A continuous sirtuin activity assay without any coupling to enzymatic or chemical reactions

Sirtuins are NAD+ dependent lysine deacylases involved in many regulatory processes such as control of metabolic pathways, DNA repair and stress response. Modulators of sirtuin activity are required as tools for uncovering the biological function of these enzymes and as potential therapeutic agents. Systematic discovery of such modulators is hampered by the lack of direct and continuous activity assays. The present study describes a novel continuous assay based on the increase of a fluorescence signal subsequent to sirtuin mediated removal of a fluorescent acyl chain from a modified TNFα-derived peptide. This substrate is well recognized by human sirtuins 1–6 and represents the best sirtuin 2 substrate described so far with a kcat/KM-value of 176 000 M−1s−1. These extraordinary substrate properties allow the first determination of Ki-values for the specific Sirt2 inhibitory peptide S2iL5 (600 nM) and for the quasi-universal sirtuin inhibitor peptide thioxo myristoyl TNFα (80 nM).

. (a) Sirtuin-mediated deacylation reaction transfers fluorescently labeled acyl residues from lysine side chain to ADP-ribose. (b) Sirtuin-mediated deacylation reaction transfers quencher-containing acyl residue from lysine side chain to ADP-ribose. In both cases sirtuin activity causes an increase in the fluorescence signal. (Fl -fluorophore, Qu -quencher, ADPr -ADP ribose) Scientific RepoRts | 6:22643 | DOI: 10.1038/srep22643 modification of this side chain and Fmoc-based solid phase peptide chemistry was employed. The peptide 1a is the best Sirt6 substrate described so far 21 .
1a and 2a were subjected to HPLC-based activity assay to assess their substrate properties and to determine if the quencher moiety is accepted by sirtuin 6. Negative controls without NAD + under identical conditions yielded  no conversion of substrates. The kinetic constants uncovered that the replacement of threonine in + 1 position of the substrate by the quencher moiety 3-nitrotyrosine did not influence the turnover number and minimally disturbed the apparent affinity to the active site of Sirt6 as reflected by the almost comparable K M values for 1a (6 μM) and 2a (17 μM) (Supplementary Fig. S3).
Sirtuins 1-5 were also tested and 1a was shown to represent a universal sirtuin substrate with k cat /K M values in the range of 10 to 50 000 M −1 s −1 (Supplementary Table S2). Therefore, we reasoned that use the fatty acid chain could be utilized as an attachment point for the very small fluorophore 2-amino-benzoylamide. Systematic variation of the distance (number of bonds) between the amide bond on the lysine side chain and the 2-amino-benzoylamide moiety (i.e. 3, 4, 4a) revealed good substrate properties for 3 only (Table 1). Peptide 4 was not a substrate for sirtuins 1 and 3-7 but showed some activity for Sirt2 in an HPLC based end-point activity assay. However, increasing the number of methylene groups to place the fluorophore to a different position yielded improvement in substrate properties for Sirt2 with an optimum for 3. Further elongation of the spacer resulted in > 1000-fold decrease of substrate properties for Sirt2 (Table 1) and complete loss of activity for sirtuin isoforms 1 and 3-6. Substrate 3 represents a quasi-universal sirtuin substrate because it is recognized by isoforms 1-6. The development of 3 resulted in slightly decreased substrate properties for sirtuins 1, 3, 4, 5, and 6 as compared to 1a but interestingly yielded an improved substrate for Sirt2 with a specificity constant of 176 000 M −1 s -1 representing the best Sirt2 substrate described so far (Table 2). Sirt7 was not able to recognize substrate 3 pointing to structural differences of the hydrophobic channel accommodating the acyl chain.
Sirtuin mediated transformation of 3 into 2b could be followed directly and continuously (λ Ex = 310 nm, λ Em = 405 nm) using a fluorescence spectrometer ( Supplementary Fig. S5). Without NAD + in the presence of sirtuin enzyme or without sirtuin in the presence of NAD + no significant change in fluorescence signal over time could be observed ( Supplementary Fig. S5). This indicated that the observed fluorescence change results directly from sirtuin-mediated deacylation and not from unspecific interactions between NAD + and/or sirtuin and 3.
