Synthesis and Biochemical Evaluation of Warhead-Decorated Psoralens as (Immuno)Proteasome Inhibitors

The immunoproteasome is a multicatalytic protease that is predominantly expressed in cells of hematopoietic origin. Its elevated expression has been associated with autoimmune diseases, various types of cancer, and inflammatory diseases. Selective inhibition of its catalytic activities is therefore a viable approach for the treatment of these diseases. However, the development of immunoproteasome-selective inhibitors with non-peptidic scaffolds remains a challenging task. We previously reported 7H-furo[3,2-g]chromen-7-one (psoralen)-based compounds with an oxathiazolone warhead as selective inhibitors of the chymotrypsin-like (β5i) subunit of immunoproteasome. Here, we describe the influence of the electrophilic warhead variations at position 3 of the psoralen core on the inhibitory potencies. Despite mapping the chemical space with different warheads, all compounds showed decreased inhibition of the β5i subunit of immunoproteasome in comparison to the parent oxathiazolone-based compound. Although suboptimal, these results provide crucial information about structure–activity relationships that will serve as guidance for the further design of (immuno)proteasome inhibitors.


Introduction
In mammals, most intracellular proteins are destined for degradation, which involves the proteasome, a multiprotease complex [1][2][3]. The 26S proteasome represents the heart of the ubiquitin-proteasome system that is responsible for the maintenance of protein homeostasis and the regulation of various cellular processes [4][5][6]. It is a nucleophilic hydrolase with N-terminal Thr1 acting as a nucleophile to cleave the peptide bond of proteins [7]. The 26S proteasome is comprised of a 20S core particle (CP) and 19S regulatory units. The 20S core is a 720 kDa large barrel-shaped structure assembled of four stacked rings, each consisting of seven subunits. The two outer α rings provide structural integrity and act like "gates" allowing the entry of unfolded proteins to the two inner β rings, which contain three catalytically active subunits responsible for proteolysis of substrates [8]. Subunit β1 shows caspase-like activity, subunit β2 trypsin-like activity, whereas subunit β5 exhibits chymotrypsin-like activity [9,10]. There are three individual CP types: the constitutive proteasome (cCP), which is expressed in all eukaryotic cells, the thymoproteasome (tCP) [11], which is exclusive to cortical thymic epithelial cells, and the immunoproteasome (iCP) [12], which is expressed in cells of hematopoietic origin, but can also be induced in other tissues. Namely, the induction of iCP in other cell types is possible during acute immune and inflammatory responses [13][14][15]. Exposure to inflammatory factors, such as tumor necrosis factor α and interferon-γ causes the expression of the iCP active subunits β (designated as β1i, β2i, β5i), which replace their constitutive counterparts [12,16]. such as tumor necrosis factor α and interferon-γ causes the expression of the iCP active subunits β (designated as β1i, β2i, β5i), which replace their constitutive counterparts [12,16].
Increased expression of cCP and iCP can lead to a number of diseases. These include many types of cancer, infections, inflammatory and autoimmune diseases (Crohn's disease, ulcerative colitis, hepatitis, and rheumatoid arthritis), as well as neurological disorders [17][18][19][20][21][22][23]. The cCP and the iCP therefore represent validated targets for the design of new pharmacologically active compounds [24][25][26][27]. The druggability of both CPs is clearly represented by the clinically used covalent inhibitors bortezomib, carfilzomib, and ixazomib, which are used for the treatment of multiple myeloma and mantle-cell-lymphoma [27]. Selective inhibition of the iCP's β5i [28] subunit or simultaneously acting on β1i and β5i catalytic activities [29,30] are both approaches that are being investigated in the treatment of autoimmune and inflammatory diseases. In addition, such strategy should cause fewer adverse effects, as the expression of iCP is induced during the course of disease processes [31,32]. By avoiding cCP inhibition, the protein degradation would thus not be inhibited in most eukaryotic cells.
