Polypharmacology Approaches against the Pseudomonas aeruginosa MvfR Regulon and Their Application in Blocking Virulence and Antibiotic Tolerance

Abstract

Pseudomonas aeruginosa is an important nosocomial pathogen that is frequently recalcitrant to available antibiotics, underlining the urgent need for alternative therapeutic options against this pathogen. Targeting virulence functions is a promising alternative strategy as it is expected to generate less-selective resistance to treatment compared to antibiotics. Capitalizing on our nonligand-based benzamide–benzimidazole (BB) core structure compounds reported to efficiently block the activity of the P. aeruginosa multiple virulence factor regulator MvfR, here we report the first class of inhibitors shown to interfere with PqsBC enzyme activity, responsible for the synthesis of the MvfR activating ligands HHQ and PQS, and the first to target simultaneously MvfR and PqsBC activity. The use of these compounds reveals that inhibiting PqsBC is sufficient to block P. aeruginosa’s acute virulence functions, as the synthesis of MvfR ligands is inhibited. Our results show that MvfR remains the best target of this QS pathway, as we show that antagonists of this target block both acute and persistence-related functions. The structural properties of the compounds reported in this study provide several insights that are instrumental for the design of improved MvfR regulon inhibitors against both acute and persistent P. aeruginosa infections. Moreover, the data presented offer the possibility of a polypharmacology approach of simultaneous silencing two targets in the same pathway. Such a combined antivirulence strategy holds promise in increasing therapeutic efficacy and providing alternatives in the event of a single target’s resistance development.

Graphical Abstract

The bacterium Pseudomonas aeruginosa is an opportunistic Gram-negative bacterial pathogen responsible for more than 50 000 infections in the U.S. each year, primarily causing nosocomial infections in immunocompromised and cystic fibrosis patients.¹ P. aeruginosa rapid expansion of resistance to almost all available antibiotics urges the development of alternative therapeutic strategies.² One promising strategy, especially in the case of multidrug resistant or pan antibiotic resistant P. aeruginosa clinical strains, is antivirulence drugs that target bacterial virulence systems or master virulence regulators.³ One such master virulence regulator is the quorum sensing (QS) cell-to-cell bacterial communication system.

P. aeruginosa possesses three major QS systems: LasR,⁴ RhlR,^5,6 and MvfR.^7–10 The MvfR QS system is a promising antivirulence target due to its critical role in inducing the expression of multiple P. aeruginosa virulence systems that promote both acute and chronic infections.^7,11–14 Moreover, as opposed to LasR no clinical isolates with frequent mutations in MvfR were reported to date.¹⁵ The transcriptional regulator MvfR (also known as PqsR) binds to 37 loci and regulates the expression of the associated genes¹⁶ including the pqsABCDE operon, whose encoded proteins catalyze the biosynthesis of MvfR inducers and of ~60 distinct low-molecular-weight compounds,^7–9,17,18 part of which are the 4-hydroxyl-2-alkyl-quinolines (HAQs)¹⁹ and the non-HAQ molecule 2-amino-acetophenone (2-AA).^12,14,20 This multistep biosynthetic pathway is summarized in Figure 1a. The first step is the conversion of the HAQs precursor anthranilic acid by PqsA and PqsD into 2-aminobenzoylacetyl-CoA (2-ABA-CoA),^21,22 which then either spontaneously cyclizes into 2,4-dihydroxyquinoline (DHQ)^22–24 or is hydrolyzed by the thioesterase PqsE (or TesB) into 2-aminobenzoylacetate (2-ABA).²³ 2-ABA and octanoyl-CoA are then condensed by the PqsBC enzyme into the MvfR activating ligand 4-hydroxy-2-heptyl-quinoline (HHQ),^24,25 which is later hydroxylated into the second MvfR activating ligand 3,4-dihydroxy-2-heptyl-quinoline (PQS) by PqsH.²⁶ 2-ABA can also decompose into DHQ²⁴ or undergo decarboxylation to 2-AA²⁴ (Figure 1a).

Figure 1.

