Identification of Compounds with pH-Dependent Bactericidal Activity against Mycobacterium tuberculosis.

To find new inhibitors of Mycobacterium tuberculosis that have novel mechanisms of action, we miniaturized a high throughput screen to identify compounds that disrupt pH homeostasis. We adapted and validated a 384-well format assay to determine intrabacterial pH using a ratiometric green fluorescent protein. We screened 89000 small molecules under nonreplicating conditions and confirmed 556 hits that reduced intrabacterial pH (below pH 6.5). We selected five compounds that disrupt intrabacterial pH homeostasis and also showed some activity against nonreplicating bacteria in a 4-stress model, but with no (or greatly reduced) activity against replicating bacteria. The compounds selected were two benzamide sulfonamides, a benzothiadiazole, a bissulfone, and a thiadiazole, none of which are known antibacterial agents. All of these five compounds demonstrated bactericidal activity against nonreplicating bacteria in buffer. Four of the five compounds demonstrated increased activity under low pH conditions. None of the five compounds acted as ionophores or as general disrupters of membrane potential. These compounds are useful starting points for work to elucidate their mechanism of action and their utility for drug discovery.

Tuberculosis (TB), caused by Mycobacterium tuberculosis, is the leading cause of death among infectious diseases.[1] Treatment for tuberculosis is 6–9 months long for drug sensitive strains and even longer for resistant strains.[1] Given the long treatment time and prevalence of the disease, there is urgent need for new drugs.

M. tuberculosis is an intracellular pathogen that can replicate in macrophages and persist within the lung for long periods.[2]M. tuberculosis has numerous mechanisms by which it can evade the immune system, including its ability to inhibit fusion of phagosomes and lysosomes, and avoid antimicrobial activities such as acidification of the compartment it occupies.[3,4]M. tuberculosis can survive and replicate within resting macrophages. However, activated macrophages can mount an effective response to M. tuberculosis infection, and if acidification of the compartment occupied by M. tuberculosis occurs, bacterial killing is observed.[5] The pH of phagosomes occupied by M. tuberculosis ranges from 4.5 to 6.2, which is dependent on the state of activation of the macrophage.[6−11] The acidic microenvironment itself may or may not be bactericidal to M. tuberculosis but can make M. tuberculosis more vulnerable to other stresses.[12−15] As a result, acidification of the microenvironment is a mechanism for inducing bacterial death. Thus, the ability of M. tuberculosis to maintain its internal pH within narrow parameters is critical for survival in macrophages.

In the host environment, M. tuberculosis must be able to sense external pH and to regulate its internal pH to maintain metabolic activity and viability. M. tuberculosis mutants defective in blocking acidification, for instance, by deletion of the periplasmic serine protease Rv3671c (MarP),[16,17] are unable to survive acid stress in vitro and are attenuated in mouse models of infection.[18−20] Survival in low pH is partly dependent on cell density, a phenomenon attributed to cell to cell signaling in other bacteria.[15,20,21] OmpA, an outer membrane protein,[22] and Ppm1 and PonA, both involved in cell wall synthesis,[15] are also involved in M. tuberculosis survival under low pH. Genes that are induced when the external pH in culture drops rapidly[23] are also up-regulated in human and mouse infected lungs and in intracellular bacteria.[24−27] Thus, pH sensing and homeostasis are important processes for M. tuberculosis survival in its host. Regulatory mechanisms such as PhoPR[28,29] and the transcription factor WhiB3 are involved in the response to low pH,[30] since they control the pH-mediated expression of genes including lipF and the apr locus.[31] In addition adenylyl cyclases may sense external pH changes.[32,33] Further evidence suggests that pH homeostasis is a viable drug target, as both nitazoxanide and the imidazopyridine series disrupt intrabacterial pH (pHIB),[34,35] as do the benzoxaninones targeting MarP,[18]and pyrazinamide, a compound that is active against M. tuberculosis under acidic conditions through an unclear mechanism of action, is a critical part of the TB treatment regimen.[36]

We are interested in developing antitubercular agents with novel mechanisms and targets. We focused on an approach to find compounds that work at low pH and might not be otherwise found in a standard screening campaign. Our screening approach was to identify compounds that disrupt the ability of M. tuberculosis to maintain its internal pH in an acidic environment using a fluorescent reporter protein that can monitor pH. We reasoned compounds disrupting pH homeostasis would have bactericidal properties under acidic conditions.

