirt3 complex with native alkylimidate (2.5 Å resolution; Table 1; Fig. 3F; Fig. S3E) is strong experimental evidence for this intermediate in Sirtuin-dependent deacetylation. It shows cis-configuration between Lys-ε-C-N and the glycosidic bond, but in contrast to NAD⁺ in productive conformation and the thio-intermediates with longer C-S bonds described above and in previous publications (13, 35), the ribose is flipped around (Fig. 3G). This feature of the native alkylimidate suggests that ribose rotation (and associated rearrangements in the diphosphate) enables moving the ribose C1′ from the C site to the acetyl-O for bond formation. However, this alkylimidate complex also did not contain bound Ex-243. Modeling Ex-243 into the C-pocket of the Sirt3/native intermediate complex by superposition with the Sirt3/NAD⁺/Ex-243 structure reveals a clash between inhibitor and the alkylimidate methyl group as molecular basis for the incompatibility (1.1 Å; Fig. 3G).

The second cyclic intermediate could also serve as a Ex-243 receptor, followed by product formation and trapping. If Ex-243 instead binds directly to the product complex, then product and inhibitor should increase each other’s affinity. Testing binding of 2′-O-acetyl-ADP ribose required >50 mM coproduct for signals, however, which is a concentration range where nonspecific effects prevent reliable quantification. The fact that we were able to assemble and crystallize a Sir2Tm/Ex-243/product complex at moderate 2′-O-acetyl-ADP ribose concentrations (5 mM) suggests, however, that the product complex is a major receptor. We conclude that Ex-243 inhibition involves binding after intermediate formation and nicotinamide release, likely to the coproduct complex, whereas the alkylimidate complex can be excluded as the receptor.

Ex-243 Induces the Closed Sirtuin Conformation.

Our Sir2Tm structures and biochemical data on Sirt1, Sir2Tm, and Sirt3 suggest Ex-243 binding to, and stabilization of, a Sirtuin/2′-O-acetyl-ADP ribose complex as its inhibition mechanism. To confirm that the inhibitor is compatible with this product in Sirt3, we solved a Sirt3 complex with 2′-O-acetyl-ADP ribose and Ex-243 (1.9 Å resolution; Table 1). The Sirt3 active site is fully occupied with 2′-O-acetyl-ADP ribose and Ex-243, with binding modes equivalent to the Sir2Tm complexes (Fig. 4A). The coproduct acetyl group appears to display partial rotational flexibility, because it is defined only in two of the three monomers of the asymmetric unit. The cofactor binding loop is trapped in a closed conformation instead of being open for product release. Comparison of our Ex-243 complexes with apo Sirt3 (PDB ID 3GLS) (13) and Sir2Tm/Sirt3 structures with occupied NAD⁺ pockets [Sir2Tm/NAD⁺/p53 peptide, 2H4F (16); Sirt3/intermediate, this study] indeed shows that Ex-243 stabilizes the loop’s closed conformation also adopted on NAD⁺ binding (Fig. 4B; Fig. S4). Sir2Tm-Phe33/Sirt3-Phe157, which was proposed to orient the NAD⁺ nicotinamide riboside and to shield the intermediate against nicotinamide and solvent (15, 16), is packed against the acetyl-ribose and interacts with the B- and C-ring of the inhibitor. The interaction and the associated Phe side chain rotation seem to contribute to this stabilization. Ex-243 thus appears to inhibit Sirtuins by trapping the protein in its closed conformation, preventing product release.

Sirtuin Isoform Selectivity of Ex-243.

To identify Sirtuin features responsible for Ex-243 selectivity, we analyzed sequence alignments and structure comparisons. The insensitivity of Sirt5 against Ex-243 inhibition is likely due to the Sirt5-specific Arg-105 next to the C-pocket, which contributes to the Sirt5-specific response to nicotinamide (36). The only Sir2Tm-specific residue contacting Ex-243 in our complex structures is Met71. It is part of an additional helix turn in Sir2Tm, whereas mammalian Sirt1, 2, and 3 have identical chain lengths and residues in this region and no side chain directly corresponding to Met71. This residue thus does not explain the increased potency for Sirt1/Sir2Tm, and the Ex-243 binding sites are otherwise highly similar for isoforms that are inhibited potently (Sirt1, Sir2Tm) or only weakly (Sirt2, Sirt3). Consistently, Ex-243 affinities are comparable for the sensitive Sir2Tm and the less sensitive Sirt3 (see above). Therefore, factors other than isoform-specific residues interacting with Ex-243 have to cause selectivity.

