Antagonism of a Zinc Metalloprotease Using a Unique Metal-Chelating Scaffold: Tropolones as Inhibitors of P. aeruginosa Elastase

Abstract

Tropolone emerged from the screening of a chelator fragment library (CFL) as an inhibitor of the Zn²⁺-dependent virulence factor, Pseudomonas aeruginosa elastase (LasB). Based on this initial hit, a series of substituted tropolone-based LasB inhibitors was prepared, and a compound displaying potent activity in vitro and in a bacterial swarming assay was identified. Importantly, this inhibitor was found to be specific for LasB over other metalloenzymes, validating the usage of tropolone as a viable scaffold for identifying first-in-class LasB inhibitors.

Tropone, and its hydroxylated derivative tropolone, are planar, 7-membered, unsaturated rings that are found in natural products with diverse biological activities, including compounds displaying antibacterial, antiinflammatory, antitumor, and antiviral activity.^[1–3] The metal-binding capacity of tropolone is well known, using its exocyclic oxygen donor atoms to bind metal ions (i.e. O,O donor ligand, Figure 1),^[4] and the employment of tropolone-based units in the design of metalloprotein inhibitors has been explored.^{[5, 6]} Tropolone has been identified as an inhibitor of several Zn²⁺-dependent metalloenzymes including carboxypeptidase A, thermolysin, matrix metalloproteases (MMP-2 and -3), and anthrax lethal factor (LF) with IC₅₀ values ranging from 0.003–1.4 mM.^{[1, 4, 7]} Tropolone has also been found to be a potent inhibitor of the dinuclear copper-dependent enzyme tyrosinase (IC₅₀ value of ~400 nM);^[8] however, a recent crystal structure of tropolone bound to tyrosinase revealed that the natural product does not act by coordinating to the metal ion.^[9]

Figure 1.

Metal-binding groups (MBGs) and derived inhibitors with IC₅₀ values listed for LasB inhibition.

In an effort to identify suitable metal-binding groups (MBGs) for targeting metalloprotein active sites, a fragment-based drug discovery (FBDD) approach has been applied via the development of chelator fragment libraries (CFLs). CFLs are specifically designed with fragments that can coordinate metal ions in the active site of metalloproteins. This approach has revealed novel scaffolds such as hydroxypyrones, hydroxypyridiones, hydroxyquinolines, and quinolone sulfonamides to be effective MBGs against a variety of metalloproteins, including MMPs, LF, and several others.^{[4, 7, 10]}

LasB^{[11, 12]} is one of several virulence factors produced by P. aeruginosa to promote infection within a host.^{[13, 14]} Previous mutation-^[15] and vaccine-based^[16] studies have revealed that LasB plays a critical role in promoting virulence through targeted proteolysis of host tissue proteins and immune system components.^[11] Moreover, LasB has also been linked to the establishment of antibiotic-resistant biofilm^[17] and swarm colonies.^{[18, 19]} Because evidence exists supporting the investigation of virulence factors as promising new antibiotic targets,^[20–22] the pursuit of non-peptidic, small molecule inhibitors of LasB is of interest. Recently, the screening of CFL-1.1 against P. aeruginosa elastase (LasB) was shown to produce several hits.^[19] Among the initial hits was 3-hydroxy-1,2-dimethylpyridine-4(1H)-thione (3,4-HOPTO, 1), a chelator that is also a potent inhibitor of MMPs, LF, and others metalloproteins.^[10] Because 3,4-HOPTO displayed activity against LasB (IC₅₀ value 81 μM, Figure 1), an existing sublibrary of 3,4-HOPTO derivatives^[10] was examined revealing compound 2 (Figure 1) as a potent LasB antagonist and the first mechanism-based compound to exhibit swarming antagonism.^[19] Despite its exceptional activity, the 3,4-HOPTO MBG (1) is known to possess activity against a broad set of metalloproteins and it is likely that 2 and related compounds will not be highly selective for LasB (indeed compound 2 is known to inhibit MMPs and LF at concentrations of 1–4 μM).^[10]

