Evaluation of a Conformationally Constrained Indole Carboxamide as a Potential Efflux Pump Inhibitor in Pseudomonas aeruginosa

Efflux pumps in Gram-negative bacteria such as Pseudomonas aeruginosa provide intrinsic antimicrobial resistance by facilitating the extrusion of a wide range of antimicrobials. Approaches for combating efflux-mediated multidrug resistance involve, in part, developing indirect antimicrobial agents capable of inhibiting efflux, thus rescuing the activity of antimicrobials previously rendered inactive by efflux. Herein, TXA09155 is presented as a novel efflux pump inhibitor (EPI) formed by conformationally constraining our previously reported EPI TXA01182. TXA09155 demonstrates strong potentiation in combination with multiple antibiotics with efflux liabilities against wild-type and multidrug-resistant (MDR) P. aeruginosa. At 6.25 µg/mL, TXA09155, showed ≥8-fold potentiation of levofloxacin, moxifloxacin, doxycycline, minocycline, cefpirome, chloramphenicol, and cotrimoxazole. Several biophysical and genetic studies rule out membrane disruption and support efflux inhibition as the mechanism of action (MOA) of TXA09155. TXA09155 was determined to lower the frequency of resistance (FoR) to levofloxacin and enhance the killing kinetics of moxifloxacin. Most importantly, TXA09155 outperformed the levofloxacin-potentiation activity of EPIs TXA01182 and MC-04,124 against a CDC/FDA panel of MDR clinical isolates of P. aeruginosa. TXA09155 possesses favorable physiochemical and ADME properties that warrant its optimization and further development.


Introduction
Pseudomonas aeruginosa is an efficient opportunistic pathogen causing serious infections in patients who are mechanically ventilated, individuals who are immunocompromised, and patients with malignancies or HIV infection [1]. It is the most common Gram-negative organism identified in nosocomial pneumonia, and it can cause infection in immunocompromised hosts and infection in those with structural lung disease such as cystic fibrosis, surgical site infections, urinary tract infections and bacteremia [2]. In the Unites States, there are approximately 300,000 hospital-acquired pneumonia (HAP) infections per year, of which an estimated 64,800 are caused by P. aeruginosa, resulting in 27,200 annual deaths [3,4]. The pathogenicity of P. aeruginosa is largely caused by several virulence mechanisms such as secreted toxins, quorum sensing and biofilm formation, as well as genetic flexibility that enables it to survive in varied environments [1,2]. Lung injury associated with P. aeruginosa infection results from both the direct destructive effects of the organism on the lung parenchyma and exuberant host immune responses [1]. Treatment of Previously, TAXIS Pharmaceutical reported on the results of a synthetic screen seeking to replace the aryl alkyl head groups of a diaminopentanamide class of EPI with more druggable fragments in order to eliminate or diminish membrane disruption as a secondary MOA and improve metabolic and serum stability. This work produced TXA01182, a novel EPI that enhances the activity of multiple classes of antimicrobials with efflux liabilities in wild-type and MDR clinical isolates of P. aeruginosa while playing a minimal role in membrane disruption and avoiding the onset of resistance. In a hit-to-lead campaign for more powerful EPIs with better drug-like properties, the diamine side chain of TXA01182 was constrained at the C2-amine and the C4 positions to form a pyrrolidine containing EPI, TXA09155, with an identical indole head group and C2-stereochemistry (S) (Figure 1). The rationale for such a modification came from the well-established fact that conformation constraints and molecular flexibility have a strong effect on the activity of flexible molecules. There are many examples where conformationally constrained analogs are more bioactive or more specific than the non-constrained molecules [30][31][32]. In addition to the better potency, constraining flexible molecules may lead to drug candidates with better toxicity profiles and PK properties. To show if such a modification increased the potencies of TXA09155, its potentiation effects on different antibiotics were compared with TXA01182. TXA09155 was first assayed for its antibacterial activity against P. aeruginosa ATCC 27853, and the compound had a MIC of 50 µg/mL. As presented in Table 1, at 6.25 µg/mL, TXA09155 reduces the MICs of all tested antibiotics two-fold better than TXA01182, thus validating our design rationale. Once it was confirmed that constraining the diamine side chain of TXA01182 resulted in a more active EPI, an in-depth evaluation of TXA09155′s potentiation abilities on multiple antibiotics over a concentration range was carried out, and the results are presented in Table 2. The ability of TXA09155 to lower the MIC of multiple antibiotics with efflux liabilities in P. aeruginosa was tested at concentrations ranging from 25 to 3.13 µg/mL (1/2 to 1/16th MIC). TXA09155 potentiated most antibiotics in a concentrationdependent manner except for levofloxacin where all active concentrations showed the To show if such a modification increased the potencies of TXA09155, its potentiation effects on different antibiotics were compared with TXA01182. TXA09155 was first assayed for its antibacterial activity against P. aeruginosa ATCC 27853, and the compound had a MIC of 50 µg/mL. As presented in Table 1, at 6.25 µg/mL, TXA09155 reduces the MICs of all tested antibiotics two-fold better than TXA01182, thus validating our design rationale. Once it was confirmed that constraining the diamine side chain of TXA01182 resulted in a more active EPI, an in-depth evaluation of TXA09155 s potentiation abilities on multiple antibiotics over a concentration range was carried out, and the results are presented in Table 2. The ability of TXA09155 to lower the MIC of multiple antibiotics with efflux liabilities in P. aeruginosa was tested at concentrations ranging from 25 to 3.13 µg/mL (1/2 to 1/16th MIC). TXA09155 potentiated most antibiotics in a concentration-dependent manner except for levofloxacin where all active concentrations showed the same level of potentiation. Fluoroquinolones levofloxacin and moxifloxacin were potentiated by up to 16-and 128-fold, respectively. Tetracyclines doxycycline and minocycline were potentiated by up to 32-and 256-fold, respectively. Cefpirome, a cephalosporin, was potentiated up to 16-fold. Aztreonam, a monobactam, was potentiated up to 32-fold. Chloramphenicol and cotrimoxazole were potentiated up to 128-fold. As anticipated for an EPI, imipenem, which is not the substrate of RND efflux pumps in P. aeruginosa, was not potentiated by TXA09155 at any of the concentrations tested. These results are in line with TXA09155, maintaining the EPI activity of its parent compound TXA01182. The lowest active concentration for TXA09155 was 6.25 µg/mL. Thus, the 6.25 µg/mL concentration was chosen for future antimicrobial potentiation assays in P. aeruginosa to ensure that its MIC did not interfere with its potentiation properties.

