We solved and refined the crystal structure of SaDHNA, and assigned SaDHNA to be a member of the HotDog fold class I superfamily of proteins, where all β-strands pointing outwards and the long main α-helix pointing towards to the core of the structure. SaDHNA associates as a dimer of homodimers, with a face-to-face (or helix-to-helix) conformation. The “HotDog” fold was first described by Leesong and co-workers for a thiol ester dehydratase-isomerase from E. coli, named FabA (pdb code 1MKA)24. Since after, a number of proteins possessing the “HotDog” fold were identified and described for several organisms25–29. Despite the absence of a consensus sequence and a sequence identity ranging only between 10 and 20%, the overall fold and N- or C-termini secondary structure elements are similar and particular characteristic for members of the “HotDog” fold protein family. Although overall low sequence identity among all thioesterases was observed in sequence alignments, the secondary structure elements are mostly well conserved. Therefore, it is expected that the active site architecture of those enzymes remains homologous as well. Investigations involving protein superfamilies have shown that the mode of catalysis, the active site location, as well as residues involved in substrate recognition and catalysis are indeed frequently conserved among these evolutionarily related proteins30–32. This explains the relative low sequence homology between SaDHNA and other known thioesterases, even though the quaternary structures and in particular the positions of the respective active sites and residues within the interface region forming the dimers are homologous.
Thioesterases from E. coli EcYbgC as well as Haemophilus influenzae HiYbgC revealed to be more active for short acyl chain substrates, in contrast to Helicobacter pylori HpYbgC, which showed to be more active towards long acyl chains, e.g. palmitoyl- and stearoyl-CoA33. Although all enzymes possess overall the same fold, HotDog fold class I, EcYbgC, HiYbgC, HpYbgC as well as SaDHNA have some differences in their structure, which explains a divergence in the substrate specificity. In fact, we observed the presence of a long tunnel associated with the binding site of the acyl moiety of the substrate for HpYbgC, which is absent in the HiYbgC. We detected a similar situation for SaDHNA, which is more active towards long acyl chains (stearoyl-CoA) in comparison to a short chain (crotonyl-CoA). The structure of SaDHNA structure reveals that activity towards long acyl chains can be associated with the presence of an extended tunnel with hydrophobic nature, involving the residues Leu35, Ile38, Tyr45; Met48, Leu122, Tyr125 and Phe126 (Figure S2), where the long acyl chain of a stearoyl-CoA may point towards this hydrophobic tunnel and is stabilized mostly through hydrophobic non-covalent interactions26.
Considering also all till now known homologous thioesterase structures, investigations performed for the native Ps4HBT and a D17N Ps4HBT mutant revealed functional residues involved in the thioesterase activity of SaDHNA. The residue His23 positioned in the N-terminus of the main α-HD helix (Ƞ2 and α1), in substitution of Tyr24 in Ps4HBT, is likely responsible for the polarization of the thioester carbonyl carbon group by a hydrogen bond with the imidazole ring of the histidine sidechain. The carbonyl sidechain from the closest residue, Asp16 (positioned within the connecting β-turn loop between Ƞ1 and the main α-HD helix), may act as a nucleophile during the thioester bond cleavage. In fact, mutagenesis we performed for the residue Asp16 (D16A) in SaDHNA truthfully altered the catalysis rate of the SaDHNA thioesterase, resulting in a 300-fold decrease of the hydrolysis. This important result highlights the significance of this particular residue for thioesterase activity of SaDHNA. Nevertheless, the mutagenesis D16A was not sufficient to completely suspend the SaDHNA thioesterase activity, showing also a divergent result compared to previously reported thioesterases34–36. This controversial result about the function of an aspartic acid in the catalysis comparing Ps4HBT, Orf6 thioesterase and SaDHNA indicates that the enzymatic mechanism of SaDHNA may not occur according to a covalent catalysis, as observed for Ps4HBT, Orf6 thioesterase but more probably by a catalysis involving a water molecule25,37,38. The mechanism involving the covalent catalysis is based on the proposed anhydride intermediate formation resulting from the nucleophilic attack performed by an acidic residue (aspartic or glutamic acid) in