HAQs and 2-AA biosynthesis pathway and inhibition in live P. aeruginosa cells. (a) HAQs and 2-AA biosynthesis pathway. PqsABCDE are encoded by the pqs operon. PqsH and TesB encoding genes are located elsewhere in P. aeruginosa chromosome. TesB is another thioesterase also able to convert 2-ABA-CoA into 2-ABA. (b,c) HHQ, PQS, 2-AA and DHQ levels measured by LC/MS in an mvf R mutant strain constitutively expressing the pqs operon (b) or in PA14 wild type strain (c) in the presence or absence of various BB inhibitors at 100 μM. Levels are normalized to that of the DMSO vehicle control. Results show the average ± SD of at least two independent replicates.

Importantly, HHQ and PQS are critical for MvfR activity in vitro; however, only HHQ appears to be essential for acute infection in vivo, as the absence of PQS assessed by using a pqsH mutant causes WT mortality in mice.⁹ On the contrary, 2-AA silences the acute infection branch of the MvfR QS system by binding and inhibiting the activity of PqsBC^13,25 and promotes antibiotic tolerance as well as chronic/persistent infections by interfering with the bacterial translation apparatus and modulating epigenetically the host immune system to promote host tolerance to infection, respectively.^12–14,27 The role of the MvfR QS pathway in both acute and chronic infections has motivated several drug discovery studies that generated inhibitors of PqsA,^28,29 PqsD,^30,31 and MvfR.^32–35 However, no synthetic PqsBC inhibitor has been identified to date. We previously reported the identification and development of a new family of molecules with a benzamide–benzimidazole (BB) core structure as highly cell-permeable inhibitors of the MvfR QS system.³² One of our most potent inhibitors, M64, was shown to target MvfR and inhibit HAQs and 2-AA synthesis with an IC₅₀ in the nanomolar range.³² The present work describes our effort to obtain further mechanistic knowledge on the BB family inhibitors’ capability in inhibiting MvfR circuitry. This work provides novel insights that are critical in the design of improved MvfR regulon inhibitors and reports the identification of the unprecedented first class of dual inhibitors of MvfR and PqsBC activities.

RESULTS AND DISCUSSION

HAQs and 2-AA Inhibition Profiling Reveals MvfR Regulon Inhibitors with Dual Targets.

We previously reported an initial target assessment of few of our most potent BB compounds and demonstrated the physical interaction of the most potent inhibitor, M64, with MvfR.³² However, the profile of a BB compound, M51, was ambiguous, which could suggest that this chemical family might target other proteins in the MvfR QS system besides the transcriptional regulator MvfR itself. In order to address this point, we performed a systematic target assessment for representative BB compounds in our collection.

First, we assessed the ability of each compound to interfere with the activity of MvfR, PqsA, PqsBC, or PqsD by quantifying the production levels of the MvfR-regulated molecules HHQ, PQS, DHQ and 2-AA. We used an mvf R isogenic mutant strain that constitutively expresses the pqsABCDE genes (mvf R- pPqsABCD)³² and thus has MvfR-independent HHQ, PQS, 2-AA and DHQ production. In this strain, inhibitors targeting PqsA or PqsD enzymes result in decreased production of all four MvfR-regulated molecules, while PqsBC inhibition causes decreased production of HHQ and PQS level and an accumulation of 2-AA or DHQ; in the mvf R- pPqsABCD strain MvfR inhibition has no impact on the production of any of these molecules. Interestingly, our inhibitors’ collection contained compounds exerting at least two types of inhibitory patterns in the mvf R- pPqsABCD strain (Figure 1b). The first group of molecules (green) exhibits the same inhibition pattern previously described with the MvfR inhibitor M64,³² that is, no inhibition of HHQ, PQS, DHQ, and 2-AA. These results suggest that inhibitors M50, M62, M34, M61, M53, and M57 do not target the PqsA, PqsBC, or PqsD enzymes, which points to MvfR as the potential target. In contrast, however, the second group of compounds (blue), although inhibiting HHQ and PQS, led to a considerable accumulation of 2-AA and DHQ (Figure 1b), clearly suggesting that they are targeting the PqsBC enzyme.