TXA09155 Plays a Minimal Role in Membrane Disruption
To test if membrane disruption played a role in the potentiation of antibiotics seen in Table 1, the possibility of membrane disruption caused by TXA09155 was examined. Membrane disruption was assayed with two approaches: (1) a flow cytometry-based propidium iodide (PI) assay to monitor inner membrane permeabilization, and (2) a nitrocefin (NCF) assay to monitor outer membrane permeabilization. NCF is a chromogenic cephalosporin that changes from yellow to red when the amide bond in the β-lactam ring is hydrolyzed by a β-lactamase. The rate of hydrolysis in intact cells is slow, as it is limited by the rate of diffusion of periplasmic β-lactamase across the outer membrane. However, in the presence of an agent that permeabilizes the outer membrane, the rate of hydrolysis will increase [33]. The impact of TXA09155 on NCF hydrolysis is minimal at concentrations below 25 µg/mL, suggesting that TXA09155 does not interact with the outer membrane of P. aeruginosa at these concentrations ( Figure 2A). This result is comparable to what was previously reported for TXA01182 [29] suggesting that incorporation of the N-2-(4-aminomethylpyrrolidin-2-yl) methyl moiety that produced TXA09155 did not introduce membrane disruption properties to the compound. Polymyxin B was used as a positive control ( Figure 2B). TXA09155 did not disrupt the bacterial inner membrane below concentrations of 50 µg/mL compared with water and polymyxin B as vehicle and positive controls, respectively ( Figure 2C). In the PI assay, log-phase P. aeruginosa cells were mixed with various concentrations of TXA09155, followed by the addition of PI. Cells whose membranes remain intact exclude PI and remain non-fluorescent, while cells with compromised membrane integrity allow PI to enter and bind to DNA, resulting in fluorescence. phase P. aeruginosa cells were mixed with various concentrations of TXA09155, followed by the addition of PI. Cells whose membranes remain intact exclude PI and remain nonfluorescent, while cells with compromised membrane integrity allow PI to enter and bind to DNA, resulting in fluorescence.

Figure 2.
Outer-and inner-membrane permeabilization studies with TXA09155. Basal levels of NCF hydrolysis (A,B) or PI fluorescence (C) are observed upon addition of TXA09155 at concentrations below 12.5 and 50 µg/mL, respectively, indicating intact outer and inner membranes. Experiments were repeated at least three times, and a representative experiment is shown.

TXA09155 Inhibits the Efflux of Ethidium Bromide and Levofloxacin
The ability of TXA09155 to inhibit efflux in P. aeruginosa was studied using two quantitative assays to establish efflux inhibition as the MOA. In the first assay, the efflux of ethidium bromide (EtBr) by P. aeruginosa cells is studied in the presence of different concentrations of TXA09155. Wild-type P. aeruginosa ATCC 27853 cells were incubated with EtBr to allow for the intracellular accumulation and were treated with CCCP to inhibit active efflux. When bound to intracellular bacterial DNA, EtBr fluoresces brightly, while any unbound EtBr outside bacterial cells exhibits little or no fluorescence. Following activation by the addition of glucose, the efflux of EtBr can be followed in real time as a decrease in fluorescence based on the concentration of TXA09155. As seen in Figure 3A, the fluorescence intensity increased proportionally with the increasing concentration of TXA09155, indicating intracellular accumulation of EtBr and supporting a role in efflux inhibition by TXA09155.
The second assay quantifies the levels of levofloxacin accumulated inside P. aeruginosa after treatment with different concentrations of TXA09155. P. aeruginosa DA7232-harboring mutations in DNA gyrase (gyrA-T83I) and topoisomerase IV (parC-S80L) were used for this study since they are highly resistant to levofloxacin and thus can tolerate and accumulate high concentrations of the antimicrobial [34]. The levofloxacin MIC for this strain decreased from 256 to 1 µg/mL in the presence of 6.25 µg/mL TXA09155, making this strain ideal for the study. In the assay, P. aeruginosa DA7232 cells are incubated with levofloxacin and TXA09155 to allow for intracellular accumulation. After TXA09155 removal and membrane permeabilization, the levels of accumulated levofloxacin can be quantified fluorescently. Similar assays have been described elsewhere with the same purpose [35,36]. TXA09155 led to the accumulation of levofloxacin inside P. aeruginosa in a concentration-dependent manner ( Figure 3B) in support of a role in efflux inhibition.