Results

Assay Development and Validation

We adapted a medium throughput screen (96-well format) that uses a pH-sensitive fluorescent reporter or pHLUOR (rGFP).[37] This approach was previously used to identify compounds that disrupt pH homeostasis from a small library of 1980 natural products.[37] We wanted to increase the throughput of the screen to enable us to screen tens of thousands of compounds, so we adapted and miniaturized the assay to 384-well format.

We generated a recombinant strain of M. tuberculosis H37Rv expressing a ratiometric GFP (rGFP) using the same vector as previously described.[37] The pHLUOR demonstrates a pH-dependent excitation wavelength; we confirmed that we could generate a similar calibration curve to determine pH from the ratio of the fluorescence of Ex395/Em510 and Ex475/Em510 using M. tuberculosis lysates in buffered solutions (Figure A). These data confirmed that the rGFP was working as expected and the standard curve was used to derive intrabacterial pH (pHIB) for all future work. The standard curve and line of best fit were confirmed with a second set of lysates at the end of the screen.

Figure 1

High throughput screen. (A) Lysates were generated from M. tuberculosis expressing pH-sensitive green fluorescent protein in buffered solution. Fluorescence was measured at excitation/emission 400/516 nm and 485/516 nm, and the ratio was calculated. Data are average and standard deviation from 2 independent experiments. The line of best fit using the least-squares method was derived to generate a standard curve. (B) Duplicate plates for minimum signal (5 μM monensin), midpoint signal (0.5 μM monensin), and maximum signal (DMSO) were tested on 3 different days using M. tuberculosis incubated in phosphocitrate buffer at pH 4.5 for 48 h. Day 1, black; day 2, dark gray; day 3, light gray. The Z′ was >0.5 for each plate. (C) For each compound, the intrabacterial pH was calculated based on the standard curve in panel A; compounds that dropped the pH < 6.5 were considered active.

We selected similar conditions to the published screen using 0.15 M phosphate citrate buffer, pH 4.5, plus tyloxapol; we used the ionophore monensin as a control.[38] After optimization for higher throughput, our final assay parameters were as follows: 25 μL of late logarithmic phase M. tuberculosis in phosphate citrate buffer at an OD590 of 0.6; 24 μL of phosphate citrate buffer, pH 4.5; 1 μL of compound; Greiner black 384-well plates with a clear bottom; 48 h incubation with compound at 37 °C. We validated the robustness of the assay according to NCGC guidelines by performing a reproducibility test. We ran duplicate plates for maximum signal (DMSO only), minimum signal (5 μM monensin), and midpoint signal (0.5 μM monensin) on three separate days[39] (Figure B). We used the standard curve to convert the ratio of the two fluorescence readings into pH. The average pH was 6.45 ± 0.07 for 5 μM monensin, 6.56 ± 0.05 for 0.5 μM monensin, and 6.88 ± 0.03 for DMSO. The assay passed reproducibility testing for both intraexperiment and interexperiment variability measures; % CV was <20%, signal/background and signal/noise were >5, and the Z′ of controls for plates was >0.5 (Figure B).

Once we had validated the screen for reproducibility, we tested several reference compounds (Table ). None of the compounds we tested were active, with the exception of carbonyl cyanide 3-chlorophenylhydrazone (CCCP), a known proton ionophore and pyrazinamide (PZA). We defined the minimum active concentration 6.5 (MAC6.5) as the minimum concentration required to decrease M. tuberculosis pHIB < 6.5. CCCP was most effective, with a MAC6.5 of 0.5 μM. PZA had activity but only at higher concentrations, with a MAC6.5 of 69 μM.

Table 1

Reference Compoundsa

compound	MAC6.5 (μM)
carbonyl cyanide 3-chlorophenylhydrazone	0.5 ± 0.0
metronidazole	>100
pyrazinamide	69 ± 26
ethambutol	>100
rifampicin	>100
ofloxacin	>100
d-cycloserine	>100
kanamycin	>100
isoniazid	>100
moxifloxacin	>100

M. tuberculosis expressing the pH-sensitive green fluorescent protein was incubated with compounds in phosphocitrate buffer at pH 4.5 for 48 h. Fluorescence was measured at excitation/emission 395/510 and 475/510, and the ratio was converted into pH using the standard curve from Figure . MAC6.5 is the concentration required to reduce the intrabacterial pH < 6.5. The assay was repeated at least twice; results are average ± standard deviation.