The most advanced iCP inhibitors that are frequently utilized in functional studies of iCP inhibition are represented in Figure 1. Please note that only a selected number of derivatives is depicted; namely, the most studied β5i-selective inhibitor PR-957 [28], β1i and β5i dual inhibitors KZR-616 [29] and 'compound 22′ [33], as well as the most selective β5i inhibitor DPLG-3 [34]. Structurally, these compounds all possess a peptidic backbone. Moreover, the former three compounds are all endowed with an electrophilic warhead, which reacts with the catalytic Thr1 of the proteasome subunits to form a covalent bond and to confer improved inhibition [24]. For a more thorough overview on subunit-selective iCP inhibitors, the reader is referred to recent reviews [32,35].
Because peptidic compounds, such as bortezomib and carfilzomib, are prone to poor metabolic stability and low bioavailability due to the unfavorable physico-chemical characteristics [36][37][38], there is a need to develop inhibitors with non-peptidic scaffolds. Despite being significantly less represented, there were some recent reports on non-peptidic inhibitors of the iCP (mostly inhibiting the β5i subunit) and the representative compounds are shown in Figure 2 [39][40][41][42][43][44]. As with peptidic compounds, irreversible inhibitors of nonpeptidic nature can be obtained through structure-guided optimization, whereby an electrophilic warhead is properly positioned onto the structure of the non-covalently binding scaffold [45]. An essential prerequisite for this strategy to work is that the position of the electrophilic moiety allows the formation of the covalent bond between the inhibitor and the catalytic Thr1. Structures of the most studied iCP-selective peptidic inhibitors. For a more thorough overview on subunit-selective iCP inhibitors, the reader is referred to recent reviews [32,35].
Because peptidic compounds, such as bortezomib and carfilzomib, are prone to poor metabolic stability and low bioavailability due to the unfavorable physico-chemical characteristics [36][37][38], there is a need to develop inhibitors with non-peptidic scaffolds. Despite being significantly less represented, there were some recent reports on non-peptidic inhibitors of the iCP (mostly inhibiting the β5i subunit) and the representative compounds are shown in Figure 2 [39][40][41][42][43][44]. As with peptidic compounds, irreversible inhibitors of non-peptidic nature can be obtained through structure-guided optimization, whereby an electrophilic warhead is properly positioned onto the structure of the non-covalently binding scaffold [45]. An essential prerequisite for this strategy to work is that the position of the electrophilic moiety allows the formation of the covalent bond between the inhibitor and the catalytic Thr1.
Recently, we discovered non-peptidic and β5i-selective inhibitors with a central psoralen core [39]. The most potent non-covalent inhibitor obtained during structure-activity relationship (SAR) studies possessed a phenyl substituent at position 4 (see Figure 3 for psoralen atom numbering). This compound was also transformed into two potent irreversible covalent inhibitors by adding electrophilic warheads at position 3, i.e., succinimidyl ester and oxathiazolone. Of these two compounds, the oxathiazolone-based inhibitor showed the most promising inhibitory characteristics ( Figure 2, 'compound 42 ) as it was a potent and selective iCP inhibitor [39]. It was demonstrated previously that oxathiazolones inhibit iCP via cyclocarbonylation of the β-OH and α-NH 2 of the active site Thr1 [41]. Nevertheless, this structural fragment is deemed hydrolytically unstable making it less suitable for further development [41]. This fact prompted us to investigate other possible warheads that could be attached at the same position of the psoralen core. Previously, we already determined that acrylamides and nitrile-based warheads led to worse inhibition of the iCP [39]. However, to further map the warhead chemical space attached onto the psoralen core, we prepared a new focused set of compounds with different electrophilic fragments attached at position 3 (Figure 3), and evaluated their influence on the inhibition of all six catalytic subunits of both CPs. The selection of warheads in this study was based both on previously well described Thr targeting warheads (e.g., vinyl sulfones, α',β'-epoxyketones) [24], as well as on biologically less represented electrophilic moieties. In addition, to minimize the influence of non-covalently binding portion of the molecule on overall inhibitory potency, we used the same core compound with a phenyl substituent at position 4 . Recently, we discovered non-peptidic and β5i-selective inhibitors with a central psoralen core [39]. The most potent non-covalent inhibitor obtained during structure-activity