Screening of CFL-1.1 against LasB had produced other, albeit less active, MBG hits. It was hypothesized that these hits, which still showed good ligand efficiencies,^{[23, 24]} could be used to develop potent, and more selective LasB inhibitors. Herein, we report the discovery of tropolone-based inhibitors of LasB. Importantly, the identified antagonists are not only more potent than the previously reported 3,4-HOPTO-derived compound 2, but are also highly selective for LasB over other metalloproteins.

From our prior screening of CFL-1.1 against LasB,^{[7, 10]} tropolone (3, Figure 2) emerged as a “hit”, with an IC₅₀ value of 387±1 μM.^[19] As mentioned earlier, tropolone is known to inhibit Zn²⁺-dependent thermolysin via metal coordination.^{[25, 26]} Importantly, acetylation of 3 to produce acetyltropolone (4, Figure 2)^[1] resulted in a ~3-fold reduction in potency (IC₅₀ value of 1.29±0.24 mM) suggesting that Zn²⁺ chelation is required for LasB inhibitory activity. Again, this parallels findings with thermolysin, where acetylation of tropolone resulted in a loss of activity.^[1] Also contained in CFL-1.1 was β-thujaplicin (5, Figure 2), a tropolone natural product containing an isopropyl group in the 4-position of the ring^{[2, 3]} The IC₅₀ value of 5 was 1.34±0.37 mM, which is ~3-fold less potent than 3, perhaps suggesting that steric repulsion at the 4-position is prohibiting proper Zn²⁺ chelation within the active site. However, 5-aminotropolone (6, Figure 2) was found to be the most potent hit, with an IC₅₀ value of 180±1 μM, suggesting that selected substitutions may be tolerated on the ring system.

Figure 2.

Tropolone fragments with IC₅₀ values listed for LasB inhibition. Standard deviations are shown in parentheses.

Encouraged by these findings, a sublibrary of compounds based on fragment 6 was constructed via a one-step condensation of 6 with isocyanates under microwave irradiation conditions (Figure 3, Figure S1^†). This sublibrary was screened using a fluorescence-based assay (Figure S2^†),^{[19, 25]} and several compounds were found to have IC₅₀ values of <10 μM (Table S1^†). The best compounds identified from this sublibrary were more potent than the previously reported 3,4-HOPTO-based inhibitors by ~2-fold.^[19] In particular, compound 7a was found to be the most potent non-peptidic small molecule inhibitor of LasB identified to date with an IC₅₀ value of 1.16±0.06 μM. Moreover, compound 7a was found to be a competitive inhibitor of the enzyme with a K_i of 336 nM (Figure S4^†). To determine the structural relevance of the urea linker used in these compounds, an analogue of 7a, containing an amide linker was synthesized (8, Figure 3) resulting in a >8-fold loss in activity. Coupling to a shorter amide linker (9, Figure 3) produced a nearly 50-fold loss in potency. These compounds demonstrate that both the urea linkage and the linker length contribute to the activity of 7a against LasB. Finally, acetylation of 7a was performed to produce compound 10 (Figure 3) with a blocked MBG (analogous to compound 4 above). Compound 10 also showed a ~50-fold reduction in activity (IC₅₀ value of 58.4±0.1 μM, Figure S3^†), providing strong evidence that metal coordination is essential for LasB inhibition by tropolone compounds. This was further verified by acetylation of another potent derivative 7e (Figure S1, Table S1^†), which resulted in an even greater loss of potency (>100-fold, data not shown).

Figure 3.

Structures and IC₅₀ values for tropolone sublibrary compounds.