High Throughput Screen

Once we had validated the assay, we selected a diverse compound set to screen. We selected the screening concentration using three plates from the library selected at random. We tested each plate at 5 μM, 10 μM, and 20 μM. We defined compounds as active if they reduced the pH below 6.5.[37] Out of 949 compounds tested at 5 μM, only 1 compound was active, while 2 compounds were active at 10 μM, and 6 compounds were active at 20 μM (data not shown). We selected 20 μM as the test concentration for the screen, since it had the highest hit rate. We screened a set of 89 273 compounds from the Eli Lilly corporate collection, which were selected to represent the chemical diversity in the entire library of >800 000 compounds (using methods proprietary to Lilly). We screened the compounds at a fixed concentration of 20 μM (Figure C). The Z′ for each plate was determined, which ranged from 0.53 to 0.8 (average 0.77). We identified 605 hits that reduced pHIB < 6.5, making a hit rate of 0.7%.

Confirmation of Hit Activity

We confirmed the activity of our hit compounds. Compounds were resupplied from Eli Lilly and tested as a 10-point serial dilution series. We calculated the MAC6.5 for each compound using a starting concentration of 200 μM. We confirmed 556 of the 605 hits, a confirmation rate of 92%

Activity of Compounds against Replicating and Nonreplicating M. tuberculosis

In order to select the best compounds for further work, we determined which compounds had activity in other antitubercular assays. We determined the minimum inhibitory concentration (MIC) against replicating bacteria grown under standard aerobic conditions, as well as against nonreplicating bacteria generated using a multistress assay.[40] We were particularly interested in finding molecules with activity against nonreplicating bacteria, which can be hard to target. Forty-eight compounds were active in nonreplicating conditions. From this set, compounds were identified and grouped into structural classes for further study, and five compounds with a range of activity from 1.4 to 100 μM against nonreplicating bacteria were selected for further study (Table ). Compounds IDR-0020850 and IDR-0054790 are benzamide sulfonamides, where IDR-0020850 has a benzothiazole substitution on the amide. Compound IDR-0099118 is a similar benzothiadiazole compound. Compound IDR-0040669 is a bissulfone. Compound IDR-0081053 is a thiadiazole. None of the described actives are known antibacterial agents. Other compounds with similar in vitro profiles showed diverse structural features different from compounds discussed here and are currently under evaluation separately.

Table 2

Key Hit Compoundsa

Compound activity was measured against M. tuberculosis under different conditions. Data are the average ± standard deviation for at least two independent experiments, except MAC6.5 was determined from a single experiment.

Activity of compounds against M. tuberculosis at low pH

We selected five compounds for further study based on their properties in vitro (Table ); in particular, these five confirmed activity from the screen by exhibiting a MAC6.5 less than 10 μM, showed activity against M. tuberculosis in the multistress model, and were less effective against aerobically grown, actively replicating bacteria. Four of the compounds had no activity against replicating bacilli, and one compound had activity, but was 50-fold less active (Table ). The multistress model uses several conditions relevant to the infection setting, namely, an acidic pH of 5.0, low oxygen, low nutrient, and nitric oxide, all of which induce the nonreplicating state.[41] In order to determine if pH was a determining factor for compound activity, we determined the MIC against replicating bacteria at pH 5.5 and pH 6.7. We selected pH 5.5, since it is the lowest pH at which we were able to measure reproducible growth. All five compounds showed activity at pH 5.5 (Table ). Two of the compounds had activity at pH 6.7, but to a lesser degree; three of the compounds had no activity at pH 6.7, confirming that they have pH-dependent activity against M. tuberculosis. The reference compounds RIF and INH were roughly equipotent against replicating M. tuberculosis under the 2 conditions. PZA was not active in either condition, up to 200 μM. We also determined if these compounds were cytotoxic to mammalian cells, and all of them were more potent in the assay that determined the MAC6.5 than they were in the cytotoxicity assay.

Lack of Nonspecific Activity

One potential mechanism for increased activity at low pH would be if the compounds exhibited ionophore/protonophore activity or were general membrane disrupters or pore formers, similar to monensin. We tested the five compounds in a number of assays to exclude this possibility. We tested for ionophore activity, using HEK-293 cells and calcium, as well as for disruption of membrane potential in HEK-293 cells (Table ). None of the compounds had activity, suggesting they are not calcium ionophores and their mode of action does not involve disrupting membrane potential in mammalian cells (Table ).