The slope of change in fluorescence intensity is linearly dependent on the enzyme concentration ( Supplementary Fig. S5). Progress curves at different concentrations were linear below 25% conversion of the substrate. We used a completely converted assay solution (controlled by LC-MS) for the generation of appropriate calibration curves ( Supplementary Fig. S9). Additionally, we were able to demonstrate that the activity assay is compatible with 96-and 384-well microtiter plate-based equipment yielding Z´-factors of 0.85 for 3 at 25 μM concentration. Kinetic constants determined with either HPLC based assay or with the assay performed in both MTP fluorescence readers and spectrophotometers yielded comparable results (Supplementary Table S2).
Due to the relatively low k cat -values of the known substrates, "classical" sirtuin activity assays are done in timeframes between 30 and 120 min and at enzyme concentrations between 0.5 and 4 μM to generate sufficient signal changes. At these conditions the basic assumption of the Michaelis-Menten-equation [S] is not valid. Moreover, the high amount of enzyme prevents the correct determination of K i -values for sirtuin inhibitors with affinities below half of the enzyme concentrations used. With substrate 3 we were able to follow enzymatic activities down to 10 nM sirtuin concentration ( Supplementary Fig. S8). We used a 96-well MTP fluorescence reader for the determination of the K i -values for different compounds (Fig. 3) including inhibitors with high affinities to sirtuin isoforms (Table 3).
The first product of the sirtuin reaction, nicotinamide (NAM), is known to be a non-competitive inhibitor with respect to both acylated peptide substrate and NAD + cosubstrate by re-binding to the active site and attacking the sirtuin bound O-alkylimidate reforming NAD + . For Sirt6 an IC 50   We determined the K i -values of NAM for Sirt3 and Sirt6 to be 93 μM and 451 μM, respectively, using NAD + at saturating conditions ( Supplementary Fig. S13). Under peptide substrate saturating conditions K i -values were found to be 45 μM and 415 μM, respectively ( Supplementary Fig. S13). The K i -value for NAM was lower than expected for Sirt6, but still higher than for other isoforms. Recently, it was shown that the IC 50 -values for NAM are dependent on the chemical nature of the acyl moiety and that different sirtuin isoforms have different acyl-dependent susceptibilities to NAM inhibition 57 . Our substrate closely resembles the physiological myristoyl substrate hence our value should reflect the sensitivity of this substrate modification. Recently, compounds Quercetin and Ex-527 were reported as Sirt6 inhibitors with inhibition of enzymatic activity of 52% and 56%, respectively, if used at 200 μM concentration 58 . We determined K i -values for these two small molecules and found considerable non-competitive inhibition with respect to the peptide substrate ( Table 3). The cyclic peptide derivative S2iL5, containing a trifluoroacetylated lysine side chain as a warhead for inhibition of sirtuin catalysis 59 , was claimed to be a Sirt2 specific inhibitor with affinities to the active site in the low nanomolar range as determined by isothermal calorimetric measurements 60 . Using 3 as substrate the determined K i -value is 560 nM and the cyclic inhibitor behaved non-competitive for the peptide substrate ( Supplementary Fig. S17). Replacement of the amide bond formed by the acyl chain and the ε -amino function of the lysine side chain by a thioxo amide bond transforms substrates into extremely slow substrates/inhibitors by generation of a stalled intermediate resembling sirtuin bi-substrate inhibitors [61][62][63] . Thioxo myristoylated and shortened derivatives of 1 were shown to be cell permeable inhibitors for Sirt6 with remarkable cross-reactivity to Sirt1-3 and reported IC 50 values in the single digit micromolar range 25 . Due to the high sirtuin concentration used in this enzymatic assay (i.e. 1 μM Sirt6) we wondered if these values are too high, not properly reflecting the K i . We determined the K i -values of 8 for sirtuins 2, 3, and 6 using substrate 3 and 96-well-based readout (Table 3) and determined higher affinities to the sirtuins especially for Sirt2 with K i -value of 80 nM. We were able to determine these K i -values because in our case the enzyme concentration was about 100 times lower as compared to the assay proposed by He et al. 25 . Additionally, Sirt3 showed high affinity to 8. We were not able to determine the K i -values if we pre-incubate the enzymes Sirt2 and Sirt3 (10 nM) with different concentrations of 8 in the presence of NAD + for 30 min enabling formation of the "stalled" intermediate without competition with the substrate peptide. Starting the reaction by addition of 3 we observed complete inhibition with low nanomolar concentrations of 8 demonstrating that the pre-formed bi-substrate inhibitor has affinities to Sirt2 and Sirt3 in the very low nano-or picomolar range. The resulting non-competitive inhibition against peptide substrate was in accordance with the suggested model of bi-substrate like inhibitors for thioxo acylated derivatives ( Supplementary Fig. S15).