relationship (SAR) studies possessed a phenyl substituent at position 4′ (see Figure 3 for psoralen atom numbering). This compound was also transformed into two potent irreversible covalent inhibitors by adding electrophilic warheads at position 3, i.e., succinimidyl ester and oxathiazolone. Of these two compounds, the oxathiazolone-based inhibitor showed the most promising inhibitory characteristics ( Figure 2, 'compound 42′) as it was a potent and selective iCP inhibitor [39]. It was demonstrated previously that oxathiazolones inhibit iCP via cyclocarbonylation of the β-OH and α-NH2 of the active site Thr1 [41]. Nevertheless, this structural fragment is deemed hydrolytically unstable making it less suitable for further development [41]. This fact prompted us to investigate other possible warheads that could be attached at the same position of the psoralen core. Previously, we already determined that acrylamides and nitrile-based warheads led to worse inhibition of the iCP [39]. However, to further map the warhead chemical space attached onto the psoralen core, we prepared a new focused set of compounds with different electrophilic fragments attached at position 3 ( Figure 3), and evaluated their influence on the inhibition of all six catalytic subunits of both CPs. The selection of warheads in this study was based both on previously well described Thr targeting warheads (e.g., vinyl sulfones, α',β'-epoxyketones) [24], as well as on biologically less represented electrophilic moieties. In addition, to minimize the influence of non-covalently binding portion of the molecule on overall inhibitory potency, we used the same core compound with a phenyl substituent at position 4′.   [39] was the most selective irreversible β5i subunit inhibitor from the initial series of psoralen-based inhibitors. It represents the parent compound for studies in this manuscript.
Recently, we discovered non-peptidic and β5i-selective inhibitors with a central psoralen core [39]. The most potent non-covalent inhibitor obtained during structure-activity relationship (SAR) studies possessed a phenyl substituent at position 4′ (see Figure 3 for psoralen atom numbering). This compound was also transformed into two potent irreversible covalent inhibitors by adding electrophilic warheads at position 3, i.e., succinimidyl ester and oxathiazolone. Of these two compounds, the oxathiazolone-based inhibitor showed the most promising inhibitory characteristics ( Figure 2, 'compound 42′) as it was a potent and selective iCP inhibitor [39]. It was demonstrated previously that oxathiazolones inhibit iCP via cyclocarbonylation of the β-OH and α-NH2 of the active site Thr1 [41]. Nevertheless, this structural fragment is deemed hydrolytically unstable making it less suitable for further development [41]. This fact prompted us to investigate other possible warheads that could be attached at the same position of the psoralen core. Previously, we already determined that acrylamides and nitrile-based warheads led to worse inhibition of the iCP [39]. However, to further map the warhead chemical space attached onto the psoralen core, we prepared a new focused set of compounds with different electrophilic fragments attached at position 3 ( Figure 3), and evaluated their influence on the inhibition of all six catalytic subunits of both CPs. The selection of warheads in this study was based both on previously well described Thr targeting warheads (e.g., vinyl sulfones, α',β'-epoxyketones) [24], as well as on biologically less represented electrophilic moieties. In addition, to minimize the influence of non-covalently binding portion of the molecule on overall inhibitory potency, we used the same core compound with a phenyl substituent at position 4′.  . Schematic representation of the work described in this study. The numbering system for the psoralen ring is shown for clarity, as well as general nomenclature for the warhead moieties used.

Syntheses of 3-Substituted Psoralens
To prepare 3-allyl-substituted psoralen, ethyl acetoacetate was used as a starting material (Scheme 1). It was first alkylated using NaH as a base to obtain compound 1, which was subjected to Pechmann reaction conditions to yield 7-hydroxycoumarin derivative 2. After OH group alkylation with 2-bromoacetophenone, the final allyl-substituted compound 4 was obtained by base-catalyzed condensation of the coumarin derivative 3 into psoralen ring (Scheme 1). A compound with 3-vinyl-based warhead attached at position 3 (compound 7) was obtained via a similar route. The crucial intermediate 7-hydroxy-4-methyl-4-vinyl-2H-chromen-2-one (5) was obtained in high yield by heating resorcinol derivative and crotonyl chloride at 60 • C in acetone. This was followed by a 2-bromoacetophenone-mediated alkylation and cyclization into psoralen yielding compounds 6 and 7, respectively.

