Design and synthesis of tailored human caseinolytic protease P inhibitors†
Human caseinolytic protease P (hClpP) is important for degradation of misfolded proteins in the mitochondrial unfolded protein response. We here introduce tailored hClpP inhibitors that utilize a steric discrimination in their core naphthofuran scaffold to selec- tively address the human enzyme. This novel inhibitor generation exhibited superior activity compared to previously introduced beta- lactones, optimized for bacterial ClpP. Further insights into the bioactivity and binding to cellular targets were obtained via chemical proteomics as well as proliferation- and migration studies in cancer cells. Mitochondria are essential for cellular energy production and represent a main source of reactive oxygen species (ROS).1 Stress caused by ROS results in accumulation of damaged proteins which block essential functions and induce the mito- chondrial unfolded protein response (mtUPR).2 Human caseino- lytic protease P (hClpP) represents a major mediator of cellular homeostasis in mtUPR by degradation of damaged proteins into peptides.3,4 These peptides are transported across the inner mitochondrial membrane to activate stress-associated transcrip- tion factors.5,6 hClpP is a serine protease expressed as inactive heptamers which assemble into an active tetradecamer by binding to hClpX.7–9 Mice bearing a ClpP knockout suffer from hearing loss, infertility and growth defects. These effects are paralleled in humans by recessive mutations of hClpP causing the Perrault syndrome.
While hClpP is not essential for viability of normal cells, a recent study demonstrated its upregulation in patients with acute myeloid leukemia (AML).11 Rapidly dividing cancer cells are exposed to multiple mitochondrial stresses and require an efficient UPR to maintain cell homeostasis. Accord- ingly, silencing of hClpP in AML cell lines increased mitochon- drial ROS levels and thereby attenuated cell viability. hClpP thus represents a promising anti-cancer target.12 Previous experi- ments with beta-lactone inhibitors, originally discovered and optimized for bacterial ClpPs,13 induced death of leukemia cells and even reduced their growth in xenografted mice.11 We recently discovered a next generation of bacterial ClpP inhibitors based on a phenyl ester scaffold.14 Of note, one molecule, AV167, displayed superior inhibitory activity against hClpP exceeding that of previous beta-lactones (Fig. 1a).We here systematically dissect the AV167 core structure andidentify novel inhibitors with superior potency for hClpP inhi- bition compared to the gold standard lactones. The inhibitors showed anti-proliferation, anti-migratory and apoptotic effects against cancer cells. Corresponding activity-based protein pro- filing (ABPP) suggested that multiple targets are likely respon- sible for these effects. Given the clinical need for novel hClpP inhibitors our scaffolds provide a unique starting point to further fine tune selectivity.
In a previous high-throughput screen against bacterial ClpP, AV167, bearing a characteristic naphthofuran moiety, protruded as the best hClpP inhibitor out of six phenyl ester hits.14 We thus selected this scaffold for further optimization. First, covalent inhibition of hClpP was validated for AV167 and compared to one of the best previous beta-lactone ClpP inhibitors (D3) (Fig. 1a). As quantitative acylation kinetics, represented inbinding velocity (Fig. S1, ESI†), we performed two alternative assays assessing the reactivity and potency. ClpP inhibitors have been previously characterized by their inhibition of ClpP peptidase activity at three different concentrations (100, 10 and 1 mM).14–16 AV167 significantly reduced hClpP turnover already at 10 mM and intact mass-spectrometric (MS) analysis of the modified enzyme revealed about 80% binding (Fig. 1d, e and Fig. S2, ESI†). Contrary, D3 only achieved 65% inhibition at thehighest concentration tested (Fig. 1d). This trend is also reflected by corresponding apparent IC50 values determined at identical incubation times. D3 displayed a 4.5-fold higher value compared to AV167 and again retained 35% residual enzyme activity at saturating concentrations indicating restricted binding (Fig. 1f and Fig. S4, ESI†).