Identification of small molecule modulators of sirtuin activity using 3 could be hampered by absorbance/fluorescence of the effectors in the range between 320 nm and 400 nm. Consequently, the known sirtuin activity modulator resveratrol could not be analyzed because of the high extinction coefficient in that range. Using HPLC-based activity assay we found no significant influence of resveratrol on Sirt1 mediated deacylation of 3 ( Supplementary Fig. S4) To be able to analyze small molecules with absorption/fluorescence in the range of 2-aminobenzoylamide fluorescence, we exchanged the 2-aminobenzoylamide fluorophore by (4-N,N-dimethylamino-1,8-naphthalimido)-acetamide resulting in derivative 6 with fluorescence excitation at 471 nm 64 . Surprisingly, this fluorophore could be used in combination with the 3-nitro-L-tyrosine quencher despite non-optimal overlap of the spectra. Compound 6 was not a substrate for sirtuins 1, 3, 5 and 6 as determined by HPLC-based assays and was a weak substrate for Sirt2 with an about 500-fold lower k cat /K M -value as compared to 3. Interestingly, Sirt4 recognized 6 better than 3 resulting in an about 10-fold improved K M -value (Supplementary Table S2) indicative of differences in the flexibility of the hydrophobic channel accommodating the acyl chain between Sirt4 and the other sirtuin isoforms. However, these results showed that the development of substrates with different spectral properties is possible enabling the simultaneous detection of enzymatic activity using substrate mixtures. We analyzed kinetics for Sirt2 and Sirt4 using a mixture of substrates 3 and 6 using

11
Sirt2 50.0 ± 9.9 n.d. 290 nm for excitation of both fluorophores and recording fluorescence spectra over time ( Supplementary Fig. S10 and S11). Furthermore development of isoform selective substrates should be possible by systematic variation of the size and position of the fluorophore in the acyl side chain. Because of the obvious limitations for most of the sirtuins in accommodating bulkier fluorophores, we decided to create a small quencher moiety at this position closely related to the well-recognized 2-amino-benzoylamide residue. Addition of a nitro function in para-position to the amino-group of the 2-amino-benzoylamide moiety generated a very efficient quencher for more bulky fluorophores like 7-methoxy-coumaryl-L-alanines (Mca) or even (4-N,N-dimethylamino-1,8-naphthalimido)-L-alanines (Dma) (data not shown). We speculated that sirtuins are less sensitive for modifications within the peptide sequence and synthesized several derivatives of 1a characterized by substitutions of residues in different positions relative to the myristoylated lysine by either Dma (12,13) or Mca (15,16). Additionally, we attached the fluorophores to the N-terminus in form of appropriately substituted acetyl residues (14,17). Analysis of sirtuin 2, 3, and 6 activity against these substrates using an HPLC-based assay revealed that all peptides are substrates but only 13 and 16 are well recognized (Supplementary Table S1). For solubility reasons (data not shown) we decided to combine the 7-methoxy-coumaryl-L-alanyl-residue with the 5-nitro-2-amino-benzoylamide quencher moiety resulting in 7. The substrate properties of 7 for Sirt2 and Sirt4 are superior to substrates described in the literature and similar to 3, demonstrating that fluorophore and quencher positions could be switched without influence on substrate properties (Supplementary Table S2 and Fig. 1).