Syntheses of 3-Substituted Psoralens
To prepare 3-allyl-substituted psoralen, ethyl acetoacetate was used as a starting material (Scheme 1). It was first alkylated using NaH as a base to obtain compound 1, which was subjected to Pechmann reaction conditions to yield 7-hydroxycoumarin derivative 2. After OH group alkylation with 2-bromoacetophenone, the final allyl-substituted compound 4 was obtained by base-catalyzed condensation of the coumarin derivative 3 into psoralen ring (Scheme 1). A compound with 3-vinyl-based warhead attached at position 3 (compound 7) was obtained via a similar route. The crucial intermediate 7-hydroxy-4methyl-4-vinyl-2H-chromen-2-one (5) was obtained in high yield by heating resorcinol derivative and crotonyl chloride at 60 °C in acetone. This was followed by a 2-bromoacetophenone-mediated alkylation and cyclization into psoralen yielding compounds 6 and 7, respectively. Scheme 1. Synthesis of compounds with allyl (4) and vinyl (7)  Compounds 4 and 7 were further used as synthons to prepare derivatives with other electrophilic moieties at position 3 (Scheme 2). The former was used in a Wacker-type oxidation of the terminal olefin by the combination of Pd(OAc)2 and Dess-Martin periodinane to prepare the compound with ethyl methyl ketone moiety, i.e., compound 8. The vinyl-substituted derivative 7 was a starting point for three different warhead-decorated psoralens, namely vinyl sulfone 9 (via NH4I-induced sulfonylation of vinyl at position 3 with DMSO), 3-bromo-4,5-dihydroisoxazole 10 [46] (via cycloaddition of the alkene with 1,1,-dibromoformaldoxime), and pinacolate ester 11 (via transition-metal-free synthesis of alkylboronate from vinyl and bis(pinacolato)diboron) (Scheme 2). Scheme 1. Synthesis of compounds with allyl (4) and vinyl (7) warheads attached at position 3 of the psoralen ring. Reagents and conditions: (a) allyl bromide, NaH (60%), THF, 0 Compounds 4 and 7 were further used as synthons to prepare derivatives with other electrophilic moieties at position 3 (Scheme 2). The former was used in a Wacker-type oxidation of the terminal olefin by the combination of Pd(OAc) 2 and Dess-Martin periodinane to prepare the compound with ethyl methyl ketone moiety, i.e., compound 8. The vinylsubstituted derivative 7 was a starting point for three different warhead-decorated psoralens, namely vinyl sulfone 9 (via NH 4 I-induced sulfonylation of vinyl at position 3 with DMSO), 3-bromo-4,5-dihydroisoxazole 10 [46] (via cycloaddition of the alkene with 1,1,-dibromoformaldoxime), and pinacolate ester 11 (via transition-metal-free synthesis of alkylboronate from vinyl and bis(pinacolato)diboron) (Scheme 2).  The synthesis of 3-propanal-substituted psoralen 15 was initiated by a coumarin derivative 12 possessing ethyl propionate moiety at position 3 (Scheme 3). The acidic hydrolysis yielded propanoic acid 13, which was transformed into aldehyde derivative 14 by first forming an acid chloride, followed by in situ reduction with hydrogen gas using Pd/BaSO 4 as a catalyst. Interestingly, an attempt to prepare α-ketoaldehyde (which is a known Thr-targeting warhead [24]) from compound 15 by Riley oxidation with SeO 2 resulted in the formation of α,β-unsaturated aldehyde derivative 16 (Scheme 3, Figures 4 and 5). cheme 2. Synthesis of compounds 8, 9, 10, and 11 with ketone, vinyl sulfone, 3-bromo-4,5-dihydroisoxazole, and pinacote ester, respectively, as warheads. Reagents and conditions: (a) Dess-Martin periodinane, Pd(OAc)2, CH3CN, H2O, 50 The synthesis of 3-propanal-substituted psoralen 15 was initiated by a coumarin derivative 12 possessing ethyl propionate moiety at position 3 (Scheme 3). The acidic hydrolysis yielded propanoic acid 13, which was transformed into aldehyde derivative 14 by first forming an acid chloride, followed by in situ reduction with hydrogen gas using Pd/BaSO4 as a catalyst. Interestingly, an attempt to prepare α-ketoaldehyde (which is a known Thr-targeting warhead [24]  To further confirm the structure of α,β-unsaturated aldehyde 16, two-dimensional NMR experiments correlation spectroscopy (COSY) and Nuclear Overhauser effect spectroscopy (NOESY) were recorded. In the COSY spectrum ( Figure 4), a clear correlation between the aldehyde proton CHO and the adjacent C2′-H was observed. In addition, the NOESY experiment showed a coupling between the CH3 protons and C4-H and C3′-H ( Figure 5).     The fact that the most advanced selective iCP inhibitors and also carfilzomib, which is a marketed cCP and iCP inhibitor, possess an α',β'-epoxyketone fragment as the Thrtargeting warhead, encouraged us to prepare two such psoralen-based compounds To further confirm the structure of α,β-unsaturated aldehyde 16, two-dimensional NMR experiments correlation spectroscopy (COSY) and Nuclear Overhauser effect spectroscopy (NOESY) were recorded. In the COSY spectrum ( Figure 4), a clear correlation between the aldehyde proton CHO and the adjacent C2 -H was observed. In addition, the NOESY experiment showed a coupling between the CH 3 protons and C4-H and C3 -H ( Figure 5).
The fact that the most advanced selective iCP inhibitors and also carfilzomib, which is a marketed cCP and iCP inhibitor, possess an α',β'-epoxyketone fragment as the Thr-targeting warhead, encouraged us to prepare two such psoralen-based compounds (Scheme 4). Both 20 and 21 were synthesized from the corresponding precursors 17, 18, and 19 by a HATU-mediated amide bond formation (Scheme 4). To prepare 3-azetidin-2-one-substituted psoralen 24, a previously synthesized compound 22 was used as a crucial intermediate. It was first N-acylated with 3-bromopropanoyl chloride to yield 3-bromopropanamide 23, and then cyclized into the β-lactam ring by using NaOtBu as a base (Scheme 5).  of compounds 17, 18, and 19 were based on previously described procedures [39], compound 17; [47], compound 18; [48], compound 19. All spectral data ( 1 H-NMR, HRMS) corresponded well to the original reports.
To prepare 3-azetidin-2-one-substituted psoralen 24, a previously synthesized compound 22 was used as a crucial intermediate. It was first N-acylated with 3-bromopropanoyl chloride to yield 3-bromopropanamide 23, and then cyclized into the β-lactam ring by using NaOtBu as a base (Scheme 5).

Biochemical Evaluation
The target compounds were evaluated for their inhibitory potencies on both CPs (Table 1) using subunit selective fluorogenic substrates (for details, see Materials and Methods Section). The data were calculated as residual activities (RAs) of individual subunits of CPs in the presence of 1 µM of each compound. This concentration was used due to poor solubility of all final compounds at higher concentrations, emphasizing the need for development of inhibitors with improved solubility. The previously described oxathiazolone derivative 'compound 42′ and carfilzomib were used as positive control using the same concentration (1 µM) to enable a better comparison between compounds. Table 1. Inhibitory potencies of compounds against all catalytically active subunits (β5i, β2i, and β1i) of the iCP and against all catalytically active subunits (β5, β2, β1) of the human cCP. In the assays, the following substrates were used: Suc-LLVY-AMC for β5i and β5; Boc-LRR-AMC for β2i and β2; Ac-PAL-AMC for β1i; Ac-nLPnLD-AMC for β1.