The ability of each compound to interfere with the activity of either MvfR or PqsBC was further interrogated in the parental strain PA14 where the expression of the pqs operon is MvfR dependent (Figure 1c). In this strain, inhibition of MvfR results in a reduced production of HHQ, PQS, 2-AA, and DHQ, whereas PqsBC inhibition only decreases HHQ and PQS production but not 2-AA or DHQ. Figure 1c shows the 2-AA, DHQ, PQS, and HHQ levels produced in the WT strain PA14 in the presence of each compound. We observed two different inhibition profiles: Compounds of the first profile (red)—M50, M62, M34, M61, M53, M57, M59, M58, M51, B1, and M52—exhibit the inhibition pattern we previously observed with the MvfR inhibitor M64,³² that is, inhibition of HHQ, PQS, 2-AA, and DHQ in this wild-type PA14 strain, suggesting they are targeting MvfR. Interestingly, however, five of these compounds—M59, M58, M51, B1, and M52—were identified as PqsBC inhibitors in Figure 1b, indicating that they may act as dual inhibitory compounds that also target MvfR in addition the PqsBC enzymatic activity. Compounds from the second profile (orange), identified as PqsBC inhibitors in Figure 1b—M27, M26, M23 M4, M8, and M55—partially inhibit the synthesis of 2-AA and DHQ in addition to fully blocking the synthesis of HHQ and PQS (Figure 1c). M22 also blocks 2-AA and DHQ production but requires higher concentrations to do so (Figure S5). These data suggest that compounds M27, M26, M23 M4, M8, and M55 and M22 may also interfere with MvfR activity in addition to that of PqsBC. However, their lower efficacy at reducing 2-AA and DHQ production compared to M59, M58, M51, B1, and M52 suggests that, overall, they are less potent MvfR inhibitors (Figure 1c). While compounds M23, M4 and M8 reduce 2-AA production, they induce the production of DHQ in the wild type PA14 strain (Figure 1c). Although the reason for this effect is not clear, changes in the bacterial culture pH and inactivation of PqsE have been shown to affect the ratio between 2-AA and DHQ.^23,24

Taken together, these data suggest that the BB compounds can be classified in three categories: (1) MvfR inhibitors (M64, M50, M62, M34, M61, M53, and M57); (2) MvfR–PqsBC dual inhibitors with high anti-MvfR and high anti-PqsBC activity (M59, M58, M51, B1, and M52); and (3) MvfR–PqsBC dual inhibitors with low anti-MvfR and high anti-PqsBC activity (M27, M26, M23, M4, M8, M55).

We report here a combination of two HAQs quantification assays in live cells that provide valuable insights on compound target(s). The mvfR isogenic mutant strain constitutively expressing the pqs operon (mvfR-pPqsABCD) is particularly useful in discriminating between MvfR and PqsA/D inhibitors, and it can also help identify PqsBC inhibitors that were overshadowed by the MvfR/PqsA/PqsD inhibition phenotype dominant in the PA14 wild type strain. Accordingly, it is possible that some MvfR, PqsA or PqsD inhibitors reported in previous studies such as^28,30,33 might also have other targets within the MvfR QS pathway. We believe that such assays would benefit the community in discriminating the target(s) of MvfR QS system inhibitors and allow a better understanding of the determinants driving the interaction of compounds with various targets in this pathway.

Target Validation.

To validate the inhibitors’ targets, we selected representative compounds from each category and assessed their ability to bind to MvfR using surface plasmon resonance (SPR). As expected, all tested inhibitors bind to MvfR (Figure 2a,c and S2). Compounds with a low anti-MvfR activity (M27, M26, and M23) bind 9 to 111 times less efficiently to MvfR than those with high anti-MvfR activity (M64, M50, M62, M59, M51) (Figure 2a). This lower binding is nonetheless significant because the K_D of those inhibitors is in the same order of magnitude as that of HHQ and PQS, two well-established MvfR native ligands.¹⁰

Figure 2.

Target validation. Compounds binding intensity to MvfR was measured via SPR (a,b). Binding to MvfR was assessed with a wide range of inhibitors concentrations. Compounds interference with PqsBC was measured via enzyme kinetics by assessing the inhibitory activity on the condensation of 2-ABA and octanoyl-CoA to HHQ by PqsBC (c,d). 2-ABA conversion into HHQ was assessed with a wide range of inhibitors concentrations. Shown is the average ± SEM of three independent replicates. (e) Binding of M59 assayed with fluorescence polarization spectrometry. The autofluorescence of M59 was used to probe binding of the molecule to PqsBC (gray fit) and octanoyl-PqsBC (black fit), each of which were titrated stepwise to the inhibitor solution. The calculated dissociation constants were 2.5 and 2.9 μM, respectively, indicating that M59 binds to both forms of the enzyme. (f) The displacement of 2-AA, a PqsBC inhibitor competitive with 2-ABA, by M59 was analyzed by measuring the 2-AA fluorescence intensity change in response to PqsBC binding. In an experiment where 2-AA was titrated into a solution containing 1 μM PqsBC, the K_D was 6.8 μM (upper fit). When 10 μM M59 was present in the protein solution, the apparent K_D of the PqsBC-2-AA complex increased (lower fit), indicating that M59 interferes with 2-AA binding.