Table 3

Ionophore and Membrane Potential Testinga

compound	calcium ionophore EC50 (μM)	membrane potential (μM)
IDR-0020850	>100	>100
IDR-0054790	>100	>100
IDR-0099118	>100	>100
IDR-0040669	>100	>100
IDR-0081053	>100	>100
A23187	0.15 ± 0.008	>100
amphotericin	>100	0.62 ± 0.1311

HEK-293 cells were seeded in 384-well plate with compounds prepared as serial dilutions. Calcium ionophore activity was measured using calcium dye and FLIPR reader; A23187 was the positive control. Disruption of membrane potential was measured using the membrane potential red dye; amphotericin B was the positive control. EC50 is the concentration required to effect 50% of the maximal response. Results are the average ± standard deviation from 2 independent runs.

Bactericidal activity against M. tuberculosis

We demonstrated that compounds disrupted pH homeostasis and were active in vitro in short-term assays (up to 7 days) dependent on bacterial growth. Our next step was to determine if this activity translated into effective bactericidal activity over the course of 21 days. We determined the kill kinetics of the five compounds against nonreplicating M. tuberculosis at near-neutral (pH 6.8) and acidic (pH 4.5) conditions (Figure ). Surprisingly, all the compounds showed activity against bacilli in buffer pH 6.8, despite having no activity against replicating bacteria, suggesting their bactericidal activity may be specific for nonreplicating organisms. In some cases this activity was pronounced; for example, compound IDR-0040669 was rapidly bactericidal, resulting in sterilization after 3 days at a concentration as low as 10 μM (Figure B). In contrast, 3 compounds, IDR-0020850, IDR-0054790, and IDR-0099118, had a slower kill that appeared to be concentration dependent. Compound IDR-0081053 was effective at sterilizing cultures in a concentration-dependent fashion but only at the highest concentrations (7 days for 100 μM, 14 days for 50 μM, and 21 days for 25 μM).

Figure 2

Kill kinetics. Compounds were tested for their ability to kill M. tuberculosis in phosphate citrate buffer at pH 6.8 (left panels) and pH 4.5 buffer (right panels): (A, B) test compounds; (C) control compounds. Data are the average and standard deviation from two independent experiments.

All compounds were more effective at killing under low pH, except IDR-0040669 (Figure ). Compounds IDR-0020850 and IDR-0054790 were slightly better at low pH, whereas compounds IDR-0099118 and IDR-0081053 were markedly more efficient at killing M. tuberculosis under acidic conditions. Thus, we confirmed that 4 of the 5 compounds had pH-dependent bactericidal activity. In these experiments, we also saw that RIF and INH were slightly more active in the acidic conditions (Figure C). This has not been noted before, but we see this reproducibly, suggesting that mechanisms other than disrupting pH homeostasis could also lead to the identification of compounds with condition-dependent, or condition-enhanced activity. Since INH inhibits mycolic acid production and therefore cell wall integrity, and inhibition of transcription can prevent an adaptive transcriptional response, these factors may contribute to enhanced activity at low pH. PZA was inactive at the concentrations we tested in both conditions. This is consistent with previous reports, which suggest that even at low pH, the effective concentration of PZA is very high (>500 μM). To rule out any strain-dependent effects, we confirmed that compounds 1–5 were similarly active against the Erdman strain under low pH (data not shown).

Compound Stability at Low pH

We confirmed the purity of the five compounds by LCMS as >95%. We determined the stability of compounds in buffer at pH 4.5 or pH 6.8 over 5 days. Four of the compounds were stable in buffers at both pH values (50% remaining after 5 days), while compound IDR-0054790 appeared to degrade to an unknown species in the low pH buffer (<20% at pH 6.8 and no detectable parental compound at pH 4.5 at day 5).