Biochemical Evaluation
The target compounds were evaluated for their inhibitory potencies on both CPs (Table 1) using subunit selective fluorogenic substrates (for details, see Materials and Methods Section). The data were calculated as residual activities (RAs) of individual subunits of CPs in the presence of 1 µM of each compound. This concentration was used due to poor solubility of all final compounds at higher concentrations, emphasizing the need for development of inhibitors with improved solubility. The previously described oxathiazolone derivative 'compound 42 and carfilzomib were used as positive control using the same concentration (1 µM) to enable a better comparison between compounds. Table 1. Inhibitory potencies of compounds against all catalytically active subunits (β5i, β2i, and β1i) of the iCP and against all catalytically active subunits (β5, β2, β1) of the human cCP. In the assays, the following substrates were used: Suc-LLVY-AMC for β5i and β5; Boc-LRR-AMC for β2i and β2; Ac-PAL-AMC for β1i; Ac-nLPnLD-AMC for β1. Given the fact that all assayed compounds possessed the same non-covalently binding portion, we were able to thoroughly assess the contributions of attached warheads to the inhibition of all catalytically active subunits of iCP and cCP. The assay results showed that all new psoralens were worse inhibitors of β5i subunit of iCP in comparison to the parent oxathiazolone-based 'compound 42 (Table 1, Figure 6). This is most probably due to the mispositioning of the electrophilic carbons of all compounds and the catalytic Thr1O γ in the β5i active site. Interestingly, all compounds inhibited β5i activity with a similar potency at 1 µM with RA values ranging from 62 to 78%. Of the 12 prepared compounds, 3-bromo-4,5-dihydroisoxazole-substituted psoralen 10 and compound 16 with an α,β-unsaturated aldehyde as the warhead were the most promising. The former showed RA value of 62 ± 5%, whereas for the latter RA was determined at 65 ± 3% (see also postulated binding modes for 10 and 16 in Figure 7). It was not surprising to see that all 12 compounds also exhibited worse inhibition of the β5 subunit of cCP, albeit these differences were much less pronounced as for the β5i subunit. Of note, compounds 9, 11, 23, and 24 were slightly better inhibitors of β1i subunit of iCP in comparison to the 'compound 42 . All psoralen-based compounds (with oxathiazolone included) did not inhibit other subunits of both CPs (i.e., β2i, β2, and β1), whereas carfilzomib completely abolished activity of all subunits at 1 µM (Table 1, Figure 6). the inhibition of all catalytically active subunits of iCP and cCP. The assay results showed that all new psoralens were worse inhibitors of β5i subunit of iCP in comparison to the parent oxathiazolone-based 'compound 42′ (Table 1, Figure 6). This is most probably due to the mispositioning of the electrophilic carbons of all compounds and the catalytic Thr1O γ in the β5i active site. Interestingly, all compounds inhibited β5i activity with a similar potency at 1 µM with RA values ranging from 62 to 78%. Of the 12 prepared compounds, 3-bromo-4,5-dihydroisoxazole-substituted psoralen 10 and compound 16 with an α,β-unsaturated aldehyde as the warhead were the most promising. The former showed RA value of 62 ± 5%, whereas for the latter RA was determined at 65 ± 3% (see also postulated binding modes for 10 and 16 in Figure 7). It was not surprising to see that all 12 compounds also exhibited worse inhibition of the β5 subunit of cCP, albeit these differences were much less pronounced as for the β5i subunit. Of note, compounds 9, 11, 23, and 24 were slightly better inhibitors of β1i subunit of iCP in comparison to the 'compound 42′. All psoralen-based compounds (with oxathiazolone included) did not inhibit other subunits of both CPs (i.e., β2i, β2, and β1), whereas carfilzomib completely abolished activity of all subunits at 1 µM (Table 1, Figure 6).

