Selective noncovalent proteasome inhibiting activity of trifluoromethyl‐containing gem‐quaternary aziridines

The ubiquitin‐proteasome pathway (UPP) represents the principal proteolytic apparatus in the cytosol and nucleus of all eukaryotic cells. Nowadays, proteasome inhibitors (PIs) are well‐known as anticancer agents. However, although three of them have been approved by the US Food and Drug Administration (FDA) for treating multiple myeloma and mantel cell lymphoma, they present several side effects and develop resistance. For these reasons, the development of new PIs with better pharmacological characteristics is needed. Recently, noncovalent inhibitors have gained much attention since they are less toxic as compared with covalent ones, providing an alternative mechanism for solid tumors. Herein, we describe a new class of bis‐homologated chloromethyl(trifluoromethyl)aziridines as selective noncovalent PIs. In silico and in vitro studies were conducted to elucidate the mechanism of action of such compounds. Human gastrointestinal absorption (HIA) and blood–brain barrier (BBB) penetration were also considered together with absorption, distribution, metabolism, and excretion (ADMET) predictions.

"post-glutamate peptide hydrolase" (PGPH) or "caspase-like" (C-L) activity ( Figure 1). [10,11] All three catalytic subunits contain an N-terminal residue represented by Thr1. The hydroxyl group of Thr1 carries out the nucleophilic attack to the peptide bond, playing a central role in the formation of the tetrahedral intermediate of the proteolytic mechanism. A water molecule then completes the degradation of the peptide, performing the hydrolysis with the release of the catalytic Thr1. [12,13] To date, three kinds of proteasomes have been investigated: the constitutive proteasome (cCP), [14][15][16] which is the most widely distributed in cells, the immunoproteasome (iCP), [17,18] and the thymoproteasome (tCP), [19] which are responsible for the antigen presentation and the positive selection of CD8 + T cells, respectively.
The employment of proteasome inhibitors (PIs) as anticancer agents is well-established nowadays. [20,21] Indeed, three PIs, bortezomib (1), [22] ixazomib (2), and carfilzomib (3) have been approved by the US Food and Drug Administration (FDA) for the treatment of multiple myeloma and mantel cell lymphoma ( Figure 2a). [23] The clinically used PIs, as well as the major part of the inhibitors reported in the literature, are covalent. They are characterized by different electrophilic warheads able to bind the hydroxyl group of the N-terminal threonine residue of the 20S catalytic β5-subunit. [24] Even if they show great efficiency in anticancer therapy, the covalent mode of action leads to severe side effects [25,26] and limited activities in solid tumors. [27] Therefore, their clinical use has been limited due to a lack of specificity and excessive reactivity and instability. On the other hand, promising diverse noncovalent inhibitors (e.g., 4, 5, and 6) have been identified, showing reduced toxicity and providing a different mechanism for solid tumors due to the rapid binding and dissociation kinetics of the noncovalent interactions ( Figure 2b). [14] Considering the results achieved so far regarding the development of noncovalent PIs [28][29][30][31] and in agreement with preliminary docking studies, the inhibitory activity of a set of bis-homologated F I G U R E 1 Schematic structure of the 26S proteasome. chloromethyl(trifluoromethyl)aziridines (CMTMAs) was evaluated toward this anticancer target. [32] 2 | RESULTS AND DISCUSSION