TXA09155 Inhibits the Efflux of Ethidium Bromide and Levofloxacin
The ability of TXA09155 to inhibit efflux in P. aeruginosa was studied using two quantitative assays to establish efflux inhibition as the MOA. In the first assay, the efflux of ethidium bromide (EtBr) by P. aeruginosa cells is studied in the presence of different concentrations of TXA09155. Wild-type P. aeruginosa ATCC 27853 cells were incubated with EtBr to allow for the intracellular accumulation and were treated with CCCP to inhibit active efflux. When bound to intracellular bacterial DNA, EtBr fluoresces brightly, while any unbound EtBr outside bacterial cells exhibits little or no fluorescence. Following activation by the addition of glucose, the efflux of EtBr can be followed in real time as a decrease in fluorescence based on the concentration of TXA09155. As seen in Figure 3A, the fluorescence intensity increased proportionally with the increasing concentration of TXA09155, indicating intracellular accumulation of EtBr and supporting a role in efflux inhibition by TXA09155.

TXA09155 Does Not Affect the Proton Gradient across P. aeruginosa Inner Membrane
RND efflux pumps couple the influx of protons from the proton gradient across the inner membrane to the efflux of drugs and other molecules [37]. Therefore, it is important to verify that the MOA by which TXA09155 potentiates the antimicrobials mentioned in Table 1 does not involve disruption of the proton gradient. The effect of TXA09155 on the proton gradient of P. aeruginosa was assessed with a fluorescence-based 3,3′diethyloxacarbocyanine iodide [DiOC2 (3)] assay that was developed to measure membrane potential changes in Gram-negative bacteria [38]. DiOC2 (3) is a positively charged fluorescent dye that accumulates within cells in a charge-dependent manner, shifting its fluorescence spectrum from green to red at high concentrations due to concentration-dependent dye stacking [39]. As shown in Figure 4A, treatment of P. aeruginosa with TXA09155 at concentrations ranging from 20 to 0.01 µM (20 µM = 7.32 The second assay quantifies the levels of levofloxacin accumulated inside P. aeruginosa after treatment with different concentrations of TXA09155. P. aeruginosa DA7232-harboring mutations in DNA gyrase (gyrA-T83I) and topoisomerase IV (parC-S80L) were used for this study since they are highly resistant to levofloxacin and thus can tolerate and accumulate high concentrations of the antimicrobial [34]. The levofloxacin MIC for this strain decreased from 256 to 1 µg/mL in the presence of 6.25 µg/mL TXA09155, making this strain ideal for the study. In the assay, P. aeruginosa DA7232 cells are incubated with levofloxacin and TXA09155 to allow for intracellular accumulation. After TXA09155 removal and membrane permeabilization, the levels of accumulated levofloxacin can be quantified fluorescently. Similar assays have been described elsewhere with the same purpose [35,36]. TXA09155 led to the accumulation of levofloxacin inside P. aeruginosa in a concentration-dependent manner ( Figure 3B) in support of a role in efflux inhibition.
2.4. TXA09155 Does Not Affect the Proton Gradient across P. aeruginosa Inner Membrane RND efflux pumps couple the influx of protons from the proton gradient across the inner membrane to the efflux of drugs and other molecules [37]. Therefore, it is important to verify that the MOA by which TXA09155 potentiates the antimicrobials mentioned in Table 1 does not involve disruption of the proton gradient. The effect of TXA09155 on the proton gradient of P. aeruginosa was assessed with a fluorescence-based 3,3 -diethyloxacarbocyanine iodide [DiOC 2 (3)] assay that was developed to measure membrane potential changes in Gram-negative bacteria [38]. DiOC 2 (3) is a positively charged fluorescent dye that accumulates within cells in a charge-dependent manner, shifting its fluorescence spectrum from green to red at high concentrations due to concentrationdependent dye stacking [39]. As shown in Figure 4A, treatment of P. aeruginosa with TXA09155 at concentrations ranging from 20 to 0.01 µM (20 µM = 7.32 µg/mL TXA09155) did not lead to membrane depolarization, ruling this out as the MOA responsible for antibiotic potentiation. Azithromycin and CCCP were used as positive and negative controls, respectively.

TXA09155 Does Not Affect the Proton Gradient across P. aeruginosa Inner Membrane
RND efflux pumps couple the influx of protons from the proton gradient across the inner membrane to the efflux of drugs and other molecules [37]. Therefore, it is important to verify that the MOA by which TXA09155 potentiates the antimicrobials mentioned in Table 1 does not involve disruption of the proton gradient. The effect of TXA09155 on the proton gradient of P. aeruginosa was assessed with a fluorescence-based 3,3′diethyloxacarbocyanine iodide [DiOC2 (3)] assay that was developed to measure membrane potential changes in Gram-negative bacteria [38]. DiOC2 (3) is a positively charged fluorescent dye that accumulates within cells in a charge-dependent manner, shifting its fluorescence spectrum from green to red at high concentrations due to concentration-dependent dye stacking [39]. As shown in Figure 4A, treatment of P. aeruginosa with TXA09155 at concentrations ranging from 20 to 0.01 µM (20 µM = 7.32 µg/mL TXA09155) did not lead to membrane depolarization, ruling this out as the MOA responsible for antibiotic potentiation. Azithromycin and CCCP were used as positive and negative controls, respectively.

TXA09155 Does Not Deplete ATP Levels in P. aeruginosa
Another approach to assess any effect a drug may have on membrane integrity is to monitor ATP levels following drug treatment. Membrane disruption leads to membrane depolarization and impairment of the proton motive force (PMF), which is essential for a variety of critical bacterial processes, such as ATP synthesis [40]. As shown in Figure 4B,

TXA09155 Does Not Deplete ATP Levels in P. aeruginosa
Another approach to assess any effect a drug may have on membrane integrity is to monitor ATP levels following drug treatment. Membrane disruption leads to membrane depolarization and impairment of the proton motive force (PMF), which is essential for a variety of critical bacterial processes, such as ATP synthesis [40]. As shown in Figure 4B, treatment of P. aeruginosa with TXA09155 did not result in ATP depletion. CCCP was used as positive control.

