Table of Contents:
Preliminary modeling studies suggested that a 2,3,4,7,-tetrahydro-1H-azepine scaffold could provide a promising starting point for the design of putative inhibitors of the aspartic protease plasmepsin II (Plm II). The azacycle was primarily supposed to address the two aspartates of the catalytic dyad by its basic nitrogen atom. Based on this scaffold, optimizations through structural variations improved inhibitor binding with affinities in the nanomolar range.
(rac)-3,5-bis-hydroxymethyl-2,3,4,7-tetrahydro-1H-azepine was selected as the starting point due to the fact that the primary hydroxyl functionalities of this core structure could readily be esterified in order to appropriately address the specificity pockets of Plm II.
In a first design cycle, a phenyl moiety for addressing the S1-subpocket of Plm II turned out to be a suitable substituent for further optimization. For the S2´-subpocket, a p-NH2-phenyl moiety was chosen to be favourable by docking studies.
A second and third design cycle was performed in which the substituent for the S1-pocket was optimized keeping the previously selected P2´-substituent unchanged. Via this modification, the binding affinity could be further increased by a factor of 700 towards Plm II and by a factor of 140 against Plm IV, compared to the initial structure. Furthermore, the affinities of these compounds towards human cathepsin D were determined. Notably, one of these synthesized compounds inhibited Plm II and Plm IV equally well and, moreover, also displayed selectivity over human cathepsin D.
Recent studies in our group revealed a certain tolerance concerning the linker length of the acceptor groups adressing the so-called flap region. Therefore, we decided to establish a new synthetic strategy leading to an azepine scaffold which, on the one hand, enabled us to synthesize compounds with different linker lengths and, on the other hand, allowed the introduction of additional substituents designated to address the S1´- and the S2-pocket of Plm II, respectively.
However, the corresponding compounds with a shortened acceptor distance again showed affinities in the micromolare range and hence no significant activity improvement. Therefore, only a simple reduction of the linker length leading to a shorter C=O distance to the flap region does not seem to have any significant impact on the inhibitory activity of these compounds. Subsequently, we tried to enhance the binding affinity through introduction of further substituents to simultaneously address the S1´-pocket and the S2-pocket.
The main synthetic challenge regarding our new core structure was to synthesize inhibitors adressing three specificity pockets of Plm II and, in addition, bearing different linker lengths. The appropriate 3-armed inhibitor with shorter C=O distance exhibited an affinity of 11.5 µM against Plm II whereas the corresponding derivative addressing only two specificity pockets showed a Ki of just 54 µM. Thus, a 5-fold increase in binding affinity could be achieved.
All inhibitors synthesized in the second design cycle with elongated acceptor distance addressing three binding pockets of the enzyme displayed affinity values in the lower micromolare range leading to the conclusion that at least in our case the introduction of a third substituent does not have any favourable influence on binding.
Nevertheless, we were able to obtain a crystal structure of one of these inhibitors in complex with the aspartic protease endothiapepsin by X-ray crystallography. Through this crystall structure we were able to verify our previous assumption that the two aspartates of the catalytic dyad are indeed addressed by the protonated endocyclic nitrogen atom of our seven-membered azacycle. Furthermore, only the R-enantiomer of the synthesized racemic inhibitor mixture could be observed in the complex. The crystal structure also confirmed our modeling studies suggesting the p-NH2 moiety establishing H-bonding to the S2´-pocket and the C=O moiety of the carboxamide group pointing towards the flap region. Unexpectedly, the p-bromo moiety points towards the S2-pocket and the p-NH2-phenylmetyhlene substituent, however, misses the flap region. These findings might account for the lack of improvement of the binding affinity by introducing a third substituent.
The fairly high sequence homology between endothiapepsin and plasmepsin allows us to draw relevant conclusions from the obtained crystal structure for further structural optimizations with respect to affinity towards the plasmepsins as well as selctivity over human cathepsin D.