We observe by x-ray crystallography that the dimethoxy phenyl ring of 3 disrupts the water network around the catalytic His121. Thus it is possible that if 3 prevents collapse of the tetrahedral intermediate, it could do so by perturbing the local environment around this key residue, preventing it from acting as the general acid. Although we are unable to isolate and quantify the binding interactions of 3 to the tetrahedral intermediate, it is noteworthy that the measured affinities of 3 to the Michaelis complex (,200 nM by SPR) and acyl enzyme (1.3 mM by SPR) are both weaker than the potency determined in enzymatic assays (11 nM IC50). We speculate that binding of 3 to the tetrahedral intermediate is the favored enzyme/substrate complex leading to potent inhibition. An unexpected feature of this inhibitor is the 2? orders of magnitude difference in inhibitory potency depending on thefluorophore employed in enzymatic assays, and the apparent lack of activity when fluorophore-free substrates are utilized. The computational models suggest one possible explanation for this difference, namely a polarized CH-p interaction between the paramethoxy group of 3 and the face of the orthogonal phenyl ring of the R110 dye, an interaction that is not possible with AMC-based substrates or substrates lacking a dye (e.g. native protein substrates) (Figure 7C). The importance of such CH-p interactions has been noted previously [29]. Furthermore, there appears to be either an edge-face or p-stack interaction between the phenyl ring of the inhibitor and the fluorophore aromatic ring. The remaining interaction energy difference can be explained by displacement of waters by the two extra rings of the R110, and/or additional hydrophobic interactions between the extra two rings of R110 and the protein. All of these interactions would be absent in a peptide substrate lacking a fluorophore at the P1′ position. It is known from studies on caspase-3 that prime side interactions can lead to a significant increase in inhibitory potency; for instance, the addition of a benzoxazole moiety on the prime side of the AcDEVD a-ketoaldehyde peptide inhibitor increases the potency ,300-fold against caspase-3 [30]. As for the inability of 3 to inhibit Lamin A cleavage, the presence of substrate residues more distal to the scissile bond (P5�P8) may alter the general conformation of the inhibitor binding site to disrupt the key L2?L4 interactions observed in Figure 5B. The importance of P5 for caspase-2 substrate recognition and catalysis has been described [31] and we speculate that the inhibitor binding site defined here may be altered by similar enzyme-substrate interactions.
Figure 7. Docking models of caspase-6/VEID-R110/3 ternary complex explains fluorophore-dependent potency of this series of compounds. (A) Docking model of the Michaelis-Menten complex formed between caspase-6 (light blue), VEID-R110 (green sticks) and 3 (wheat sticks). (B) Docking model of the tetrahedral intermediate between caspase-6, VEID-R110 (green sticks) and 3 (wheat sticks) with substrate covalently bound to Cys163. (C) Depiction of monovalent VEID substrates with R110 or AMC fluorophores. This class of inhibitors also shows sensitivity to the peptide sequence of the substrate, and unprecedented selectivity for caspase-6. To better understand this selectivity profile, we superposed the caspase-3/DEVD coordinates onto the caspase6/VEID/3 ternary structure (Figure S3). Three residues lining the binding site of 3 provide a structural rationale for the selectivity of these inhibitors (Cys264 and Ala269 in the L4 loop and His209 in the L3 loop); we believe that Ala269 is the primary driver of caspase selectivity (amino acids depicted in Figure 5C). Ala269 is Phe256 in caspase-3 and Phe282 in caspase-7. These larger residues would hinder compound binding by clashing with the benzyl side chain of all inhibitors from this series. Our models also explain the substrate peptide sequence sensitivity of these inhibitors. The smaller Val residue in the substrate (DEVD)2R110 would produce a weaker hydrophobic interaction between the substrate and the benzyl side chain, while the larger Trp and His residues in the substrate (WEHD)2R110 would prevent inhibitor binding by clashing with the inhibitor side chain. The substrate-dependent variation in potency minimizes the utility of these inhibitors as tools to understand target biology. This finding may also suggest that peptide surrogates used in biochemical assays have potential to contribute to misleading SAR for other series of inhibitors. This phenomenon is not specific tocaspase-6. A common assay system used to profile the activity of the histone deacetylase enzymes also incorporates a proximal fluorophore attached to the C-terminus of a tetrapeptide. The crystal structure of this Arg-His-Lys-Lys-Coumarin substrate with HDAC8 illustrates direct interactions of the fluorophore with amino acid residue side chains [32]. Several reports make claim that SIRT activation by Resveratrol is an artifact of this fluorogenic assay [33,34], although follow up work confirms the original findings [35]. Thus, it is advised that a detailed mechanistic characterization of hits, as described here, be performed early in the triage stage of lead identification campaigns, particularly when inhibitors with unusual mechanisms are found. In summary, the mechanistic and structural information described here explains the selective and substrate-specific inhibition of caspase-6 by a novel series of inhibitors. Uncompetitive inhibition is a proven strategy for other targets including MEK1/2 [36?8] and IMPDH [39,40] but because these compounds recognize a specific substrate-enzyme complex, they do not potently inhibit cleavage of other more physiologically relevant substrates. These particular inhibitors provide new insight into caspase selectivity, a topic of significant importance in drug discovery. This mechanism of uncompetitive inhibition is uniquefor any caspase family member and suggests that the discovery of inhibitors of specific, biologically relevant, enzyme-substrate complexes may be achievable. The observed binding of 3 to the acyl-enzyme when no fluorophore occupies the prime side (Figure 5 and Figure 6) suggests that elaboration of this series could lead to biologically relevant caspase-6 inhibitors. The work described herein provides a template for identification of uncompetitive caspase inhibitors as well as effective triage strategies of lead matter with novel mechanisms.