| Pharmacology/Biology
The preparation of the targeted CMTMAs was levered on the homologation logic of imine surrogates with nucleophilic lithium carbenoids [33,34] previously reported. [32] Compounds 8-19 ( Figure 3) were then evaluated in vitro against human 20S proteasome, and the results are reported in Table 1.
Comparing the biological activity of the CMTMAs with the previously reported chloro(trifluoromethyl)aziridines (CTMAs), [29] a clear correlation of how the substituents influence the activity was not found. Even if the two series of compounds differ only for a methylene unit, a different percentage of inhibition was found for F I G U R E 3 Chemical structures of compounds 8-19 and reference compound 20. the derivatives embodied with an anthracenyl (16), pyrrolidinyl (19), diazophenyl (18), and morpholinyl (10) moieties. Indeed, for the CTMAs the best activity was observed for the anthracene-(IC 50 = 13.6 µM) and diazophenyl-(IC 50 = 14.1 µM) bearing compounds, [29] whereas for the CMTMAs the presence of a phenyl group in ortho position (12), a morpholinyl (10), and a pyrrolidinyl (19) functionalities led to the most active molecules. Neither the position nor the presence of electron-withdrawing or electron-donating groups clearly explained the variation of the activity. Despite the unclear correlation between the various substituents linked to the aziridine moiety, comparing the results obtained from the docking studies carried out in the previous work and this one, it appears that the interactions with the various residues of the receptor site are almost conserved, albeit with different portions of the molecules, for example, with residues Ala20, Ala27, Ser28, and Glu132. Each of these residues interacts with both the previous ligands and those reported in this work, albeit through interactions of a different nature.
Preliminary assays (i.e., screening at 20 μM) performed on various cysteine and serine proteases (i.e., human cathepsins B and L, and bovine pancreatic α-chymotrypsin) to assess off-target reactivity, as well as on proteasome β1 and β2 subunits, did not display any inhibitory activity (inhibition range 0%-5%), demonstrating a marked selectivity of compounds 8-19 against proteasome β5 subunits.

| Molecular modeling studies and absorption, distribution, metabolism, and excretion (ADMET) prediction
In silico studies were performed using AutoDock Vina implemented in the yet another scientific artificial reality application (YASARA) software. The docking studies were performed on all the protonation states of the compounds at pH 7.4, previously calculated using the Marvin software. The peptide Suc-Leu-Leu-Val-Tyr-AMC (20, Figure 3) was used as a reference proteasome substrate. The calculated inhibition constants are reported in Table 1 and, for the best-performing compounds, 10, 12, and 19, agree well with the experimental ones. Figure 4 shows the 3D and 2D poses of the three best compounds, 10, 12, and 19. From the interactions dictated at the poses within the receptor pocket, it is evident that the aromatic ring of these compounds acts as the anchor portion; in fact, we note in all three compounds π-cation and π-anion interactions, in particular, compound 10 establishes these interactions with residues Glu132, Glu134, and Arg137, compound 12 with residues Asp126 and Arg137, and compound 19 with residue Arg137. The presence of other similar interactions, such as hydrophobic, π-alkyl, and with halogens, reflects the similar inhibitory activity of the three compounds toward the proteasome. For example, compounds T A B L E 1 Docking scores and predicted K i on proteasome subunit β5 of Saccharomyces cerevisiae (PDB ID: 4HRD) as well as inhibitory activities of compounds 8-19 compared with reference [20].

| CONCLUSION
In conclusion, the inhibitory activity toward the proteasome of a new class of aziridine derivatives has been reported. In silico studies predicted a noncovalent mechanism of action and promising "drug-like" features as anticancer agents. In vitro enzymatic assays revealed a selective inhibition over the β5 proteasome subunits. Moreover, preliminary studies were conducted on several cysteine and serine proteases (i.e., human cathepsins B and L, and bovine pancreatic α-chymotrypsin) to assess off-target reactivity and did not show any inhibitory activity. as we previously reported. [43] MeLi-LiBr (1.5 M ethereal solution) was titrated immediately before use, according to the literature. [44] The InChI codes of compounds 8-19, together with some biological activity data, are provided as Supporting Information.
4.1.2 | General procedure for the synthesis of trifluoroacetimidoyl chlorides [45] To a solution of Ph 3 P (3.0 equiv) in 1,2-dichloroethane, CCl 4 (4.0 equiv), The inhibitory activity of the compounds was evaluated by a standard fluorometric method. [36] Human 20S proteasome, isolated and purified from erythrocytes, was obtained from a commercial source

| Molecular docking and dynamics
The studied molecules were drawn using Marvin Sketch and were downloaded from the Protein Data Bank (www.rcsb.org). Only chains K and L were used. Water molecules were also removed. All amino acid residues were kept rigid, whereas all single bonds of ligands were treated as fully flexible for both proteins. A 10 Å simulation cell around all atoms of the cocrystallized ligand was used.
AMBER 14 force field was used for the simulation. [29] The MD simulations of the complexes were performed with the YASARA structure package according to our previously reported procedures. [46,47]