TXA09155 Is Active in P. aeruginosa Mutants Overexpressing MexAB-OprM and MexXY-OprM Efflux Pumps
Next, the ability of TXA09155 to inhibit P. aeruginosa mutants overexpressing MexAB-OprM and MexXY-OprM was evaluated. Loss of function mutations in nalB and mexZ resulting in overexpression of MexAB-OprM and MexXY-OprM, respectively, have been identified in P. aeruginosa clinical isolates and are associated with resistance to cephalosporins, fluoroquinolones, and aminoglycosides [41][42][43][44][45][46][47]. An EPI should reverse the antimicrobial susceptibility lost in bacteria overexpressing efflux pumps. Therefore, the EPI activity of TXA09155 was studied using antibiotics that display efflux liabilities in these two RND efflux pumps (Table 3 and Supplementary Material Table S1). The level of cefpirome and levofloxacin potentiation seen with TXA09155 in MexAB-OprM and MexXY-OprM overproducing strains K1455 and K2415, respectively was comparable to the effect of inactivating both pumps by deleting oprM (strain K3698). The level of cotrimoxazole, doxycycline, minocycline, and chloramphenicol potentiation by TXA09155 was greater than the effect seen by inactivating MexAB-OprM and MexXY-OprM (compare K1455/K3698 ratios to K1455 ± TXA09155 and K2415 ± TXA09155). In agreement with these observations, TXA09155 did not significantly potentiate cefpirome or levofloxacin in the MexAB-OprM and MexXY-OprM-deficient strain K3698, but potentiated cotrimoxazole, doxycycline, minocycline, and chloramphenicol significantly in the same strain. Doxycycline and minocycline belong to the tetracycline class of antibiotics and thus inhibit protein synthesis through reversible binding to bacterial 30S ribosomal subunits [48]. Chloramphenicol is a broad-spectrum antibiotic that reversibly binds to the bacterial 50S ribosomal subunit inhibiting protein synthesis [49]. Cotrimoxazole is the antibacterial combination product of trimethoprim and sulfamethoxazole. These components inhibit enzyme systems involved in the bacterial synthesis of tetrahydrofolic acid and block the initiation of protein synthesis among other cellular processes [50][51][52][53]. Together, these data suggest that in addition to inhibiting efflux in MexAB-OprM and MexXY-OprM overproducing strains, TXA09155 may have a secondary MOA synergistic with protein synthesis inhibitors. Finally, imipenem, which is not the substrate of RND efflux pumps in P. aeruginosa, was not potentiated by TXA09155 in any of the strains tested. The ability of TXA09155 to potentiate levofloxacin in P. aeruginosa clinical isolates was tested in two outsourced studies. For both studies, TXA09155 was tested at a fixed concentration of 6.25 µg/mL. In the first study performed by Micromyx (Kalamazoo, MI, USA), 209 recent P. aeruginosa clinical isolates for which the levofloxacin MIC ranged between 0.03 to >32 µg/mL were collected from three hospitals across the US. Isolate selection was unbiased for resistance mechanisms. In this panel, TXA09155 decreased the levofloxacin MIC at which 50% of strains are inhibited (MIC 50 ) from 1 to 0.12 µg/mL and decreased the MIC 90 from 16 to 2 µg/mL ( Figure 5A and Supplementary Material Table S2). In the second study performed by IHMA (Switzerland), 300 recent MDR P. aeruginosa clinical isolates for which the levofloxacin MIC ranged between 0.06 to 64 µg/mL were collected from hospitals across 63 countries. In this panel, TXA09155 decreased the levofloxacin MIC 50 from 0.5 to 0.06 µg/mL and decreased the MIC 90 from 32 to 8 µg/mL ( Figure 5B and Supplementary Material Table S3). In both studies, the percentage of levofloxacin-resistant isolates dropped by 24% and 10%, respectively, highlighting the potential the TXA09155 scaffold has at reducing resistance and increasing susceptibility to antibiotics affected by efflux, such as levofloxacin.
collected from hospitals across 63 countries. In this panel, TXA09155 decreased the levofloxacin MIC50 from 0.5 to 0.06 µg/mL and decreased the MIC90 from 32 to 8 µg/mL ( Figure 5B and Supplementary Material Table S3). In both studies, the percentage of levofloxacin-resistant isolates dropped by 24% and 10%, respectively, highlighting the potential the TXA09155 scaffold has at reducing resistance and increasing susceptibility to antibiotics affected by efflux, such as levofloxacin.

TXA09155 Shows Superior Potency against Multidrug-Resistant P. aeruginosa Clinical Isolates from the CDC-FDA
The significant levofloxacin potentiation displayed in the previous P. aeruginosa panels containing resistant isolates was a promising sign that suggested that TXA09155 might retain its activity against MDR P. aeruginosa. Therefore, the EPI activity of TXA09155