the active site, releasing the CoA thiol in the absence of a water molecule. On the other hand, the general base catalysis mechanism requires the directed activation of a water molecule which consequently acts as a nucleophile on the CoA thiol group39,40. Indeed, by a careful inspection of the native SaDHNA structure, there is a water molecule close to the proximity of BCA substrate and the sidechain of residue Asp16, inside of the tunnel of the active side, which, in fact, could support the general base mechanism in SaDHNA. Interestingly, although the mutagenesis of the aspartic acid was sufficient to completely inactivate the thioesterases from P. profundum and Pseudomonas 4HBT, for SaDHNA this mutant had substantially decreased activity, but was not fully inactivate. This surprising result indicates that SaDHNA might use the aspartic acid together with another residue as an alternative during thioesterase activity. Hydrolases, including thioesterases, frequently use the Ser-His-Asp catalytic triad to perform a bond cleavage. Within this triad, aspartic acid is an important activator during nucleophile attack, followed by serine and histidine residues. The imidazole ring of histidine possesses a pKa of approximately 6 to 7 and allows this residue to switch between protonated and unprotonated states at a physiological pH. This individual property enables histidine to participate in general acid-base catalysis and to enhance the nucleophilicity of the hydroxyl and thiol groups. Protonated nitrogen of the imidazole ring can act as a general acid while unprotonated nitrogen acts as nucleophile, and consequently, performs as a general base41. In absence of aspartic acid in the active site, basically nitrogen from the imidazole ring of His23 might abstract a proton of the nucleophile (a water molecule of SaDHNA native structure closed to the proximity of BCA substrate) and henceforward induce the nucleophilic attack on the carbonyl carbon of the polarized substrate (electrophile) as modelled in Figure S3. On the other hand, complementary mutagenesis we performed for the residue Glu31 (E31N) resulted in no detectable activity. The orientation of the uncharged polar sidechain of an asparagine residue might interfere with the binding of substrate in the active site, in which Glu31 is more likely to act only as a supportive residue required for the substrate binding and not being involved in the thioesterase activity. Initial evidences obtained from our investigations support the general base catalysis.
Finally, selected peptide inhibitors were successfully screened by assays and docking studies using the atomic structure of SaDHNA. In general, stable peptide-protein interactions involve hydrogen bonds, as well as complementary interactions, such as hydrophobic van der Waals interactions, leading to a high selectivity and binding affinity42. Although designed peptide inhibitors based on atomic structures have demonstrated effectiveness to inhibit bacterial protein synthesis43 and transcription44, so far, no attempt has been made to rationally design peptide inhibitors of a DHNA thioesterase. In this study we designed two peptide ligands. During docking analysis Pep-1 was predicted to bind inside of the SaDHNA active site, producing a stable interaction via hydrogen bonds, as well as noncovalent interactions and via aromatic ring stacking (π stacking), which may also contribute to the peptide stability inside the binding pocket. This stable interaction might prevent the substrate binding by blocking the active site entrance for other substrates. On the other hand, Pep-2 was predicted not to bind inside the active site, but on the surface of SaDHNA. In contrast to traditional drug target sites, e.g. enzyme active sites, protein surface regions are usually more flat and mostly have less well-defined binding pockets to bind small molecules or peptides45,46. Thoden and co-workers47 observed that the coenzyme A ribose of both 4-hydroxybenzoyl-CoA substrates and the 4-hydroxyphenacyl-CoA inhibitor were positioned in a cleft located on the solvated surface of the dimer. This important observation suggests that the interaction of Pep-2 with the SaDHNA surface region might interfere with the nucleotide moiety binding of the substrate and is reflected in the overall thioesterase activity47,48. Our spectroscopy assay data obtained support the inhibitory properties of the rationally designed peptides Pep-1 and Pep-2, which might be considered as potential lead compounds for further investigations to provide more insights about potential SaDHNA inhibition mechanisms.