Inspection of the published crystal structure of Sirt6 in complex with the myristoylated H3K9 substrate (PDB ID 3ZG6) and the respective electron density maps revealed that both conformations of the amide bond between the fatty acid and the lysine side chain amino function could be fitted, but the published coordinates are given in a conformation resembling the cis conformation of peptide bonds 21 . In the recently reported structures of Sirt2 complexed with a thioxo myristoylated inhibitor closely related to 8 (PDB ID 4Y6Q) or a thioxo myristoylated peptide derived from Histone H3 (PDB IDs 4Y6L and 4R8M) the conformation of the thioxo amide bond was in trans conformation 24,57 which is the preferred conformation of secondary amide/thioxo amide bonds in aqueous solutions. We synthesized 8 in order to analyze if there is any isomer-specificity during binding to the active site of Sirt2 and Sirt6. Cis/trans isomerizations of secondary amide bonds are too fast compared to the time needed for "classical" sirtuin activity assays preventing such analyses. Our assay allowed enzymatic measurements within short time and the isomerization of thioxo amide bonds is slower at lower temperatures 65 . Moreover, the UV-absorption of the π -π * transition for the cis conformer of thioxo amide bond is slightly red-shifted enabling determination of cis/trans isomerization rates using UV-spectroscopy 66 . We determined the isomerization rate for 8, 10 and 11 at different temperatures subsequent to increasing the cis content in the photo-excited state using UV-light (Supplementary Figs S19-33) 66,67 . The re-equilibration to the ground state (nearly 100% trans-conformation) could be followed using UV-spectroscopy at 260 nm yielding activation parameters (Supplementary Figs S19-33). In order to analyze isomer-specific inhibition of Sirt2 by 8 we optimized the assay conditions to measure the enzymatic activity using 3 without significant cis/trans isomerization during the assay. The cis content of 8 is 2.5% in assay buffer and up to 25% in the photo-excited state as measured by HPLC (Supplementary Fig. S34). We determined the rate of re-equilibration for 8 and calculated the resulting cis content subsequent to different times of darkness (Supplementary Table S3). This setup enabled the determination of inhibition of Sirt2 mediated deacylation of 3 at different cis contents of 8 ranging from 2.5% to 25%. We found no significant difference in inhibitory effect depending on the cis content of 8 indicating an unexpected plasticity of the active site of Sirt2 to accommodate both conformations with similar affinities. This result indicated that the Sirt2/myristoyl-peptide complexes, which were modeled as in trans conformation 24,57 would also be compatible with cis. IC 50 -values of Sirt6 were determined as 1.06 ± 0.12 μM and 1.87 ± 0.18 μM for 2.5% and 25% cis isomer of 8, respectively, pointing to a small preference for the cis conformation of the amide bond ( Supplementary Fig. S36). Nevertheless, because of the suboptimal substrate properties of 3 for Sirt6 the assay duration was 30 min which allows significant re-equilibration of the photo-induced change of the cis/trans equilibrium. Therefore, we introduced an additional methyl group at the lysine nitrogen resulting in a tertiary thioxo amide 9 which is an inhibitor with similar affinities to the active site of Sirt6 as compared to 8 (Supplementary Table S4). The rate constant for the cis/trans isomerization of the tertiary thioxo amide bond of 9 (5.4 × 10 −4 s −1 at 20 °C) was much slower than that of the secondary thioxo amide bond of 8 ( Supplementary Fig. S35). HPLC analyses revealed a cis content of about 50% and there was no change detectable subsequent to photo-excitation at the π -π * transition of the tertiary thioxo amide bond. We tested several different organic solvents to change the cis content but found no sufficient differences (Supplementary Table S5). Therefore, the two isomers were separated by HPLC at low temperatures (4 °C).
We were able to enrich the faster migrating cis isomer to 72.4% and the trans isomer to 70.3% (Fig. 4). The frozen isomers (− 70 °C) were stable for several days (Supplementary Fig. S39).
Determination of inhibition of Sirt6 by 9 using samples with different cis content showed minimal preference for the cis isomer (IC 50 values of 0.6 μM and 1.7 μM for 72.4% and 29.7% of cis of 9, respectively Supplementary Fig.  S38). These results again demonstrated the plasticity within the active site of sirtuins, at least for Sirt6, enabling both isomers to bind with similar affinities. Inspection of the electron density maps of PDB 4R8M and 3ZG6 suggest that there is sufficient space around the lysine side chain amide/thioxo amide bond to fit both isomers. Recently, Sirt6 coordinates of 3ZG6 were re-refined by Denus lab and it was established that the myristoylated peptide should be in a trans conformation regarding the amide bond between the lysine side chain and the acyl moiety 57 .