Discussion

We performed a high throughput screen using a pH-sensing strain of M. tuberculosis to identify compounds that interrupt the pH homeostasis. From our hits, we focused on compounds that were active against nonreplicating M. tuberculosis and might therefore be able to contribute to shortening the duration of therapy. Although this screen was tested against live cells of M. tuberculosis, one major limitation was that activity from the screen might not translate into the ability to kill bacteria. We were able to find compounds with bactericidal activity, which in 4/5 cases was pH-dependent, confirming the validity of the approach. We used a cutoff for hits in our screen of pH 6.5 as previously described,[37] although it is not known how low the pHIB needs to drop, or for how long, to result in death. In Mycobacterium bovis BCG or Mycobacterium smegmatis, a reduction to pH 6 by CCCP had a lethal effect.[42] We demonstrate here that a reduction in M. tuberculosis to pH 6.5 is sufficient, since compounds had bactericidal activity.

The five compounds we selected were more active against M. tuberculosis at low pH. Four of the compounds were more active against replicating M. tuberculosis in the acidic condition compared to neutral pH. Two compounds were more active against nonreplicating M. tuberculosis in the acidic condition compared to the neutral condition. This increased activity at low pH could be partially due to M. tuberculosis being more generally vulnerable in the low pH condition. However, the increase in susceptibility was more marked for these compounds than for reference compounds. Further study regarding mechanism of action of these five compounds would shed light on why some compounds are pH-dependent under replicating conditions and others are pH-dependent under nonreplicating conditions.

In summary, we screened a set of ∼89 000 compounds, identified compounds that interrupt pH homeostasis, and confirmed 556 hits. Five compounds were selected and demonstrated pH-dependent bactericidal activity, confirming the utility of the screen. Further work to elucidate their mechanism of action and their utility as starting points for drug discovery is warranted.

Materials and Methods

Growth of M. tuberculosis

M. tuberculosis H37Rv was cultured in Middlebrook 7H9 (Difco) medium supplemented with 0.05% w/v Tween 80 and 10% v/v oleic acid, albumin, dextrose, and catalase (OADC; Becton Dickinson) (7H9-Tw-OADC). Recombinant M. tuberculosis expressing a codon-optimized pH-sensitive green fluorescent protein (GFP) under the control of a constitutively expressed promoter was constructed in H37Rv and cultured with hygromycin B (Hyg) at 50 μg/mL.[11,16]

Generation of rGFP Standard Curve

M. tuberculosis expressing a codon-optimized pH-sensitive green fluorescent protein (rGFP) was grown to late logarithmic phase (OD590 ≈ 1), washed, resuspended in phosphate buffered saline (PBS), and lysed using a Fastprep instrument (MP Biomedicals). Cell-free extracts were recovered and diluted to 10 μg/μL total protein in phosphate citrate buffers adjusted to pH 5.5–8.5 in increments of 0.5 units. Fluorescence was measured at excitation/emission 395/510 and 475/510, and the ratio was calculated. The ratio was plotted versus the pH to generate a standard curve, and a three parameter nonlinear fit line was generated.

Screen

M. tuberculosis was cultured in 7H9-Tw-OADC-Hyg to late logarithmic phase, washed, and resuspended in phosphate citrate buffer, pH 4.5 (0.0896 M Na2HPO4, 0.0552 M citric acid) plus 0.05% v/v tyloxapol to an OD of 0.6. To prepare assay plates, 24 μL of buffer was dispensed into sterile black, clear bottom 384-well plates (Greiner). Controls or compounds were added as follows: 1 μL of 250 μM monensin (Sigma-Aldrich) in column 1; 1 μL of 25 μM monensin in column 23; 1 μL of DMSO in columns; and 1 μL of 40 μM test compound in columns 3–22. An additional 25 μL of buffer was dispensed in column 24. Plates were inoculated with 25 μL of culture in columns 1–23 using a Multidrop Combi (Thermo Scientific). Plates were incubated for 2 days in a humidified incubator at 37 °C. The following QC criteria had to be satisfied for each plate and each run: % CV of maximum acidification control <20, % CV of minimum acidification control <20, % CV of background <20, and Z′ of controls >0.5.

Dose Response

Compounds were tested as 10-point, 3-fold serial dilutions, typically starting at 200 μM. Compound response curves were plotted using four-parameter logistic nonlinear regression, and the concentration at which the pH reached 6.5 was calculated and recorded as the minimum active concentration 6.5 (MAC6.5).