We then assessed the ability of one compound in each category to interfere directly with PqsBC enzymatic activity by quantifying the in vitro conversion of 2-ABA into HHQ using purified PqsBC protein. As expected, the MvfR–PqsBC dual inhibitors M59 and M27 block the ability of PqsBC to convert 2-ABA into HHQ, with an EC₅₀ of 13.4 and 12.5 μM, respectively (Figure 2c,d). Moreover, Figure 2e,f show that M59 binds to PqsBC and displaces 2-AA, a natural PqsBC inhibitor structurally unrelated to BB compounds acting competitively with the physiological substrate 2-ABA, confirming further M59 ability to interfere with PqsBC. These data indicate that the compounds reported here are significantly more potent PqsBC inhibitors than 2-AA whose EC₅₀ for PqsBC inhibition was previously found to be 46 μM.²⁵ Moreover, the anti-PqsBC activity of 2-AA in live cells is weak as its IC₅₀ for HHQ inhibition is around 400 μM.¹³ Interestingly, Figure 2c shows that the MvfR inhibitor M64 also interferes with PqsBC activity although the inhibition is weaker (EC₅₀ ~ 185 μM) compared to that of M59 and M27. This suggests that other inhibitors from the first category (M50, M62, M34, M61, M53, and M57) may also be MvfR–PqsBC dual inhibitors with a weak anti-PqsBC activity. To exclude any direct reactions of inhibitors with the PqsBC substrates, we analyzed inhibitor–substrate mixtures by UV spectroscopy and assessed 2-ABA and octanoyl-CoA stability by HPLC. Data verified the absence of reactivity and confirmed the stability of both, 2-ABA and octanoyl-CoA in the presence of any of the inhibitors (Figure S1 and Table S1). Overall, these physical interaction and enzymatic activity data confirm the existence of MvfR–PqsBC dual inhibitors.

One puzzling question is how compounds with the same core structure can bind to such different targets, one being a transcriptional regulator and the other a biosynthetic enzyme. It is worth noting that the common point between both proteins is HHQ. Indeed, PqsBC catalyzes HHQ production while MvfR binds to HHQ. Therefore, it is possible that compounds with the BB core structure share some physical properties with HHQ allowing them to bind HHQ related proteins/targets.

Polypharmacology, involving a single or multiple drugs acting on the same pathway has been shown to have a significant impact on the treatment efficacy for various diseases,^36,37 including bacterial infections.^38,39 Although there seems to be no synergistic benefit with this series of inhibitory compounds, the possibility of blocking two different targets in the same pathway offers an advantage in the event of a single target’s resistance development.^39,40 For example, if a resistance mutation occurs in PqsBC, a dual inhibitor would retain both antivirulence and antipersistence potency via its anti-MvfR activity. In the case of a resistance mutation in MvfR, dual inhibitors would still retain an antivirulence potency thanks to their anti-PqsBC activity, although they might not be able to retain the antipersistence activity. The current view in the antivirulence field is that resistance is still expected to occur but at much lower frequencies than that promoted by traditional antibiotics. By definition, virulence contributes to pathogen fitness in vivo, implying that an antivirulence resistant mutant would theoretically outcompete sensitive cells during treatment. However, some important microbial population genetics and ecological concepts such as public/private use of virulence factors, fitness localization, and population structure suggest that selective pressure to antivirulence may only be applied in specific settings, as opposed to that of antibiotics, which apply selective pressure in all settings.^3,41,42 It is worth pointing out that thus far no MvfR mutations have been reported in P. aeruginosa clinical isolates sequenced, suggesting that a functional MvfR QS system is critical for P. aeruginosa infections. However, one cannot exclude this from occurring once MvfR QS system inhibitors are used in the context of long-term treatment regiments (i.e., CF patients). Therefore, a polypharmacology antivirulence approach could present a significant advantage for the treatment of P. aeruginosa infections.