Here we present a continuous sirtuin activity assay allowing convenient measurement of highly accurate data. The sensitivity of the activity assay enables the reliable determination of K i -values for inhibitors with affinities below 100 nM. Because of the demonstrated compatibility with 384-well MTP readout we expect that this assay principle will find widespread application in drug discovery projects. Additionally, the superior substrate properties of 3 allow the investigation of isomer specificity in the binding of inhibitors to the active site of sirtuins enlarging the portfolio of tools in sirtuin research. For HPLC separations solvents consisting of water (solvent A) and ACN (solvent B), both containing 0.1% TFA, were used. Analytical runs were performed on an Agilent 1100 HPLC (Boeblingen, Germany) with a quaternary pump, a well-plate autosampler and a variable wavelength detector. Separations were performed on a 3.0 × 50 mm reversed phase column (Phenomenex Kinetex XB C-18, 2.6 μm) with a flow-rate of 0.6 mL/min. A Merck-Hitachi High Speed LC system (Darmstadt, Germany) with a Merck Hibar Li Chrospher ® RP-8 column (250-25 mm, 5 μm) was used for preparative separations (flow-rate: 8 mL/min). Eluted compounds were analyzed by MALDI mass spectrometry. NMR spectroscopy was carried out using Varian Gemini 2000 spectrometer in deuterated chloroform.  (30 mmol) in acetone (60 mL) was added. Next solid NaHCO 3 (75 mmol) was added in portions over 1 hour. The stirred reaction mixture was left in a melting ice bath overnight. The blue precipitate was filtered, subsequently washed with H 2 O, ethanol and diethyl ether (Et 2 O). After air drying yield of the complex was 87%. To the copper complex of ε -nosyl lysine (5 mmol) a solution of ethylenediaminetetraacetic acid (EDTA) disodium salt (6.5 mmol) in 40 ml of H 2 O was added. This suspension was stirred and heated at 70-80 °C until no blue complex was left and then cooled to room temperature. Afterwards, solid NaHCO 3 (10.5 mmol) was added to the formed suspension of ε -nosyl lysine and followed by solution of Fmoc-N-hydoxysuccinimide ester (Fmoc-OSu) (10.5 mmol) in 30 mL of acetone. The mixture was stirred vigorously overnight, diluted with 250 mL of 1% solution of NaHCO 3 and extracted with Et 2 O (3 × 100 mL). Ether washings were back extracted with diluted NaHCO 3 solution and discarded. Combined aqeous phases were acidified with 10% HCl and extracted with dichloromethane (DCM) (3 × 50 mL). Combined organic phases were washed with water and dried over Na 2 SO 4 . Solvent was evaporated to afford target compound as white foam. Yield: 91%.  13  Synthesis of carboxymethyl dithiomyristoate. Carboxymethyl dithioester was prepared in accordance to Leon et al. 68 . To the solution of myristic acid (4 mmol), HBTU (4 mmol) and N,N-diisopropylethylamine (DIPEA) (8 mmol) in DCM (30 mL) was added piperidine (4.2 mmol). After 4 hours reaction mixture was diluted with water and extracted with DCM. Extracts were washed with diluted HCL, diluted NaHCO 3 and with water, dried over Na 2 SO 4 and DCM was evaporated. To the residue toluene (10 mL) was added followed by Lawesson's reagent (2 mmol). Reaction was heated at 106 °C for 3.5 hours. Solvent was evaporated and the residue flash-chromatographed (silica gel, ethyl acetate (EtOAc)/petr.ether 1:9) (Yield: 74%).
N,N-pentamethylenesuccinamic acid (10 mmol) was refluxed in anhydrous MeOH (15 mL) containing 2 drops of H 2 SO 4 for 4 hours. Solvent was evaporated and residual oil dissolved in EtOAc, washed with NaHCO 3 solution, water and dried over Na 2 SO 4 . Evaporation of EtOAc afforded product as a colorless oil (Yield: 84%).
To a solution of Methyl 3-(N,N-pentamethylenethiocarbamoyl)propanoate ( 2 mmol) in anhydrous THF (8 mL) MeI (10 mmol)was added. The reaction was conducted for 48 h in darkness Yellow-colored THF was decantated, the crystals were briefly washed with dry THF and dissolved in dried DMF (3 mL). Dried H 2 S was bubbled into solution for 2 h and mixture was left at 0 °C for 24 h. After addition of H 2 O (100 mL), product was extracted with EtOAc, washed several times with water, brine and dried over Na 2 SO 4 . Evaporation of EtOAc in vacuo gave the crude product as a yellow oil (Yield: 81.5%).
