Structural Characterization of 17β-Hydroxysteroid Dehydrogenase Type 14 and Inhibitor Optimization Using Crystallography and Computational Techniques
17β-Hydroxysteroid dehydrogenase type 14 (17β-HSD14) is the latest identified subtype of 17β HSDs. In vivo this enzyme oxidizes the hydroxyl group at position 17 of estradiol (E2) and 5 androstenediol (5-diol) in the presence of NAD+ as cofactor. Two isoforms of this cytosolic protein exist that dif...
|PDF Full Text
No Tags, Be the first to tag this record!
|17β-Hydroxysteroid dehydrogenase type 14 (17β-HSD14) is the latest identified subtype of 17β HSDs. In vivo this enzyme oxidizes the hydroxyl group at position 17 of estradiol (E2) and 5 androstenediol (5-diol) in the presence of NAD+ as cofactor. Two isoforms of this cytosolic protein exist that differ only in sequence position 205: S205 and T205. So far, the protein has not been thoroughly investigated in detail and its physiological role remains unknown. Prior to this thesis, the 17β-HSD14 apoenzyme (S205) had already been crystallized. The determined structure revealed a very broad and open active site and the conserved catalytic triad and the Rossmann fold motif. However, all C-terminal tails and for some chains also amino acids in the flexible loop (189-212) were not defined in the electron density. Moreover, it is impossible to derive information regarding a potential substrate from this apo structure. Therefore, the renewed structural determination of the 17β-HSD14 apo protein as well as in complex with its cofactor and substrate was of utmost importance.
After successful establishment of the expression and purification protocols for 17β HSD14 protein, the two enzyme isoforms (S205 and T205) were characterized biochemically. The structures of the S205 apoenzyme and the binary complexes with NAD+ of both isoforms were determined. In these complex structures the flexible loop adopts a unique closed conformation differing from the apo structure. Binding of the cofactor is accompanied by a shift of the flexible loop and of the C-terminal Tyr253’ of the adjacent monomer, thereby reducing the size of the active site. The ternary complex of the enzyme with estrone (E1) and NAD+ was also determined. E1 binds to the active site in an atypical fashion, in so far as its A-ring and not the enzymatically modified position 17 close to the nicotinamide moiety of NAD+.
Enzyme inhibitors are useful tools to study the consequences of enzyme inhibition in vivo. This allows to clarify whether this enzyme may be interesting as a new drug target for a certain disease. In addition, potent and selective 17β HSD14 inhibitors may help understand the selectivity issue with other 17β HSDs. As no 17β HSD14 inhibitor was known prior to this study, the goal was to identify and optimize nonsteroidal 17β-HSD14 inhibitors. To that, a library of 17β-HSD1 and 17β HSD2 inhibitors was screened against 17β-HSD14. The most promising hit was taken as the starting point for further chemical modification applying a ligand based approach. Newly designed compounds were synthesized and subsequently tested for their 17β HSD14 inhibitory activity. Prior to this thesis, no human 17β HSD structure in complex with a nonsteroidal ligand was published. The crystal structures confirmed that the inhibitors bind to the substrate binding site and allowed to rationalize the strong affinity of these inhibitors.
Subsequently, two different structure-based strategies were pursued for inhibitor design. The first structure based modifications of the initial pyridine-based scaffold led to a ten-fold more potent inhibitor. The goal of the second structure based optimization strategy was to extend the central pyridine core to interact with the empty binding pocket adjacent to the steroid A and B-ring. The predicted binding mode was verified by co-crystal structures and the low nanomolar potency was confirmed by biophysical characterization. The new crystal structures revealed how small changes of the inhibitors affect the adopted binding mode. The characterization of the most promising 17β HSD14 inhibitors against 17β HSD1, 17β-HSD2, and 17β HSD10 revealed varying degrees of selectivity. In addition, some of these inhibitors showed very low cytotoxicity and did not interact with the multi-drug resistance protein Pgp, indicating these compounds might not be effluxed from the brain and that the risk of potential side effects is reduced. This suggests these inhibitors as tool compounds for further investigation in vivo.
To explain the selectivity profiles of the ligands towards 17β HSD14 and other 17β HSDs we conducted a structural comparison. The typical V-like shape of the binding pocket of 17β HSD14 is determined by His93 and Gln148, which are not present in 17β HSD1, 17β HSD8 and 17β HSD10. In addition, the latter three enzymes have a rather flat binding pocket. This suggests that matching the characteristic three-dimensional requirements of 17β-HSD14 and optionally addressing His93 and/or Gln148 will increase the selectivity toward this target. Such inhibitors were predicted by docking a library of about 400 17β-HSD1 and 17β-HSD2 inhibitors with GOLD followed by in vitro screening of docking hits and related compounds. Remarkably, predicted binding modes were in poor agreement with the subsequently determined crystal structures due to the adaptability of the binding pocket caused by the flexible loop.
Finally, a large fragment screening campaign by X-ray crystallography with the aim to discover new inhibitor scaffolds bound to 17β HSD14 was performed. This resulted in two fragments that could be clearly identified in the electron density. However, these fragments did not significantly inhibit 17β HSD14. In order to enhance affinity, fragment growing and fragment linking strategies were applied, resulting in two new inhibitors with better affinity than the starting fragments.
In summary, both isoforms of 17β-HSD14, S205 and T205, were characterized biochemically and structurally resulting in four new crystal structures. The first two classes of inhibitor for this enzyme were discovered and the ligands were thoroughly profiled. In addition, the structures of 12 nonsteroidal inhibitors in complex with the protein were elucidated for the first time for this protein family. The fragment screening by determining 96 fragment-soaked structures, resulted in two fragment hits that were successfully optimize culminating in two inhibitors more active than their precursor fragments.