Because the 3,4-HOPTO chelating motif found in 2 (Figure 1) was known to inhibit other Zn²⁺ metalloproteins, including MMPs,^{[7, 10]} the specificity of 2 versus tropolone 7a was examined against a panel of metalloenzymes. Three Zn²⁺-dependent metalloproteins, MMP-2, MMP-9, and human carbonic anhydrase II (hCAII), and one Cu²⁺-dependent metalloenzyme, tyrosinase, were examined as a gauge of selectivity. Specificity was determined by treating each enzyme with 50 μM of either 2 or 7a and monitoring percent enzymatic activity. Both 2 and 7a had a modest effect on hCAII, each reducing activity by ~25% at 50 μM (Figure S5^†). In contrast, 2 was found to completely suppress the enzymatic activity of MMP-2 and MMP-9 at 50 μM, while 7a showed only weak inhibition (<40% at 50 μM, Figure 4). Finally, neither 2 nor 7a showed any inhibition against tyrosinase. The loss in activity of 7a against tyrosinase is a key observation, as unsubstituted tropolone (3) is a very potent inhibitor of this enzyme (IC₅₀ value ~400 nM, Figure S5^†). Similar results were obtained with several other tropolone-derived inhibitors (data not shown). Therefore, by derivatizing tropolone into a more advanced lead compound (7a), a highly selective LasB inhibitor has been obtained that, unlike HOPTO-derived inhibitors, is selective for LasB over other Zn²⁺-dependent metalloproteins, and is also inactive against the Cu-dependent tyrosinase, for which the tropolone MBG has an inherently high affinity. This illustrates that by judicious selection of both MBG and backbone target-specific metalloprotein inhibitors can be designed.

Figure 4.

Inhibition of Zn²⁺-dependent MMP-2 (solid) and MMP-9 (lines) by a broad-spectrum MMP inhibitor (+ control^†), 2, and 7a.

Based on the enhanced potency and specificity of compound 7a, its activity was further characterized in a cellular assay. As stated above, LasB activity is required for the induction of swarming in P. aeruginosa.^{[18, 19]} Importantly, the ability of P. aeruginosa to form swarm colonies has been linked to the development of antibiotic resistance,^{[27, 28]} indicating that small molecule inhibitors of LasB could be used as adjuvants with traditional antibiotics to enhance the susceptibility of antibiotic-resistant P. aeruginosa to these drugs.^[29] To examine the anti-swarming activity of compound 7a, P. aeruginosa strain PA14 was grown on swarm agar plates containing either DMSO (control) or 25 μM of 7a. As shown in Figure 5, this tropolone-based inhibitor was able to completely inhibit the swarming phenotype at this concentration, exhibiting swarming inhibitory properties comparable to 2.^[19] Importantly, 7a was found to be non-cytotoxic to PA14 at a concentration of 25 μM (Figure S6^†). Finally, compound 10, which has an acetylated tropolone MBG, was found to be much less effective at inhibiting swarming (Figure S7^†). Thus, these results demonstrate the potential of this natural product-based chelating moiety for the design of antimicrobial metalloprotease inhibitors.

Figure 5.

Swarming of P. aeruginosa strain PA14 in the absence (left, DMSO control) or presence of 7a (right, 25 μM).

Conclusions

In conclusion, the first tropolone-based metalloprotein inhibitors have been developed by a chelator-focused FBDD approach. These compounds are the most potent non-peptidic small-molecule inhibitors of LasB reported to date and show excellent activity in a cell-based swarming assay. Importantly, the tropolone MBG-derived inhibitors are more active and more selective than the previously identified HOPTO-based compounds. The work presented here is consistent with earlier studies on tropolone-based metalloprotein inhibitors. While the majority of the previous tropolone-based inhibitors were identified by screening of natural products, this study demonstrates how use of chelator fragment libraries and sublibraries can rapidly identify leads for the development of such inhibitors. The present findings clearly suggest that identification of privileged chelating scaffolds for a given metalloenzyme can lead to the realization of both potent and selective metalloprotein inhibitors.