The problem: inhibiting a deacylase without a classic warhead
SIRT2 is an NAD+-dependent lysine deacylase implicated in cancer and neurodegeneration — it strips acetyl and long-chain acyl marks from substrates such as α-tubulin, NF-κB and FOXO. Drugging it selectively is hard: the human sirtuins (SIRT1–7) share a conserved catalytic core, and the most potent peptide inhibitors to date are mechanism-based — thioacyl-lysines (e.g. thiomyristoyl) that get partway through catalysis and form a stalled covalent intermediate with NAD+. That works, but a covalent NAD+-adduct is a liability for a chemical probe or a degrader payload.
Friedrich, Jung, Schutkowski and co-workers (Halle / Freiburg) ask a sharper question: can you make an acyl-lysine that binds like a substrate, aligns perfectly for catalysis, and then simply refuses to turn over — without forming any covalent intermediate?
The strategy: branch the acyl chain and stall the chemistry

SIRT2 deacylation begins when the acyl-lysine’s amide oxygen attacks the C1′ of the NAD+ ribose (nicotinamide is cleaved; His187 orchestrates the transition state). The authors reasoned that steric bumps right at the reacting carbons would let the amide dock in the catalytically competent geometry but block the collapse of the intermediate — converting a substrate into a “slow substrate” with a long residence time and minimal turnover. No thiocarbonyl, no stalled covalent adduct: a cleaner mechanism of inhibition.
The chemistry: methyl branches and cyclopropanes on a thia-acyl chain

The warhead is an Nε-acyl-lysine built on a TNFα-derived nonapeptide (with a 3-nitrotyrosine chromophore for HPLC readout of turnover). For synthetic access, a methylene in the acyl chain is swapped for a thioether — acrylic-acid HFIP esters undergo 1,4-addition of 1-dodecanethiol to give 4-thiapalmitoyl chains, which are coupled to resin-bound peptide. Branching is then installed at the reacting carbons:
- Unbranched (1): a true substrate — quantitatively deacylated within 24 h.
- Cα gem-dimethyl (4) and Cβ gem-dimethyl (5): dimethyl substitution abolishes catalysis (3% and 6% conversion over 24 h). Peptide 5 — and its truncated di-lysine version KK3 (5a) — is the lead: apparent Ki ≈ 0.9 nM, in-vitro IC50 5 nM, cellular NanoBRET EC50 13 nM.
- Cβ cyclopropyl (7): a rigid, planar bioisostere of the gem-dimethyl — sub-nanomolar IC50, among the most potent SIRT2 inhibitors in the assay.
Extending the idea, the authors show that naturally/drug-derived branched acyls — phytanoyl- and ibuprofenyl-lysine — also inhibit SIRT2, suggesting branched acylations may be endogenous slow-substrate signals.
The structural proof: aligned on NAD+, caught mid-reaction

Ternary SIRT2–peptide–NAD+ co-crystals nail the mechanism. In the SIRT2–5–NAD+ complex (PDB 8QTU, 1.80 Å), the branched acyl-lysine threads the hydrophobic channel (lined by Phe96, Phe119, Phe143/190/234 and the selectivity pocket) with NAD+ bound simultaneously and the amide nitrogen making the canonical hydrogen bond to the Val233 backbone. In the companion SIRT2–4–NAD+ structure (8QT4, 1.55 Å), the amide oxygen sits 3.3 Å from the NAD+ ribose C1′ — essentially identical to the 3.1 Å of a native myristoyl substrate (4Y6O). So minimal turnover is not caused by misalignment or amide-bond torsion; the branch stalls the chemistry after the enzyme has done everything right. Comparing SIRT2 with SIRT1/SIRT3 reveals isoform-specific acyl-chain trajectories that underlie selectivity (dimethyl branching favours SIRT2 over SIRT1).
Why it matters
- A cleaner inhibition logic for sirtuins. “Slow substrate” trades the covalent stalled intermediate of thioacyl inhibitors for a non-covalent, long-residence block — attractive for chemical probes and SIRT2-targeting degraders where a reactive NAD+ adduct is unwanted.
- Potency comes from the channel, not the backbone. Truncated di-lysine analogues (KK) keep single-digit-nanomolar potency, pinpointing the hydrophobic acyl channel as the affinity driver — a clear handle for future small-molecule mimetics.
- Cellular validation. Optimized inhibitors show low-nanomolar target engagement (NanoBRET) and induce robust α-tubulin hyperacetylation, confirming on-target SIRT2 inhibition in cells.
- A design primitive worth stealing. Steric branching at the reacting atom — gem-dimethyl or a cyclopropane bioisostere — is a general way to freeze an enzyme in a competent-but-unproductive state. Expect it to travel beyond sirtuins.
Reference: Friedrich, F.; Meleshin, M.; Kalbas, D.; Kollmer, F.; Papenkordt, N.; Mauerer, A.; Ruprecht, J.; Hilscher, S.; Zessin, M.; Schiene-Fischer, C.; Zhang, L.; Graf, L.; Sippl, W.; Einsle, O.; Jung, M.; Schutkowski, M. “Alkyl-Branched Acyl-Lysine Peptides as Potent Cell-Permeable Slow-Substrate Inhibitors of SIRT2.” J. Med. Chem. 2026. DOI: 10.1021/acs.jmedchem.6c01275.
All visuals are original to Astinova Bio Labs: warhead structures and the mechanism schematic drawn with RDKit/Matplotlib; the binding view rendered with 3Dmol.js from the deposited co-crystal PDB 8QTU. No paper figures were reused.
Astinova Bio Labs designs and makes molecules like these — warhead and mechanism-of-inhibition design, peptide and small-molecule synthesis, and structure-based analysis. Talk to us about your target.