Ionophore and Membrane Potential Assays

HEK-293 cells (ATCC CRL-1573) were seeded at 15000 cells/well onto poly(d-lysine) coated black/clear bottom 384-well plates in 25 μL of minimum essential medium, 10% FBS, 100 IU/mL penicillin, 100 μg/mL streptomycin, and 2 mM Glutamax. Compounds were diluted in DMSO and tested as 10-point, 3-fold serially diluted samples starting at 100 μM. For the calcium ionophore assay, 25 μL of calcium 5 dye (Molecular Devices) was suspended in Hank’s Balanced Salt Solution (HBSS) with 20 mM HEPES, pH 7.4, and added to 384-well cell plates containing HEK-293 cells, then incubated for 60 min at 37 °C and 5% CO2. Compounds were then added to the dye-loaded cell plates. Using a FLIPR Tetra plate reader (Molecular Devices), fluorescent signal (excitation 470–490 nm; emission 515–575 nm) was detected at 1 s intervals for 300 s. Emission maximum results, a measure of intracellular calcium, were plotted using XLfit software (IDBS, Inc.). Calcium ionophore A23187 (Sigma) was used as a positive control. The membrane potential assay was carried out as described above for the calcium ionophore assay except membrane potential red dye R8126 (Molecular Devices) was used and detected at excitation 510 nm and emission 565–625 nm to indicate voltage depolarization changes across the cell. Amphotericin B (Sigma) was used as a positive control.

Replicating and Nonreplicating Minimum Inhibitory Concentration (MIC) Determination

Compounds were tested in nonreplicating and replicating conditions against the mc26220 strain as previously described.[40,41] Briefly, logarithmically replicating M. tuberculosis was exposed to compound in a modified Sauton’s medium at pH 5.5 in 1% O2 and 5% CO2 with 0.05% μM butyrate in the presence of 0.5 mM nitrite for 6–7 days, then diluted 5-fold and outgrown in 7H9 with glucose and glycerol at pH 6.6 in 21% O2 and 5% CO2 for 7–10 days. For replicating data, compounds were tested as described in the outgrowth step.

MIC Determination at pH 5.5 & pH 6.8

Middlebrook 7H9-OADC medium with 0.05% tyloxapol was adjusted to pH 5.5 using hydrochloric acid and filter sterilized. To allow sufficient growth at pH 5.5, the starting OD590 was 0.04 and plates were incubated for 6 days, while the starting OD590 was 0.02 and plates were incubated for 5 days for samples at pH 6.8. MICs were determined by measuring growth by OD590 and three parameter nonlinear fit.

Kill Kinetics

Late logarithmic phase H37Rv (OD590 0.6–1.0) was harvested and resuspended in phosphate citrate buffer plus 0.05% tyloxapol. Cultures were inoculated to ∼106 CFU/mL, compounds were added (final concentration 2% DMSO), and cultures were incubated standing at 37 °C. Aliquots were plated for CFU every 7 days. Plates were incubated at 37 °C for 4 weeks before counting.

Compound Stability

Compounds were diluted to 20 μM in phosphate citrate buffer with tyloxapol at pH 4.5 or pH 6.8 and incubated at 37 °C. Samples were taken and injected into an Agilent 1100 HPLC system using the following: Phenomenex Gemini C18 column, 3 mm × 50 mm; flow rate 0.45 mL/min; gradient 5–95% acetonitrile in water over 8 min with all solvents containing 0.05% formic acid. UV detection was monitored at 214 and 254 nm using an Agilent diode array detector (G1315B). Mass spectroscopy data was generated via electrospray ionization in positive mode using an Agilent LC/MSD (G1956B). The area under the UV peaks was determined using Agilent ChemStation software (B.01.03-SR2) to estimate % parent remaining.

Cytotoxicity

HepG2 human liver cells (ATCC) were seeded in DMEM, 10% FBS, 1 mM sodium pyruvate, 2 mM Glutagro (Corning), 100 IU/mL penicillin, and 100 μg/mL streptomycin containing 25 mM glucose and incubated in 37 °C, 5% CO2. Three-fold serially diluted compounds were added to the cells 24 h post cell seeding to 1% DMSO final concentration, and cells were incubated for 72 h. Next, CellTiter-Glo was added, and RLU were measured using a Synergy4 plate reader. Raw data were normalized by the average RLU value from 1% DMSO treated wells and expressed as % growth. Growth inhibition curves were fitted using the Levenberg–Marquardt algorithm, and the concentration that produced 50% of the growth inhibitory response was reported as the IC50.