TXA09155 Shows Superior Potency against Multidrug-Resistant P. aeruginosa Clinical Isolates from the CDC-FDA
The significant levofloxacin potentiation displayed in the previous P. aeruginosa panels containing resistant isolates was a promising sign that suggested that TXA09155 might retain its activity against MDR P. aeruginosa. Therefore, the EPI activity of TXA09155 was further tested against 34 MDR P. aeruginosa clinical isolates obtained from the CDC & FDA Antibiotic Resistance (AR) Isolate Bank [54]. The genomes of these MDR isolates have been sequenced and include mutations associated with MexAB-OprM overexpression (nalC-G71E and mexR-V126Q), mutations that lead to fluoroquinolone resistance (gyrA-T83I and gyrA-T133H), as well as other known resistant mechanisms (Supplementary Material  Table S5) [34,[55][56][57][58][59][60][61]. The levofloxacin MIC in these 34 MDR isolates ranged from 0.5 to 128 µg/mL. Again, TXA09155 was tested at a fixed concentration of 6.25 µg/mL. TXA09155 was able to decrease the levofloxacin MIC 50 from 32 to 2 µg/mL and decrease the MIC 90 from 64 to 8 µg/mL ( Figure 5C and Supplementary Material Table S4). More importantly, the TXA09155/levofloxacin combination lowered the percentage of levofloxacin-resistant isolates from 85% to 35%. The activity of TXA09155 outperformed the activity of its parent compound TXA01182 (24), which lowered the percentage of levofloxacin-resistant isolates from 85% to 56% when combined with levofloxacin, again confirming that constraining the diamine side chain of TXA01182 resulted in a more active EPI. MC-04,124, a known EPI active against P. aeruginosa [62], failed to decrease the MIC 50 , MIC 90 , and levofloxacinresistance percentage when tested at the same concentration as TXA09155 and TXA01182, which highlights the superior potency of the TAXIS EPI scaffolds.

TXA09155 Lowers the Frequency of Resistance to Levofloxacin
In addition to reducing the levels of intrinsic resistance, a potent EPI is also expected to significantly reverse acquired resistance as well as decrease the frequency at which antimicrobial resistance emerges. To test whether TXA09155 increases the selective pressure on resistant mutant emergence when combined with antimicrobials, the effect of TXA09155 on the frequency of resistance (FoR) of P. aeruginosa ATCC 27853 to levofloxacin was tested. TXA09155 reduced the FoR to levofloxacin more than 149-fold ( Table 4). The undetectable levels of resistance seen in the TXA09155/levofloxacin combination would be of great value in a clinical setting, particularly in cystic fibrosis patients infected with P. aeruginosa, where patients are colonized by hypermutable strains that persist for years [63].

Genes Involved in Resistance to TXA09155 Alone or Combined with Levofloxacin
A study was conducted to understand how P. aeruginosa becomes resistant to TXA09155, alone or in combination with levofloxacin. To understand resistance to TXA09155 alone, P. aeruginosa ATCC 27853 was exposed to four times or one time the MIC of TXA09155 (200 or 50 µg/mL, respectively). All resistant colonies were verified by determining the MIC to TXA09155. Resistance to TXA09155 did not result in cross-resistance to other known antibiotics (Supplementary Material Table S6). Resistance to four times the MIC of TXA09155 alone arose at a frequency of 1.40 × 10 −6 and resulted in single point mutations in the phoQ gene, the sensor kinase component of the two-component regulatory system PhoP-PhoQ (Table 5, strain EPIR1S, EPIR9S, and EPIR20L). This two-component system regulates the expression of more than 100 genes in P. aeruginosa, including genes that code for efflux components (mexX), and plays a role in the resistance to cationic antimicrobial peptides and aminoglycoside antibiotics [64,65]. Resistance to one time the MIC of TXA09155 alone occurred at a frequency of 1.9 × 10 −3 and resulted in a single point mutation that introduced an early stop codon in the ompH gene (strain EPIR43). OmpH is a homolog of outer membrane chaperone Skp [66]. It has been found to be overexpressed in P. aeruginosa isolates resistant to ampicillin and kanamycin [67]. Deletion of ompH leads to hyper susceptibility to ertapenem, ceftazidime, levofloxacin, tigecycline, and cotrimoxazole [68]. In agreement with these findings, EPIR43 gained susceptibility to levofloxacin, ceftazidime, tigecycline, doxycycline, meropenem, amikacin, and azithromycin (Supplementary Material Table S6). The level of susceptibility gained to levofloxacin and doxycycline matches the level of potentiation observed with the lowest active concentration of TXA09155 (Table 1). In E. coli, more than 30 envelope proteins have been found to interact with the Skp chaperone [69,70]. E. coli 10omph genetically interacts with AcrD, a component of the AcrAD-TolC aminoglycoside efflux pump [71,72]. The amino acid sequence of AcrD shares 62% similarity with P. aeruginosa MexB. Thus, it is possible that in P. aeruginosa, OmpH is involved in folding an efflux pump component targeted by TXA09155. If so, loss of function of ompH in P. aeruginosa could result in misfolded efflux pump components and therefore diminished antibiotic efflux leading to hypersensitivity. Alternatively, TXA09155 could be directly inhibiting the function of OmpH, resulting in misfolded efflux pump components and antibiotic hypersusceptibility.
Resistance to the TXA09155/levofloxacin combination was studied in the P. aeruginosa strain DA7232, which as mentioned above, is resistant to levofloxacin due to its DNA gyrase and Topoisomerase 4 mutations [34]. Since this strain is resistant to fluoroquinolones, it was hypothesized that resistance would arise predominantly from mutations related to efflux or additional mechanisms of action in this genetic background. P. aeruginosa DA7232 was exposed to 3.13 µg/mL of TXA09155 (lowest concentration that potentiated levofloxacin eight-fold) and 256 µg/mL of levofloxacin (1X-MIC) to obtain resistant colonies. Resistance to the combination occurred at a frequency of 2.48 × 10 −8 and resulted in single point mutations in the tryptophan-tRNA ligase gene, trpS (strain EPIR24L), suggesting a possible role in protein synthesis inhibition for TXA09155. If protein synthesis inhibition is a secondary mechanism of action of TXA09155, this would explain why it shows synergy with ribosome inhibitors doxycycline, minocycline, and chloramphenicol in efflux-deficient strain K3698 (Table 2).