Inhibition of P. aeruginosa Virulence and Antibiotic Tolerance.

To assess the potential of our compounds to reduce P. aeruginosa virulence, we evaluated the ability of selected inhibitors from each category to block P. aeruginosa acute virulence against A549 human lung epithelial cells and RAW264.7 macrophages in vitro using cell viability as a readout. Cell survival was quantified 3 h postinfection with PA14 in the presence or absence of each inhibitor. Infection with P. aeruginosa resulted in 79.2% lung epithelial cell death (Figure 3a) and 74.4% macrophage cell death (Figure 3b). In contrast, treatment with inhibitors from all three categories increased 3.1 to 3.8 times lung epithelial cell survival and 1.9 to 2.6 times macrophage survival to PA14 infection (Figure 3a,b). These compounds show no cytotoxic effect (Figure S4) and importantly reduce P. aeruginosa virulence in an MvfR QS system dependent manner, as they do not significantly change the lung cell survival rate when added to the mvf R mutant cells (Figure S3), indicating no off-target effects (Figure S4). Moreover, no significant difference in the survival of lung epithelial cells or murine macrophages was observed between compounds that exhibit low (M27, M26, M23) and high anti-MvfR activity (M64, M50, M59, M58, M51) (Figure 1c).

Figure 3.

Antivirulence efficacy in lung epithelial cells and macrophage infection assays Survival of A549 human lung epithelial cells (a) or RAW264.7 macrophage cells (b) to PA14 infection in the presence of 50 μM of dual inhibitors with high anti-MvfR activity and low anti-PqsBC activity (green), dual inhibitors with high anti-MvfR activity and high anti-PqsBC activity (red), dual inhibitors with low anti-MvfR activity and high anti-PqsBC activity (orange), or the DMSO vehicle control (black). Results show the average ± SEM of at least three independent replicates. Statistical significance to the DMSO control was assessed using one way ANOVA + Dunnett’s post-test. No statistical difference was observed when comparing the inhibitors with each other (p > 0.05, One Way ANOVA + Tukey post-test).

Next, we evaluated the potential of our compounds to inhibit antibiotic tolerance by assessing bacterial survival to the β-lactam antibiotic Meropenem. Data presented in Figure 4 indicate that the dual inhibitors with low anti-MvfR activity (M27, M26, and M23) do not significantly reduce tolerance to Meropenem. However, the dual inhibitors that exhibit high anti-MvfR activity and consequently block 2AA production (M64, M50, M59, M58, and M51) reduce antibiotic tolerance by more than 70% compared to vehicle control (Figure 4).

Figure 4.

Inhibition of antibiotic tolerance. Tolerance to 10 μg/mL of the β-lactam antibiotic Meropenem in the presence of 10 μM of dual inhibitors with high anti-MvfR activity and low anti-PqsBC activity (green), dual inhibitors with high anti-MvfR activity and high anti-PqsBC activity (red), dual inhibitors with low anti-MvfR activity and high anti-PqsBC activity (orange), or the DMSO vehicle control (black). Results show the average ± SEM of at least three independent replicates. Statistical significance to the DMSO control was assessed using one-way ANOVA + Dunnett’s post-test.

Overall, these data indicate that all three categories of inhibitors have a similar therapeutic potential in the context of acute P. aeruginosa infections likely because they all efficiently block HHQ and PQS production. However, dual inhibitors with a high anti-MvfR activity are more potent at blocking antibiotic tolerance than those with a low anti-MvfR activity because they restrict 2-AA production more efficiently. Therefore, MvfR–PqsBC dual inhibitors with a high anti-MvfR activity have an increased therapeutic potential against pro-acute infection related molecules HHQ and PQS, as well as the pro-persistent and immunomodulatory molecule 2-AA.

Role of Inhibitors Structure in Target Recognition.