Based on these promising properties, AV167 was selected for structure–activity relationship (SAR) studies to(i) dissect the structure for essential enzyme recognition motifs and (ii) identify possible sites for functionalization with a benign tag for target identification via ABPP.17–19 Synthesis of five derivatives bearing an 8-methoxynaphthofuran moiety (TG11) or different substituted benzofurans (TG24–TG27) resulted in inactive compounds (Fig. 1b and Scheme S1, ESI†). Thus, modifications at a different site were subsequently explored by introducing a bromine substituent (TG43) and an alkyne tag at 2-position (TG42) (Fig. 1b, c and Scheme 1). Synthesis of these derivatives started with Pechmann reaction of 2-naphthol and ethyl 4-chloroacetoacetate to the corresponding b-naphtho- coumarin (TG04) which rearranged under basic conditions tonaphthofuranylacetic acid (TG06). Ester protection and sub- sequent electrophilic bromination with NBS led to the precursor (TG38) for Sonogashira coupling with TMS-protected acetylene. Deprotection of the carboxylic ester and TMS-protecting group could be achieved in one step with aqueous LiOH solution. Esterification with oxadiazolyl phenol (TG05), derived from methyl paraben via two step synthesis, resulted in com- pound TG42 (Scheme 1). Satisfyingly, probe TG42 and its bromo precursor TG43 exhibited elevated potency compared to AV167 (1.54 mM) with apparent IC50s of 0.39 and 0.57 mM, respectively (Fig. 1b, f and Fig. S4, ESI†). Intact MS revealed about 50% modification to be sufficient to achieve full inhibition (Fig. S2, ESI†).
As AV167 bears an activated ester moiety we systematically tested the inhibition profiles of various ester derivatives (Fig. 1b and d). As expected alcohols exhibiting electron-donating properties such as phenol and p-methoxyphenol (TG28, TG29) turned out to be poor inhibitors while derivatives with electron withdrawing groups including p-nitro- and p-methylsulfonylphenol (TG30, TG51, TG52, TG54) showed inhibition profiles comparable to the parent p-oxadiazolylphenol of AV167. All compounds with activated ester moieties and naphthofuran substituents in 2-position (TG42, TG43, TG53) displayed potent IC50s ranging from 0.33–0.57 mM, indicating sub-stoichiometric inhibition of hClpP (used at 1 mM in this assay) (Fig. 1f and Schemes S1–S5, ESI†). Contrary to AV167, these compounds showed only minor inhibition of S. aureus ClpP, demonstrating a crucial discrimination of species selec- tivity caused by substituents attached to the naphthofuran ring at 2-position (Fig. 1f and Fig. S3, ESI†).To assess the ability of these compounds to block proteolytic activity hClpP was activated via an acyldepsipeptide derived chaperone mimic.20,21 Binding of this activator fragment to hydrophobic pockets induces pore opening and digest of FITC casein. While beta-lactone D3 showed no activity, AV167, TG42 and TG52–TG54 efficiently blocked proteolysis (Fig. 2a). This result was further corroborated by a second proteolysis assay in presence of the ClpX chaperone.22,23 Escherichia coli ClpX (EcClpX) forms a functionally active complex with hClpP and is preferred in these assays compared to the largely uncharacterized human ClpX.7 For enzyme activity assessment, degradation ofquantification (Fig. 2d).
TG42 significantly enriched hClpP in Jurkat cells, however, a large number of other prominent targets were additionally obtained, suggesting that the electrophilic probe is of either limited selectivity in human proteomes or has restricted access to mitochondrial compartments (Fig. 2e and Fig. S6, ESI†). In contrast, labeling of Huh7 cells (Fig. S7, ESI†) and Jurkat cells27 with lactone D3 resulted in no enrichment of hClpP at all. Thus, despite its superior in vitro potency and labeling performance compared to previous inhibitors, further optimization of TG42 is required to enhance cellular selectivity for mitochondrial hClpP. Nevertheless, given the crucial role of hClpP and several of the identified targets such as Reticulon28 and Protein LYRIC29 in apoptosis (Fig. 2e and Fig. S6, ESI†) we investigated the biological consequences of TG42 in Huh7 cancer cells and demonstrated concentration-dependent cell death with induction of apoptosis (Fig. 3a and b). Cancer cell migration could also be significantly reduced at 30 mM TG42 treatment (Fig. 3c).Selective eradication of cancer cells is an important task in drug development to reduce side effects of chemotherapy. hClpP is a promising target in this endeavor lacking inhibitors for therapeutic applications. AV167 represented a starting point to achieve this goal bearing a characteristic curved naphthofuranmoiety suitable to durably inhibit ClpP peptidolytic as well as proteolytic activities. Importantly, substitution at 2-position of the naphthofuran moiety not only enhanced in vitro potency but also represented a crucial signature motif differentiating between binding into human or bacterial ClpP isoforms. TG42 exhibited ZK53 superior inhibition and selectivity towards hClpP compared to a reference beta-lactone. Further fine-tuning of the inhibitor scaf- fold is needed to enhance its cellular selectivity.This work was supported by the Deutsche Forschungsge- meinschaft SI1096/8-1 (ClpP) and CIPSM.