Because the coproduct contributes to the Ex-243 binding site, isoform differences in binding geometry and affinity could contribute to selectivity. The conserved Sirtuin catalytic core and mechanism involve similar interactions to NAD⁺, indicated by several crystal structures and comparable K_M and K_d values for different Sirtuins (this study and refs. 9 and 36). However, coproduct formation and cofactor loop closure are involved in Ex-243 inhibition, and factors influencing loop conformation and dynamics and catalytic rate thus can influence inhibition. Because Ex-243 inhibits through formation of a complex with Sirtuin and coproduct, analyzing changes in k_cat/K_M on inhibition shows whether product release is rate limiting, as previously shown for yeast Hst2 (37), or whether a step preceding inhibitor binding is rate limiting (38). Indeed, for the less Ex-243–sensitive Sirt3, k_cat/K_M remains unchanged (Fig. 4C), indicating inhibition before the rate-limiting step. In contrast, k_cat/K_M increases for the highly Ex-243–sensitive Sir2Tm (Fig. 4C), indicating inhibitor binding after the rate-limiting catalytic step, which leads to more effective inhibition. Different kinetics of catalysis by Sir2Tm/Sirt1 and Sirt3 thus appear to cause their different Ex-243 sensitivities, and it seems challenging or impossible to identify individual isoform features responsible for the respective inhibitor sensitivity from the many subtle contributions to active site dynamics. This result suggests that rather than optimizing the observed interactions, novel contacts to more remote, isoform-specific residues need to be incorporated in Ex-243 to obtain highly isoform selective Sirtuin inhibitors, and our structures provide a molecular basis for designing such derivatives.

Chemicals and Peptides.

Chemicals were from Sigma if not stated otherwise. O-acetyl-ADP ribose (mixture of 2′ and 3′ form) (44) was a gift from Sirtris, a GSK company. Peptides were synthesized and HPLC purified by GL Biochem: ACS2-K642-peptide [TRSG-Lys(Ac)-VMRRL], p53-K382-peptide [RHK-Lys(Ac)-LMFK], deacetylated p53-K382 peptide (STSRHK-Lys-LMFKTE), H3-K116-peptide [IHA-Lys(Ac)-RVT], and CPS1-K527-peptide [FKRGVL-Lys(Ac)-EYGVKV]. Thio-acetyl-ACS2-K642-peptide [TRSG-Lys(thioAc)-VMRRL] was synthesized by standard Fmoc-based solid phase peptide synthesis and purified using RP-8 HPLC. Fluorenylmethoxycarbonyl-Lys(thioAc)-OH was used for introducing the thioacetylated lysine residue.

Protein Expression and Purification.

Sir2Tm was expressed and purified as previously described using a full-length construct in pET11a (Addgene plasmid 25815) (31, 33).

Recombinant human Sirt1, Sirt3, and Sirt5 were purified as described previously (12, 45). Briefly, His-tagged full-length Sirt1 (in pET15b), His-Trx-tagged Sirt3 residues 118–399 (in pVFT3S), and His-tagged Sirt5 residues 34–302 (in pET151/D-TOPO) were expressed in Escherichia coli BL21DE3 Rosetta2 (Novagen) and purified by affinity chromatography with Talon resin (Clontech). For Sirt3, the tag was removed by incubation with Tobacco Etch Virus protease, and the protein was purified through a second affinity chromatography. Finally, proteins were subjected to gel filtration on Superose12 column (GE Healthcare), concentrated, and stored at −80 °C. Sirtuin concentrations were determined through UV absorption at 280 nm, and Sirt1 concentrations were corrected for contaminants by SDS/PAGE and densitometry. His-tagged Sirt3 for activity assays (residues 114–380) was produced as previously described (46).

Crystallization and Structure Solution.

Sirtuin complexes were crystallized by the vapor diffusion method. Protein solution (Sir2Tm, 10 mg/mL; Sirt3, 11 mg/mL) was mixed with an equivalent volume of reservoir and equilibrated against reservoir at 20 °C. If not stated differently, crystals were transferred to cryoprotectant solution [reservoir supplemented with 25% (vol/vol) glycerol and the respective ligands] and flash frozen in liquid nitrogen.