Finally, we assessed whether some compound structural determinants could be associated with selective target recognition. Figure 5 shows the chemical structure for every inhibitor of the three categories. Notably, almost all the compounds with a high anti-MvfR activity (columns 1 and 2) contain a nitro group at the position 5 of the benzimidazole ring. In contrast, this nitro group is lacking from all compounds with a low anti-MvfR activity (column 3). This suggests that the nitro group may play a critical role for the interaction with MvfR. This is most obvious when comparing the binding and inhibitory activity of M26 versus M51, two identical compounds but for the nitro group which is lacking in M26 (Figure 5). Indeed, the addition of the nitro group on M26 increases 25 times the compound binding to MvfR (Figure 2a) and 14.8 times the inhibition of 2-AA production in PA14 (Figure 1c). Moreover, replacing the methyl group at the position 5 of the benzimidazole ring of M27 by a nitro group as in M50 (Figure 5) increases 9.5 times the binding to MvfR (Figure 2a) and 14.3 times the inhibition of 2-AA production in PA14 (Figure 1c). Similarly, the addition of the nitro group on M55, as in M50 (Figure 5), increases 15.4 times the inhibition of 2-AA production in PA14 (Figure 1c). Notably, HHQ-ligand-based analogues harboring strong electron-withdrawing groups such as NO₂, CN, or CF₃ on the benzene moiety of the quinolone structure were also found to be strong MvfR inhibitors,^35,43 supporting our data on the importance of the nitro group for BB compounds to interact with MvfR. Future in-depth protein–inhibitor interaction studies will provide more insights on this aspect. Interestingly, the presence of the nitro group also appears to decrease the compounds anti-PqsBC activity. Indeed, adding the nitro group to B1 as in M62, or to M55 as in M50, or replacing the methyl group of M27 by a nitro group as in M50 reduce 4.6, 8.9, or 2.3 times, respectively, the HHQ production in the mvf R mutant strain constitutively expressing the pqs operon (Figure 1b). Overall, these data demonstrate that the nitro group at the position 5 of the benzimidazole ring is critical for the compounds to selectively recognize MvfR over PqsBC, and they suggest modifications on the benzamide moiety to further modulate target selectivity.

Figure 5.

Dual inhibitors structures sorted on the basis of their potency on each target.

CONCLUSIONS

This study provides exciting new insights into the mode of action and therapeutic potential of compounds with a BB core structure in the context of quorum sensing inhibition in P. aeruginosa. We previously reported a series of nonligand based BB compounds interfering with the P. aeruginosa MvfR QS system.³² In this study, we assessed further the mode of action of this BB compound series in this pathogen. While they all inhibit the MvfR QS system, we discovered that the transcriptional regulator MvfR is not their only target in the MvfR pathway. Our analysis shows that several of our BB compounds also interfere with the PqsBC enzyme, inhibiting its ability to convert 2-ABA into the MvfR-activating ligand HHQ. These compounds represent the first class of inhibitors ever reported to interfere with PqsBC, and they are the first synthetic inhibitors ever shown to interfere with both MvfR and PqsBC. These dual inhibitors harbor an exciting therapeutic potential because of their ability to block both acute and chronic virulence-related functions in P. aeruginosa and to simultaneously inhibit more than one target. As such, they offer new avenues to overcome the probability of resistance that might be presented with single target inhibitor in this pathogen.

METHODS

Bacterial Strains, Plasmids, and Growing Conditions.

PA14 (UCBPP-PA14) is a P. aeruginosa human clinical isolate.⁴⁴ The strain mvf R-pPqsABCD which has constitutive and MvfR-independent pqs operon expression was previously described in ref 32. Unless noted otherwise, all bacterial strains were grown in 5 mL of LB Lenox medium (Fisher Scientific) at 37 °C under 200 rpm orbital shaking using glass tubes (VWR). Tetracycline (75 μg/mL) was added when growing the mvf R- pPqsABCD strain to maintain the pDN18 plasmid.

HAQs and 2-AA Quantification.

HAQs and 2-AA levels were quantified in bacterial culture supernatants by LC/MS as described in refs 9,45.

Binding to MvfR via Surface Plasmon Resonance.

Purification of the MvfR ligand binding domain (MvfRc87) was performed as described in ref 32. MvfRc87 protein (50 μg/mL) was diluted in 10 mM sodium acetate buffer (pH 5.5) and immobilized on a CM7 Series S Sensor Chip using an Amine Coupling reagent kit (GE Healthcare) at the level of 3000–5000 Response Units (RU). PBS (pH 7.4) containing 0.05% P20 surfactant was used as the running buffer during protein immobilization.