The cyclic peptide inhibitor S2iL5 was synthesized by standard Fmoc-based solid phase peptide synthesis as described by Yamagata et al. 60 .
To obtain the expression plasmid of human (His) 6 -SUMO-Sirt4(29-314), the respective DNA fragment was PCR-amplified using gene-specific primers from the plasmid pET101-Sirt4, which carries the Sirt4 gene, and cloned into the BsaI, XbaI sites of pE-SUMO yielding the plasmid pE-SUMO-Sirt4 .
The protein was overexpressed in E. coli BL21 (DE3) cells at 18 °C. The purification of the protein was performed using affinity chromatography on Ni-NTA resin in 10 mM Tris-HCl, pH 7.5, 0.5 M NaCl. The matrix-bound (His) 6 -SUMO-Sirt4(29-314) was eluted by imidazole in the buffer and further purified by gel filtration in 10 mM HEPES, pH 7.8, 150 mM KCl, 1.5 mM MgCl 2 , and stored at − 20 °C for use.
HPLC based activity assay. For the determination of kinetic constants for all sirtuin mediated reactions solutions containing 20 mM TRIS/HCl pH 7.8, 150 mM NaCl, 5 mM MgCl 2 (assay-buffer), 500 μM NAD + and varying substrate concentrations (0.5-100 μM) were used. Deacylation was started by adding human sirtuin to reach a final concentration of 0.01-0.5 μM. Enzyme-catalyzed reaction was stopped using TFA (1% final concentration) after 1 min to 180 min of incubation at 37 °C depending on substrate reactivity. The cleavage rate of the different TNFα peptide derivatives was analyzed using analytical reversed phase HPLC. 40 to 80 μl of compounds or reaction solutions were injected and separated using a linear gradient from 5% to 95% solvent B within 6min. The product and substrate peaks were quantified using absorbance at 220 nm or 365 nm (absorption of 3-Nitrotyrosyl moiety). The peak areas were integrated and converted to initial velocity rates calculated from the ratio of product area to total peak area. Linear regression of conversions plotted against time yielded reaction rates in μM/min (relative conversion below 20% of substrate). Non-linear regression according to Michaelis-Menten of the reaction rates at different substrate concentrations yielded K M -and k cat -values using the program SigmaPlot 8 (Systat Software, San Jose, USA). All measurements were done in duplicates.
Continuous fluorescence assay. The fluorescence measurements were performed on a Hitachi F-4500 fluorescence spectrophotometer (Tokyo, Japan) at λ Ex = 310 nm and λ Em = 405 nm (slit Ex = 5 nm, slit Em = 2.5 nm, PMT = 700 V for 3, 4 and 5 as well as slit Ex = 10 nm, slit Em = 10 nm, PMT = 950 V for 4a). Each reaction mixture contained assay-buffer, 0.5 mM NAD + and various peptide concentrations (0.1-100 μM) and was preincubated for 5 minutes at 37 °C. The reaction was started by adding human sirtuin (0.1-0.5 μM) and observed for 5-10 minutes. Product formation could be monitored by increase of relative fluorescence. This signal was converted into product concentration via calibration lines. The slope of the linear regression of product formation against time yielded the reaction velocity rates in μM/s. K M and k cat were obtained by non-linear regression according to Michaelis-Menten. All measurements were done in duplicates. For determination of reaction velocity rates in μM/s calibration lines were necessary. Therefore a reaction mixture was prepared, containing assay-buffer, 2 μM Sirt2, 500 μM NAD + and 100 μM of 3, 4, 4a or 5 was incubated overnight at 37 °C. The reaction mixture was analyzed with HPLC, to control if the entire peptide substrate was turned to product. Additionally the mixture was diluted (0.1-25 μM) and measured with Hitachi F-4500 fluorescence spectrophotometer at the same conditions as described above.