General Chemistry Methods
Reagents and solvents were obtained from commercial sources (Acros Organics

General Chemistry Methods
Reagents and solvents were obtained from commercial sources (Acros Organics (Thermo Fisher Scientific, Waltham, MA, USA), Sigma-Aldrich (Merck KGaA, Darmstadt, Germany), TCI Europe (Tokyo Chemical Industry, Tokyo, Japan), Alfa Aesar (Thermo Fisher Scientific, Waltham, MA, USA), Fluorochem (Fluorochem Ltd., Derbyshire, UK) and were used as received. Carfilzomib were purchased from MedChemExpress. For reactions involving air or moisture sensitive reagents, solvents were distilled before use and these reactions were carried out under nitrogen or argon atmosphere. Reactions using microwaves were performed on a standard monomode microwave reactor MONOWAVE 200 (Anton Paar, Graz, Austria). Reactions were monitored using analytical thin-layer chromatography plates (Merck 60 F254, 0.20 mm), and the components were visualized under UV light and/or through staining with the relevant reagent. Normal phase flash column chromatography was performed on Merck Silica Gel 60 (particle size 0.040-0.063 mm; Merck, Germany). 1 H and 13 C-NMR spectra were recorded at 295 K on a Bruker Avance III 400 MHz spectrometer (Bruker, Billerica, MA, USA) operating at frequencies for 1 H-NMR at 400 MHz and for 13 C-NMR at 101 MHz. The chemical shifts (δ) are reported in parts per million (ppm) and are referenced to the deuterated solvent used. The coupling constants (J) are given in Hz, and the splitting patterns are designated as follows: s, singlet; br s, broad singlet; d, doublet; app d, apparent doublet; dd, doublet of doublets; ddd, doublet of doublets of doublets; t, triplet; dt, doublet of triplets; td, triplet of doublets; m, multiplet. All 13 C-NMR spectra were proton decoupled. Mass spectra data and high-resolution mass measurements were performed on a Thermo Scientific Q Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA). The purity of the compounds used in biochemical assays was determined with analytical normal-phase HPLC on an Agilent 1100 LC modular system (Agilent, Santa Clara, CA, USA) that was equipped with a photodiode array detector set to 254 nm. A Kromasil 3-CelluCoat column (150 mm × 4.6 mm; 3 µm particle size) was used, with a flow rate of 1.0 mL/min and a sample injection volume of 5-20 µL. An isocratic eluent system of A (hexane) and B (isopropanol) was used; the ratio used is described for each compound below. The purities of the test compounds used for the biological evaluations were ≥95%, unless stated otherwise.

Molecular Modelling
Compounds were prepared for docking using LigPrep (Schrödinger Suite 2020-2, Schrödinger, LLC, New York, NY, USA, 2020) to account for all possible tautomers and ionization states at pH 7.0 ± 2.0. The X-ray structure (PDB: 5M2B, [43]) of yeast 20S proteasome with human β5i and β1 subunits in complex with noncovalent inhibitor Ro19 was used for docking. The binding site is defined by the chain K (β5i) and neighbouring L (β1), so all other chains were removed. Protein Preparation Wizard [50] was used to add hydrogen atoms, protonate residues at pH 7, refine the H-bond network and to perform a restrained minimization. The receptor's grid box required for docking calculations was centred on the corresponding co-crystallized ligand. Noncovalent docking was performed using Glide [51], with the following parameters: XP (extra precision), flexible ligand sampling, perform postdocking minimization. Covalent docking was performed with CovDock program [52] using the pose prediction mode with default setup and Thr1 defined as the reactive residue. Nucleophilic addition to a double bond (oxathiazolone) was selected as the reaction.

Conclusions
Here, we showed that the introduction of 12 new electrophilic warheads at position 3 of the psoralen ring led to compounds with abrogated inhibition of the iCP (especially β5i subunit). As already described in the Introduction, it is imperative that the initial noncovalent binding of a given compound is followed by the positioning of the electrophilic 'warhead' near the desired nucleophilic amino-acid residue of the protein to achieve covalent interaction. Poor inhibition results were in our cases most probably due to the mispositioning of the electrophilic carbon and the catalytic Thr1O γ (Figure 7). The oxathiazolone thus remains the optimal electrophilic moiety for this compound class. Despite somewhat disappointing results, the obtained data will help steer our future research in the field of psoralen-based iCP inhibitors, e.g., when designing inhibitors which simultaneously inhibit two iCP subunits as it was established that simultaneous inhibition of β1i and β5i is necessary to achieve significant anti-inflammatory effects.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.