Time-Kill Assay
In addition to assessing the potentiation activity of TXA09155 in vitro, its potentiation of a minimally bactericidal concentration of moxifloxacin (1×-MIC) was probed against P. aeruginosa ATCC 27853 with time-kill studies. Figure 6 shows time-kill curves with moxifloxacin alone or combined with TXA09155. By itself, TXA09155 did not affect the growth of P. aeruginosa ATCC 27853 at the highest concentration tested (gray curve). TXA09155 enhanced moxifloxacin's killing kinetics in a concentration-dependent manner (green, purple and black curves). After 3 h of incubation, the combinations killed more bacteria by a magnitude of ≥3 logs compared to moxifloxacin alone (orange curve). After 6 h of incubation, the combinations increased bacterial killing by a magnitude of ≥3.9 logs com-pared to moxifloxacin alone. At 24 h, the combinations achieved ≥ 4 logs of killing more than moxifloxacin alone. These results suggest that the killing kinetics for the combination of TXA09155/moxifloxacin are faster than those of moxifloxacin alone. The potentiation of a minimally bactericidal concentration of moxifloxacin (1×-MIC) by TXA09155 seen in Figure 6 is consistent with similar studies performed with a TXA01182/levofloxacin combination in P. aeruginosa [29]. of a minimally bactericidal concentration of moxifloxacin (1×-MIC) was probed against P. aeruginosa ATCC 27853 with time-kill studies. Figure 6 shows time-kill curves with moxifloxacin alone or combined with TXA09155. By itself, TXA09155 did not affect the growth of P. aeruginosa ATCC 27853 at the highest concentration tested (gray curve). TXA09155 enhanced moxifloxacin's killing kinetics in a concentration-dependent manner (green, purple and black curves). After 3 h of incubation, the combinations killed more bacteria by a magnitude of ≥3 logs compared to moxifloxacin alone (orange curve). After 6 h of incubation, the combinations increased bacterial killing by a magnitude of ≥3.9 logs compared to moxifloxacin alone. At 24 h, the combinations achieved ≥ 4 logs of killing more than moxifloxacin alone. These results suggest that the killing kinetics for the combination of TXA09155/moxifloxacin are faster than those of moxifloxacin alone. The potentiation of a minimally bactericidal concentration of moxifloxacin (1×-MIC) by TXA09155 seen in Figure 6 is consistent with similar studies performed with a TXA01182/levofloxacin combination in P. aeruginosa [29].

TXA09155 Has a Favorable Physiochemical and ADME Profile
Along with its microbiological evaluation, TXA09155 was evaluated for its physiochemical and in vitro ADME properties. In general, TXA09155 follows Lipinski's rule-offive, a rule describing molecular properties believed to be important for a drug's pharmacokinetics in the human body, which is highly soluble (>155 µM at pH 7.4). Table 6 summarizes the in vitro chemical absorption, distribution, metabolism, excretion, and toxicity (ADMET) profile of TXA09155. Overall, the metabolic stability of TXA09155 is good across two species (human and rat). It did not inhibit cytochrome P450 isoforms having IC 50 values greater than 100 µM for CYP1A2, CYP2C19, CYP2C9, and 67.2 and 28.5 µM for CYP2D6 and CYP3A4, respectively. The observed cytotoxicity IC 50 values of TXA09155 in CellTiterGlo TM assay (293T, A549) were all greater than 40 µM. Additionally, TXA09155 did not display hemolytic activity in human RBCs in the 12.5 to 100 µM concentrations range. The only safety concern for TXA09155 was its hERG liability, which has an IC 50 of 16 µM in the patch clamp assay.

Synthesis Reagents, Analytical Instruments and Reaction Conditions
All reactions, unless otherwise stated, were conducted under nitrogen atmosphere. Reaction monitoring and follow-up were performed using aluminum backed Silica G TLC plates (UV254 e Sorbent Technologies, Norcross, Georgia), visualized with ultraviolet light. Flash column chromatography was performed on a Combi Flash Rf Teledyne ISCO using hexane, ethyl acetate, dichloromethane, and methanol. The 1 H (300 MHz) spectra were performed in CDCl 3 , Methanol-d 4 , and DMSO-d 6  The synthetic scheme for the preparation of TXA09155 is outlined in Figure 7. In order to maintain the same C2-amine stereochemistry as in TXA01182 (S-stereochemistry), the synthesis started from commercially available (2S,4R)-4-hydroxypyrrolidine-2-carboxylic acid 1. For initial proof of concept regarding gain of activity with a conformationally constrained scaffold, the stereochemistry of the 4-hydroxy group was not taken into consideration at this stage. Esterification of 1 followed by benzylation afforded intermediate compound 3. Tosylation of the hydroxy group and then replacing it with the CN-group changed the stereochemistry at the C4. Reduction of the methyl ester of 5 with LiAlH 4 afforded the hydroxyl methyl derivative 6. Two-step Mitsunobu protocol was then followed to convert the hydroxyl group of 6 to the amine group as shown in compound 8 in 87% yield over two steps. Coupling of 8 with known indole acid 9 afforded the amide compound 10 in 60% yield. Nitrile reduction and debenzylation were achieved by two-step hydrogenation procedure using Raney-Ni and Pd/C as catalysts. Finally, TXA09155 was synthesized as a HCl salt in 64% yield over 3 steps.