The microtiter plate fluorescence measurements were performed on a Tecan Infinite M200 microplate reader (Maennedorf, Switzerland) at λ Ex = 320 nm and λ Em = 408 nm (lag time 9 μs, integration time 20 μM, gain 160, 170 or 182). The reactions (total volume 100 μl) were measured in black low-binding 96-well microtiter plates (NUNC). Assay-buffer, 500 μM NAD + and 0.07-200 μM peptide substrate were pre-incubated at 37 °C for 5 min. The reaction was started by adding human sirtuin (0.01-0.5 μM). The signals were converted into product concentration via calibration lines and the resulting data were evaluated as described above (single fluorescence measurement). The determination for the kinetic constants of NAD + was performed in the same way, except that the peptide concentration was fixed (5, 25 or 200 μM) and the NAD + concentration was varied (10-1500 μM). All measurements were done in duplicates. The reaction mixture for the calibration lines was prepared as described for the single fluorescence measurements. After complete turnover of peptide substrate 3, the solution was diluted (0.2-20 μM) and measured on Tecan infinite M200 microplate reader at λ Ex = 320 nm and λ Em = 408 nm (lag time 9 μs, integration time 20 μM, gain 182 (G182), 170 (G170) and 160 (G160)) K i values of the inhibitors were determined by recording k cat and K M values for 3 in the presence of varying inhibitor concentrations (0.01-600 μM). The resulting plots were analyzed by a competitive inhibition and non-competitive inhibition model using the program Sigma Plot 8. The linear regression of the apparent K M -values against the corresponding inhibitor concentration yielded the inhibitor constant K i for competitive inhibition. The K i for non-competitive inhibition was determined by linear regression of 1/apparent V max against the corresponding inhibitor concentration. The negative K i value can be determined as intersection with the X-axis from these plots.

Photo-induced change of cis content of thioxo peptides. Excitation experiments of thioxo peptides
were done in a cuvette at 254 nm under stirring with a UV-lamp (UV handheld lamp, Carl Roth). For irradiation a distance of 5 cm between cuvette and UV-lamp was chosen. UV-spectra were recorded between 230 and 325 nm using a spectrophotometer (Specord M500).
For determination of temperature dependent cis/trans isomerization a 50 μM solution of thioxo peptide was incubated for 10 min at different temperatures (10-70 °C). UV-spectra were recorded at ground state (GS) and after irradiation at 254 nm (irradiation time 45 s to 5 min) at photostationary state (PSS). Several UV-spectra over time were recorded to determine rates of cis/trans isomerization. Using a differential spectrum (UV spectrum GS -UV spectrum PSS) activation parameter and isomerization velocity could be examined.
The cis/trans content of a 50 μM thioxo peptide solution was changed by 5 min irradiation at 254 nm and the resulting solution was analysed by HPLC. Additionally, several solvents were tested to enhance cis content. As solvents H 2 O, acetic acid, TFA, trifluoro ethanol (TFE), 0.5 M LiCl in H 2 O/ethanol (EtOH)/TFE, methanol (MeOH), formic acid, N-methyl pyrolidon (NMP), DMF, Dimethylsulfoxid (DMSO) and tetrahydrofuran (THF) were chosen. Cis content was determined via HPLC of a 500 μM solution of 9.
For the separation of isomers, 5-6 mg of 9 were dissolved in 50% ACN and equilibrated overnight. For better separation HPLC-solvents were cooled down to 4 °C and a linear gradient of 45% solvent B to 55% solvent B in 70 min was used. Eluted fractions were immediately frozen in liquid nitrogen. HPLC-based determination of cis content was done directly after preparative separation.
The examination of the isomer specific inhibition of 8, 9, 10 and 11 was examined via HPLC using reaction solutions composed of 500 μM NAD + , 30 μM peptide, 0.5 μM sirtuin and 0.5-40 μM inhibitor in GS, PSS or GS* in assay buffer. After 30 min incubation at 20 °C reaction was stopped using 10% TFA solution. Inhibitor solutions were irradiated at 254 nm for 5 min. Separated isomers were applied in concentrations from 1 to 10 μM. The influence of cis content on sirtuin inhibition using fluorescence spectrometer was determined with 500 μM NAD + , 5 μM 3, 0.1 μM Sirt2 and 0.1 μM 9. Reactions were done at 20 °C with 9 in GS and PSS (after 5 min irradiation at 254 nm). 9 in PSS was applied immediately after irradiation (transfer time ~5 s) and started directly by adding sirtuin. Reactions were measured within 1 min to avoid re-isomerization.
Z' factor analysis. The Z′ factor is a dimensionless, simple statistic parameter for high-throughput screening assays 72 . It is defines as the ratio of separation band to signal dynamic range of the assay and used the signal variation at the two extremes of the activity range (0 and 100% activity).