Sir2Tm was crystallized in complex with acetyl-p53-peptide (1 mM) and Ex-243 (1.5 mM) in the presence of 5% (vol/vol) DMSO and with 25% (wt/vol) PEG 4000, 50 mM Li₂SO₄, and 100 mM Tris, pH 8.5, as reservoir. Crystals were then soaked with NAD⁺ (1 mM) for 2 min in reservoir supplemented with acetyl-p53-peptide (1 mM), Ex-527 (1.5 mM), and 25% (vol/vol) ethylene glycol, and crystals were flash frozen in liquid nitrogen. Sir2Tm in complex with products and Ex-243 was cocrystallized by mixing protein with deacetylated p53-peptide (5 mM), O-acetyl-ADP ribose (5 mM), Ex-527 (1 mM), and 10% (vol/vol) DMSO and using 20% (wt/vol) PEG 6000 and 0.1 M Bis-Tris, pH 5.9 as reservoir.

For the Sirt3/NAD⁺/Ex-243 complex, protein was mixed with acetyl-ACS2-peptide (1 mM), NAD⁺ (10 mM), Ex-527 (1 mM), and 10% (vol/vol) DMSO and the reservoir was 20% (wt/vol) PEG 3350 and 0.2 M ammonium nitrate. The structure of Sirt3 in complex with ADP ribose and Ex-243 was obtained by crystallizing Sirt3 in presence of NAD⁺ (5 mM), Ex-527 (1 mM), and 10% (vol/vol) DMSO in reservoir [0.08 M KH₂PO₄, 18% (wt/vol) PEG 8000, and 20% (vol/vol) glycerol] providing cryoprotection. The Sirt3/thio-intermediate complexes were obtained by soaking Sirt3/thio-acetyl-ACS2-peptide complex crystals (1 mM thio-acetyl-ACS2 peptide; reservoir: 25% (wt/vol) PEG 3350, 0.2 M LiSO₄, 0.1 M Na citrate, pH 5.6) with NAD⁺ (1.25 mM), and optionally Ex-527 (1 mM) and 10% (vol/vol) DMSO in cryoprotectant solution [reservoir plus 25% (vol/vol) glycerol] for several hours. The native intermediate complex with Sirt3 was produced similar to its thio-analog, using 5 mM acetyl-ACS2-peptide and 0.2 M (NH₄)₂SO₄, 0.1 M Bis-Tris, pH 5.5, and 21% (wt/vol) PEG 3350 as reservoir. The Sirt3/acetyl-ACS2-peptide crystals were soaked with NAD⁺ (5 mM), Ex-527 (1 mM), and 10% (vol/vol) DMSO in reservoir supplemented with 25% (vol/vol) glycerol for 80 min. Sirt3/2′-O-acetyl-ADP ribose/Ex-243 complex was crystallized by mixing Sirt3 with O-acetyl-ADP ribose (10 mM), Ex-527 (1 mM), and 10% (vol/vol) DMSO and as reservoir 20% (wt/vol) PEG 6000 and 0.1 M Hepes, pH 7.

X-ray diffraction data were collected at 100 K at beamline MX14.1 (operated by Helmholtz-Zentrum) at the BESSY II electron storage ring (47) using an MX-225 CCD detector (Rayonix). Diffraction data were integrated, scaled, and merged using X-ray Detector Software (XDS) (48) (Table 1). Structures were solved through Patterson searches with Phaser (49) or Molrep (50) using the protein parts of a Sir2Tm/peptide/nicotinamide complex (PDB ID 1YC5) or Sir2Tm/NAD⁺/p53-peptide complex (PDB ID 2H4F) as a search model for Sir2Tm and of a human Sirt3/thio-acetyl-ACS2-peptide alkylimidate intermediate complex (PDB ID 3GLT) as a search model for Sirt3. Structures were refined using Refmac (51), and models were manually rebuilt in Coot (52), with 5% of the reflections excluded from refinement for R_free calculation (53). Individual isotropic Debye–Waller factors were refined, and solvent molecules were added in a later stage of model refinement. For the low-resolution Sir2Tm/products/Ex-243 structure, non-crystallographic symmetry and translation libration screw-motion restraints and a high-resolution reference structure (Sir2Tm/p53-peptide complex, PDB ID 3JR3) were included in refinement. Parameter files for building and refining the ligands Ex-243, reaction intermediate, and 2′-O-acetyl-ADP ribose were generated using MarvinSketch (www.chemaxon.com) and ProDrg (54). Model quality was assessed using MolProbity (55) and validation tools in Coot.

