NE-Thioacetyl-Lysine-Containing Tri-, Tetra-, and Pentapeptides as SIRT1 and SIRT2 Inhibitors

N‹-Thioacetyl-lysine-containing tri-, tetra-, and pentapeptides, based on the R-tubulin and p53 protein sequences, were studied as SIRT1 and SIRT2 inhibitors. The potency of the pentapeptides depended on the selection of the side chains. The removal of N- and C-terminal residues of the pentapeptides yielded tripeptides with retained SIRT1 inhibitory activity but decreased SIRT2 inhibitory activity. The most potent SIRT1 inhibitors were equipotent with the reference compound (6-chloro-2,3,4,9-tetrahydro-1H-carbazole-1- carboxamide) with the IC50 values of 180-330 nM.

Introduction

Silent information regulator human type 1 and 2 (SIRT1 and SIRT2a) enzymes belong to class III histone deacetylases (HDAC) that require nicotinamide adenine dinucleotide (NAD+) as a cosubstrate to catalyze the deacetylation reaction.1-3 SIRT1 is a nuclear enzyme,4 while SIRT2 has been claimed to move from cytoplasm to nucleus during mitosis.5,6 SIRT1 and SIRT2 have been reported to deacetylate both histone and nonhistone proteins.7 The biological function of SIRT1 has been linked to type 2 diabetes and heart diseases,8,9 while SIRT2 might have a role with glioma tumorigenesis and Parkinson’s disease.10,11

It has been claimed recently that compounds which could mimic the binding of the acetylated peptide substrate might be potent and selective inhibitors of Sir2 deacetylases over other NAD+-metabolizing enzymes.12 In general, peptides have shown advantages over small molecules in terms of specificity and affinity for different therapeutic targets.13

A few preferred substrates, such as p53 for SIRT1 and R-tubulin for SIRT2, have been reported.14,15 In recent years, it has been claimed that the sirtuins might recognize certain amino acid side chains of specific substrates near the N‹- acetylated-lysine residue.16,17 Recently, peptides containing N‹- thioacetyl-lysine have been reported to inhibit SIRT1, SIRT2, and SIRT3 on a low micromolar level.18-20 These results together with the evaluated inhibition mechanism of N‹- thioacetyl-lysine-containing peptides have given a starting point for the research of substrate-based SIRT1 and SIRT2 inhibi- tors.20

A new series of shorter N‹-thioacetyl-lysine-containing tri-, tetra-, and pentapeptides was designed based on two substrate sequences; human R-tubulin (38-42) and human p53 tumor suppressor protein (380-384) (Figure 1), in order to determine important amino acid side chains and minimum peptide length for SIRT1 and SIRT2 inhibitory activity. Fatkins and Zheng have reported earlier pentapeptide 6 with an IC50 value of about 10 µM (the exact value was not reported).19 Garske and Denu have shown that for the peptide substrates the amino acids beyond positions -2 and +2 (the second residues toward the N-terminus and C-terminus, respectively, calculating from the N‹-acetyl-lysine) are not necessary for efficient binding and activity of sirtuins.17 In this study we show that several pentapeptides are potent SIRT1 and SIRT2 inhibitors. Further truncation by removal of the -2 and +2 amino acids is tolerated by SIRT1 but not by SIRT2.

Peptides were synthesized manually or on a peptide synthe- sizer using an Fmoc strategy with TBTU or HBTU and DIPEA as the coupling reagents and p-alkoxybenzyl alcohol resin or amino resin with the 4-hydroxymethylphenoxyacetic acid linker as the solid phase. NR-Fmoc-N‹-thioacetyl-lysine was synthesized as described in the literature.18 All other amino acids used were common natural amino acids. Reference compound 24 (EX-527) was synthesized as described in the literature and tested as a racemate.21 The inhibitory activities were tested in a Fluor de Lys fluorescence-based assay.

Results and Discussion

The synthesized peptides and their inhibitory activities are presented in Table 1. Peptides 1-5 based on the human R-tubulin (38-42) sequence and peptides 6-9 based on the human p53 protein (380-384) sequence were synthesized to study the effect of the peptide length. Pentapeptide 6 has been
the p53-based peptides gave better inhibitory activities for SIRT1 and SIRT2 than the R-tubulin-based peptides. This observation prompted us to synthesize peptides 10-15, in which one or two amino acids of the R-tubulin sequence were replaced by the corresponding amino acids from the p53 sequence. Peptides 16-23 with the alanine replacements on the R-tubulin and p53 sequences were synthesized to study the importance of the individual amino acid side chains.