Minimum Inhibitory Concentration (MIC) Assay for Potentiation of Antimicrobial Activity against P. aeruginosa
MIC assays were performed as described previously [29].

Flow Cytometry Assay for Permeabilization of Inner Cell Membranes to Propidium Iodide (PI) in P. aeruginosa
A flow cytometry assay used for assessing potential inner membranes permeabilization of P. aeruginosa bacterial cells to PI was conducted using the LIVE/DEAD BacLight Kit from Invitrogen (Waltham, MA, USA). Briefly, log-phase P. aeruginosa ATCC 27853 bacterial cells grown in BHI broth were diluted 5-fold in PBSM (1X PBS, 1MgCl 2 ) to an approximate concentration of 6.5 × 10 7 CFU/mL. The bacteria were aliquoted into tubes and mixed with TXA09155 at concentrations ranging from 1 to 1/8th times the MIC (50 to 6.25 µg/mL). Water alone was used as solvent control. Polymyxin B was used as a positive control. Intracellular PI fluorescence was detected by flow cytometry using a CytoFlex (Beckman Coulter Inc., Brea, CA, USA). The 488 nm laser was used for excitation, with the PC5.5 and FITC channels being used for emission. For each sample, the fluorescence of 10,000 individual bacterial cells was measured, and the percent of cells that stained positive for PI fluorescence was calculated.

Nitrocefin Cellular Assay for Outer Cell Membrane Permeabilization Assessment in P. aeruginosa
Assessment of outer membrane permeabilization was carried out as described previously [29].

Fluorescence-Based Cellular Assay for Inhibition of Pump-Mediated Efflux of Ethidium Bromide (EtBr)
Efflux of EtBr from P. aeruginosa in the presence of TXA09155 was carried out as described previously [29].
3.8. Levofloxacin Accumulation Assay P. aeruginosa DA7232 was grown to an OD 600 of 0.6 and treated with levofloxacin (64 µg/mL) alone or in combination with compounds at sub-inhibitory concentrations (1/2, 1/4th and 1/8th MIC). The treated bacterial culture samples were incubated on ice for 15 min, centrifuged, washed with 1X PBS (pH 7.4), and then resuspended in 1 mL of glycine-HCl buffer (pH 3.0) overnight, followed by centrifugation. The fluorescence of 100 µL of supernatant was read at 490 nm following a 355 nm excitation in a SpectraMax iD5 spectrophotometer (Molecular Devices, San Jose, CA, USA).

Membrane Polarization Assay
P. aeruginosa membrane polarization assays were conducted as described previously [38] using strain ATCC 27853. Briefly, exponentially grown P. aeruginosa culture (OD 600 of 0.7-0.8) was harvested by centrifugation, and the pellet was resuspended to an OD 600 of 1.0 in sodium phosphate buffer (0.13 M NaCl, 7 mM Na 2 HPO 4 , 3 mM NaH 2 PO 4 ; pH 7.0). To minimize the exclusion of DiOC 2 (3) dye, bacteria were treated with 10 mM EDTA for 5 min, centrifuged and resuspended to a final OD 600 of 1.0 in the resuspension buffer (130 mM NaCl, 60 mM Na 2 HPO 4 , 60 mM NaH 2 PO 4 , 10 mM glucose, 5 mM KCl, 0.5 mM MgCl 2 ). Cells were loaded with 30 µM DiOC 2 (3) and added to a 96-well opaque microplate for a final volume of 200 µL in each well. Prior to the addition of cells, an indicated compound and the controls were added to the wells. Compounds were incubated with DiOC 2 (3) loaded cells for 30 min at 37 • C. DiOC 2 (3) fluorescence was recorded in a microplate reader using 450 nm excitation and 670 nm emission (for red fluorescence). Background correction for each concentration of each compound was obtained from reading the fluorescence of compounds in the resuspension buffer without cells. After background correction, fluorescence was normalized to that of the vehicle (DMSO) only. After a 30 min incubation at 37 • C, DiOC 2 (3) fluorescence was recorded using a SpectraMax iD5 spectrophotometer (Molecular Devices, San Jose, CA, USA) following excitation at 450 nm. Red fluorescence intensity was recorded at 670 nm emission.

Determination of Intracellular ATP Levels
P. aeruginosa intracellular ATP levels were determined as described previously [36] using an ATP determination kit (catalog number A22066; Thermo Fisher Scientific, Waltham, MA, USA). A bacterial culture of strain ATCC 27853 was grown to the mid-log phase (OD 600 = 0.7), washed, and resuspended in the same volume of PBSM (1X PBS, 1mM MgCl 2 ). The bacterial culture was treated with sub-inhibitory concentrations of compounds (1/4th and 1/8th MIC) for 3 h at 37 • C. After treatment, the cells were lysed in chloroform, which was subsequently removed by boiling at 80 • C. The persistent ATP from cell lysate was measured in a 96-well black flat-bottom plate and plotted as relative luminescence units. A culture treated with CCCP (12.5 and 6.25 µg/mL) was included as positive controls. Error bars represent the standard deviation of triplicates.

Frequency of Resistance (FoR) Study
The frequency of resistance studies was carried out as described previously [29].

Whole-Genome Sequencing
DNA extraction, library preparation, and whole-genome sequencing of P. aeruginosa parent strains ATCC 27853 and DA7232, and single isolates EPIR1S, EPIR9S, EPIR20L, EPIR43, and EPIR24L were performed by CD Genomics (New York, NY, USA) using Illumina. Sequencing reads were mapped using the published genome of P. aeruginosa ATCC 27853 (GenBank accession number CP011857) or the published genome of P. aeruginosa PAO1 (parent strain of DA7232) as reference genomes with the BWA-MEM tool from the Galaxy web platform (https://usegalaxy.org/ last accessed on 1 August 2021.) [75]. Variations in the genomes between resistant strains and parent strains were identified using the LoFreq tool from the same platform.