Continuous Deacetylation Assay.

The coupled deacetylation assay was performed as described (31). Briefly, reaction mixtures containing substrate peptide, NAD⁺, inhibitor, and 5% (vol/vol) DMSO were supplemented with 0.75 µM Sirt1, 2.5 µM Sir2Tm, or 2.5 µM Sirt3, respectively, and the decrease in absorbance at 340 nm was monitored using a Bio-Tek plate reader. For IC₅₀ value determination, substrate and NAD⁺ concentrations were adjusted to the respective Sirtuin’s K_M values (∼10-fold K_M: Sirt1, 500 µM acetyl-p53-peptide, 1 mM NAD⁺; Sir2Tm, 1 mM acetyl-p53-peptide, 1 mM NAD⁺; Sirt3 1 mM acetyl-ACS2-peptide, 3.5 mM NAD⁺), and Ex-527 concentrations were varied. Peptide and NAD⁺ titration experiments with Sirt3 were performed using 1 mM NAD⁺ and 500 µM acetyl-ACS2-peptide, respectively, at varying Ex-527 concentrations. Product formation rates were determined by linear regression and fitted with GraFit7 (Erithacus Software) using a tight binding equation for IC₅₀ determination and noncompetitive (peptide titration) or uncompetitive (NAD⁺ titration) inhibition equations for kinetics.

MS-Based Deacetylation Assay.

Peptide and NAD⁺ titration experiments with Sir2Tm were performed using an MS deacetylation assay as described (36), with 2.5 µM Sir2Tm and 1 mM NAD⁺ or 500 µM acetyl-p53-peptide, respectively, varying Ex-527 concentrations, and 5% (vol/vol) DMSO. Briefly, aliquots of a reaction were stopped after different time points, and substrate and product were separated and quantified by nano-LC-electrospray ionization mass spectrometry on an LTQ XL (Thermo Fisher scientific), followed by linear fitting to obtain product formation rates. Turnover rates were fitted with GraFit7 using equations for noncompetitive (peptide titration) or uncompetitive (NAD⁺ titration) inhibition.

For testing product formation under saturating Ex-527 or Ro-31-8220 concentrations, respectively, varying Sirtuin amounts (Sirt1, 3.4/6.8/13.6 µM; Sir2Tm, 4/8/16 µM; Sirt3, 5/10/20 µM) were mixed with substrate peptide (Sirt1 and Sir2Tm, 200 µM acetyl-p53-peptide; Sirt3 50/200 µM acetyl-ACS2-peptide), 500 µM NAD⁺, inhibitor (Sirt1 and Sir2Tm, 150 µM Ex-527; Sirt3, 500 µM Ex-527/Ro-31-8220), and 5% (vol/vol) DMSO and incubated at 37 °C, and the reaction was stopped after 30 min by addition of 0.25% (vol/vol) trifluoroacetic acid. Samples were analyzed, and product was quantified as described previously (36). Calibration lines (Fig. S3 F and G; regression coefficient >0.95) were generated with different ratios of the respective peptide pairs (deacetylated/acetylated) for absolute quantification.

Binding Analysis by Microscale Thermophoresis.

Affinities were measured by microscale thermophoresis (56). Sirtuins were labeled by incubation with twofold molar excess of FITC (Thermo Fischer) in 20 mM Hepes, pH 7.5, and 150 mM NaCl at 4 °C overnight. Free dye was removed with a Nap25 column (GE Healthcare). Sirtuin was mixed with varying concentrations of Ex-527, and thermophoresis was measured (excitation wavelength 470 nm, emission wavelength 520 nm, LED-power 10–20%, laser-power 10%) using a Monolith NT 115 (NanoTemper Technologies) in the absence and presence of 5 mM NAD⁺ and 1 mM acetylated peptide (Sir2Tm, acetyl-H3-peptide; Sirt3, acetyl-ACS2-peptide). Dissociation constants were determined with GraFit7 (Erithacus Software) by nonlinear fitting (one-site and two-site fitting equations). Each experiment was repeated at least twice.