Figure 1. Human R-tubulin 38-42 (1) and human p53 tumor suppressor protein 380-384 (6) sequences possessing an N‹-thioacetyl-lysine.

At the -2 position, the serine residue of the R-tubulin-based sequence 1 was replaced by either a histidine or an alanine residue, resulting in peptides 10 and 16, respectively. Serine and histidine side chains are able to form hydrogen bonds while an alanine side chain is not. All three peptides were equipotent against SIRT1 and SIRT2, indicating that the hydrogen bonding of the -2 side chain is not prerequisite for good inhibitory activity.

Removal of the -2 amino acid resulted in the R-tubulin- based tetrapeptide 3, which, interestingly, was equipotent against SIRT1 but almost three times less potent against SIRT2 than pentapeptide 1. The same effect was observed with the p53- based sequences; tetrapeptide 8 was equipotent against SIRT1 but almost eight times less potent against SIRT2 than pentapep- tide 6. It seems that SIRT1 and SIRT2 differ in their binding interactions with the -2 amino acid of the peptides. For SIRT2, the presence of an amino acid at this position was more important than the identity of its side chain, indicating that the main chain interactions of the -2 residue may be more relevant than the side chain interactions.

At the -1 position, the aspartic acid residue of the R-tubulin- based peptide 1 was replaced by either a lysine or an alanine residue, resulting in peptides 11 and 17, respectively. Both peptides were clearly more potent than the parent peptide 1 for SIRT1 and SIRT2. In fact, peptides 11 and 17 were slightly more potent for SIRT1 and equipotent for SIRT2 as compared to the most potent p53-based peptide 6. It is clear that a negatively charged aspartic acid residue at position -1 was the reason for the lower inhibitory activities of the R-tubulin-based sequences 1-4 compared to the p53-based sequences 6-9. However, it is noteworthy that the sequence 1 is based on a SIRT2 substrate, and the studied replacements of the aspartic acid residue increased the SIRT2 inhibitory activity only 5-fold but SIRT1 inhibitory activity about 25-fold.

At the +1 position, the threonine residue of the R-tubulin- based peptide 1 was replaced by either an alanine or a leucine residue, resulting in peptides 18 and 12, respectively. For SIRT1, the replacement by alanine gave an equipotent peptide, but the replacement by leucine resulted in a clearly less potent peptide. For SIRT2, both replacements resulted in less potent peptides. In the p53-based sequences, peptide 6 with a leucine residue and peptide 22 with an alanine residue at position +1 are equipotent.

At the +2 position, the isoleucine residue of the R-tubulin- based peptide 1 was replaced by a methionine or an alanine residue, resulting in peptides 13 and 19, respectively. The replacement by a methionine residue slightly increased and the replacement by an alanine residue slightly decreased the inhibitory activity. The same trend was observed with the p53- based sequences. The +2 methionine residue increased the inhibitory activity compared to that of isoleucine (15 vs 12) or alanine (6 vs 23).

Removal of the amino acid at position +2 in the R-tubulin- based peptide 1 resulted in peptide 2, which had slightly decreased inhibitory activities for SIRT1 and SIRT2. The effect of the removal of the amino acid at position +2 was also confirmed with the p53-based sequences 6 and 7. The inhibitory activity is not significantly dependent on the presence of an amino acid at position +2. NR-Fmoc-N‹-thioacetyl-lysine and NR-acetyl-N‹-thioacetyl-
lysine have been reported to show no inhibitory activity toward SIRT1,18 but the inhibitory activities of tri- or tetrapeptides against SIRT1 and SIRT2 have not been reported before. Removal of the amino acids from the -2 and +2 positions did not affect the inhibitory activity for SIRT1 but significantly decreased the inhibitory activity for SIRT2. The R-tubulin-based tripeptide 4 and the p53-based tripeptide 9 had the IC50 values of 12.0 and 0.57 µM for SIRT1 and 175 and 151 µM for SIRT2, respectively. Removal of the amino acids from positions +1 and +2 resulted in substantial loss of activity against both enzymes. The R-tubulin-based tripeptide 5 had a significantly decreased inhibitory activity: 11.8% inhibition at 200 µM against SIRT1 and 13.9% inhibition at 200 µM against SIRT2.