The interactions between test compounds and MvfRc87 ligand binding domain were analyzed by Biacore T200 evaluation software 2.0 (GE Healthcare). During the measurement, 10 mM HEPES (pH 7.4) containing 150 mM NaCl, 3 mM EDTA, 0.05% P20 surfactant, and 4% DMSO was the running buffer. Injections were performed at a flow rate of 30 μL/min with 60 s of contact time and 420 s (Single-cycle kinetics) or 180 s (Multicycle kinetics) of dissociation time. Each injection was followed by an extra wash with 50% DMSO. Solvent correction was performed according to sensorgram analysis. The zero-concentration curve was subtracted from the other sensorgrams. The affinities of compounds were determined with the “Steady State Affinity” yielding the Binding Affinity Constant (K_D) and the maximum binding capacity (R_max) expressed as Response Units (RU).

Inhibition of PqsBC Enzymatic Activity.

PqsBC was purified and catalytic activity was determined as described previously.²⁵ 2-ABA was synthesized according to previous work.^24,25 Octanoyl-CoA and 2-AA were purchased from Sigma-Aldrich (St. Louis, MO, U.S.A.). For evaluating enzyme inhibition, 20–100 nM PqsBC was incubated with various concentrations of the respective inhibitor for 5 min before measuring the residual activity. 2-ABA and octanoyl-CoA concentrations were 150–200 μM and 20–40 μM, respectively, and the DMSO concentration was 1%. All assays were conducted in triplicate.

Binding to PqsBC via Fluorescence Spectroscopy.

Dissociation constants of PqsBC-inhibitor complexes were determined by fluorescence spectroscopy or fluorescence polarization spectrometry, using the fluorescence properties of the respective inhibitor molecule. 2-AA displacement from PqsBC by inhibitors was monitored by the 2-AA fluorescence intensity change resulting from binding of the molecule to PqsBC.²⁵ All assays were conducted in a buffer containing 50 mM HEPES, pH 8.0, 50 mM NaCl, and 1% DMSO, using a Jasco FP-6500 fluorescence spectrometer with polarization accessory.

Cell Viability Assays.

Cells survival to PA14 infection was assessed as previously described in ref 32. Briefly, bacterial cells were grown until midexponential phase (OD_{600 nm} = 2) in the presence or absence of each inhibitor at 50 μM. The cells were then washed and added to host cells at a MOI of 100. Three hours postinfection, bacterial cells were killed with 500 μg/mL Gentamycin and washed away twice with PBS. Cells were incubated in 100 μg/mL MTT for 16 h at 37 °C in 5% CO₂, then MTT was dissolved in DMSO and OD_{570 nm} was measured. All cells were maintained in 5% CO₂ at 37 °C. A549 (human lung epithelial cell line, ATCC, U.S.A.) and RAW264.7 cells (mouse macrophage cell line, IMGENEX, U.S.A.) were maintained in F12K and DMEM medium (Life Technologies, U.S.A.), respectively. The media were supplemented with 10% heat-inactivated FBS, penicillin/streptomycin, 2 mM L-glutamine, and 10 mM HEPES (all from Gibco). The cells were seeded in T-75 tissue culture flasks (Falcon, U.S.A.) and used between passages 2 and 3.

Antibiotic Tolerance.

P. aeruginosa cells were grown at 37 °C 200 rpm in 10g/L TSB media until midexponential phase (OD_{600 nm} 2) then exposed to 10ug/mL Meropenem (Sandoz, U.S.A.) for 24 h under the same incubating conditions. Before (t = 0) and after (t = 24 h) Meropenem addition, a 200 μL sample of each culture was collected, diluted and plated on LB agar plates to quantify the total number of bacteria (t = 0) and the surviving bacteria (t = 24 h). Colony-forming units (CFUs) were counted after 24 h of incubation at 37 °C. The fraction of antibiotic tolerant cells was then calculated as the ratio of the amount of total bacteria (t = 0) divided by the amount of surviving bacteria (t = 24 h). Data are expressed as the percentage of antibiotic tolerant cells relative to the DMSO vehicle control, which represents a survival fraction of 2.3 × 10⁻⁶ cells.

Statistical Analyses.

Statistical significance was assessed using unpaired one-way ANOVA + Dunnett’s post-test or one-way ANOVA + Tukey post-test as indicated using GraphPad Prism software.

Supplementary Material

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acschembio.6b01139

Further supporting data on compounds targets, cytotoxicity, and off-target assessments (PDF)