Time-Kill Studies
Time-kill studies were carried out as described previously [29] with the following changes. When indicated, moxifloxacin was added to the prepared bacterial suspensions at one time the MIC (2 µg/mL). TXA09155 was added to bacterial suspensions at 1/8th, 1/12.5th, and 1/16th times the MIC (6.25, 4, and 3.125 µg/mL, respectively).

Metabolic Stability Studies and hERG Assay
The metabolic stability of TXA09155 was determined in human, mouse, and rat liver microsomes by Eurofins (St. Charles, MO, USA) using the substrate depletion method (in vitro half-life (t 1/2 ) method). Quantification was performed by LC/MS/MS. In this assay, depletion of the parent compound is measured over time to determine the compound's in vitro t 1/2 and hepatic extraction ratios (ER). This is used to predict metabolic clearance. Inhibition of the hERG channel by TXA09155 was determined by Charles River Laboratories (Cleveland, OH, USA). This assay utilizes electrophysiology measurements on the electric current passing through the hERG channel heterologously expressed in a stable CHO cell line. Channels are open by a hERG-specific voltage protocol, and the compound effects are directly characterized by their block of the hERG current. The assay is performed on an automated platform, Q-Patch.

Conclusions
This work was designed to improve the activities of previously reported EPI TXA01182 by adopting the principle of constraining molecular flexibility. A common practice during the medicinal chemistry optimization of hit/lead compounds involves the application of conformation constraints since this can have a strong effect on the binding affinity of flexible molecules. Conformation constraint can be achieved by cyclization, enantiomeric substitutions, or by the introduction of stereochemically constraining chemical groups. With this in mind, the medicinal chemistry team at TAXIS Pharmaceuticals sought to constrain the diamine sidechain moiety of previously reported EPI TXA01182 by cyclization of the C2-amine and C4 positions. This resulted in TXA09155, an EPI twice as active in regard to potentiating the MIC of antibiotics from multiple classes with known efflux liabilities in wild-type P. aeruginosa. This gain in activity appears to be of clinical relevance since the activity TXA09155 displayed in wild-type P. aeruginosa translated across hundreds of clinical isolates with 334 of them being MDR isolates. The gain of activity in TXA09155 could be explained by the conformation constraint of its diamine sidechain or by a possible secondary mechanism of action. TXA09155 was able to significantly potentiate ribosome inhibitors doxycycline, minocycline, and chloramphenicol in strains deficient in MexAB-OprM and MexXY-OprM efflux pumps while not potentiating fluoroquinolones or cephalosporins in the same strain. In addition, resistance to the TXA09155/levofloxacin combination was achieved with a point mutation in the tryptophan-tRNA ligase trpS, suggesting a possible role in protein synthesis inhibition by TXA09155. Resistance to TXA09155 alone did not result in cross resistance to antipseudomonal antibiotics, which would be an advantage at the clinical setting where antibiotic treatment can result in multidrug resistance associated with mutations in multidrug efflux pumps [76,77]. Finally, the ADMET profile of a drug scaffold plays a key role in drug discovery and development. A high-quality drug candidate should show appropriate ADMET properties. Not surprisingly, many drug candidates fail to become drugs due to suboptimal ADMET properties. TXA09155 showed an overall good ADMET profile, suggesting that this EPI class has the potential to become a drug. A detailed SAR study and optimization effort resulting in further optimized candidates from this class will be reported next. This report will address acute toxicity, cytotoxicity, nephrotoxicity, hERG binding, PK parameters and lastly in vivo efficacy.
Although a portion of this work concentrates on TXA09155/levofloxacin activity against MDR clinical isolates, an MDR phenotype is only one of the multiple factors that affects the outcome of P. aeruginosa treatment at the clinic. P. aeruginosa possesses an overabundance of virulence determinants, including the production of biofilm, pigments, exotoxins, proteases, flagella, and secretion systems, the majority of which show no relevant correlation with an isolate's MDR phenotype [78]. Additionally, about half of P. aeruginosa isolates from burn victims are carbapenem-resistant due to various mechanism [79], highlighting the potential beneficial outcome that TXA09155/levofloxacin treatment would have on these cases. Finally, time-consuming practices of pathogen identification and antibiotic susceptibility testing delay diagnosis and force clinicians to employ empirical antibiotic treatment, risking the development of antibiotic resistance [80].  Table S1: P. aeruginosa efflux liability of antibiotics tested in this manuscript; Table S2: Cumulative susceptibility study of P. aeruginosa clinical isolates to TXA09155, levofloxacin and the combination; Table S3: Cumulative susceptibility study of P. aeruginosa clinical isolates to TXA09155, levofloxacin and the combination; Table S4: Cumulative susceptibility percentage of P. aeruginosa clinical isolates to TXA09155, levofloxacin and the combination; Table S5: Levofloxacin potentiation comparison between TXA09155, TXA01182 and MC-04,124 on multidrugresistant clinical isolates of P. aeruginosa; Table S6: Susceptibility of P. aeruginosa mutants resistant to TXA09155 or TXA09155/levofloxacin combination to various antimicrobials. References [26,[46][47][48][49][50][51][52] are cited in the supplementary materials

Conflicts of Interest:
All the authors are associated with TAXIS Pharmaceuticals Inc. and thus declare conflicts of interest.