Conclusions

In conclusion, the p53-based and R-tubulin-based N‹-thio- acetyl-lysine-containing tri-, tetra-, and pentapeptides are potent SIRT1 and SIRT2 inhibitors. The p53-based sequences gave overall better inhibitory activities than the R-tubulin-based sequences, mainly due to the unfavorable aspartic acid residue in position -1 of the R-tubulin sequence. The studied series shows that the correct selection of side chains is important for good inhibitory activity. In fact, the difference in the IC50 values between the most potent and the least potent pentapeptide inhibitors is 130-fold for SIRT1 and 63-fold for SIRT2. Six of the most potent SIRT1 inhibitors, 6, 8, 11, 14, 17, and 22, had the IC50 values in the range 180-330 nM, equipotent to the reference compound 24. The most potent SIRT2 inhibitor 22 had an IC50 value of 3.8 µM. These peptides are among the most potent SIRT1 and SIRT2 inhibitors published so far. In addition, tetrapeptides 3, 7, and 8 and tripeptide 9 maintained their SIRT1 inhibitory activity. As it has been claimed that it is difficult to design peptidomimetics for peptides larger than four amino acids,22 peptides 3, 7, 8, and 9 provide a promising starting point for the development of small peptidomimetic SIRT inhibitors.

Experimental Section

Manual Peptide Synthesis. Peptides 1-15, 18, 20, and 23 were synthesized manually using the Fmoc strategy with TBTU/DIPEA as the coupling reagent and Wang resin as the solid phase (Fluka). The swelled (1 h) resin-bound NR-Fmoc-amino acid (300-350 mg) was deprotected with 5 mL of 20% (v/v) piperidine in dimethyl- formamide (DMF) for 10-15 min and washed 5 times with DMF.

a The sequences were tested with free amino- and carboxy-terminals. b 95% Confidence intervals for IC50 values are given in parentheses. Each experiment was repeated at least three times. c Inhibition-% at 200 µM ( standard deviation. The IC50 value could not be determined due to the weak inhibitory activity. d Compound 24 was tested as a racemate.

The NR-Fmoc-amio acids (Fluka) in 3-5 mL of DMF were preactivated with TBTU (4 equiv)/DIPEA (10 equiv), added on the resin, shaken for 45-50 min, and washed 5 times with DMF. Cycles of deprotection and coupling with the subsequent amino acids were repeated until the desired peptide-bond resin was completed. The resin was washed with acetic acid, dichloromethane, and methanol and was evaporated under vacuum overnight. The product was cleaved with 5 mL of Reagent K (82.5% trifluoroacetic acid (TFA), 5% H2O, 5% phenol, 5% thioanisol, and 2.5% ethanedithiol; 5 mL per 300-350 mg per resin) for 45-90 min. The crude product was precipitated with cold ether, collected by centrifugation (5 min at 4000 rpm), washed yet with cold ether, and dried in vacuum.Peptide Synthesizer. Peptides 16, 17, 19, 21, and 22 were synthesized on an Apex 396 DC multiple peptide synthesizer (Advanced ChemTech, Louisville, KY) using the Fmoc strategy with HBTU/DIPEA as the coupling reagent and Wang resin and NovaSyn TGA as the solid phase (Fluka, Novabiochem, and GL Biochem (Shanghai)). The cleavage mixture was 95% TFA, 3% ethanedithiol, 1% tris-isopropylsilane, and 1% H2O.

HPLC Purification. Peptides were purified by preparative HPLC (Shimadzu LC-10AVp, Fennolab, Fenno Medical Oy), and the purity of the peptides was checked with an analytical HPLC (Agilent 1100 HPLC (Agilent Technologies, Germany) or Shimadzu LC-10Avp (Fennolab, Fenno Medical Oy)) and NMR (Bruker Avance 500, Bruker Biospin, Switzerland). The molecular weight of the peptides was confirmed by using ESI-MS (LTQ linear ion trap, Thermo Fisher Scientific or 6410 Triple Quad MS,EX 527 Agilent Technologies).