Marie H.
Foss
a,
Ye-Jin
Eun
a,
Charles I.
Grove
b,
Daniel A.
Pauw
c,
Nohemy A.
Sorto
b,
Jarred W.
Rensvold
a,
David J.
Pagliarini
a,
Jared T.
Shaw
b and
Douglas B.
Weibel
*ad
aDepartments of Biochemistry and Biomedical Engineering, 433 Babcock Drive, Madison, WI 53706, USA. E-mail: weibel@biochem.wisc.edu; Tel: +1 (608) 890-1342
bDepartment of Chemistry, University of California-Davis, Davis, CA 95616, USA
cDepartment of Cell and Molecular Biology, University of Wisconsin–Madison, Madison, WI 53706, USA
dDepartment of Biomedical Engineering, University of Wisconsin–Madison, Madison, WI 53706, USA
First published on 18th July 2012
FtsZ is a homolog of eukaryotic tubulin that is widely conserved among bacteria and coordinates the assembly of the cell division machinery. FtsZ plays a central role in cell replication and is a target of interest for antibiotic development. Several FtsZ inhibitors have been reported. We characterized the mechanism of these compounds in bacteria and found that many of them disrupt the localization of membrane-associated proteins, including FtsZ, by reducing the transmembrane potential or perturbing membrane permeability. We tested whether the reported phenotypes of a broad collection of FtsZ inhibitors disrupt the transmembrane potential in Bacillus subtilis strain 168. Using a combination of flow cytometry and microscopy, we found that zantrin Z1, COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcinnamaldehyde, COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundtotarol, sanguinarine, and viriditoxin decreased the B. subtilis transmembrane potential or perturbed membrane permeability, and influenced the localization of the membrane-associated, division protein MinD. These studies demonstrate that small molecules that disrupt membrane function in bacterial cells produce phenotypes that are similar to the inhibition of proteins associated with membranes in vivo, including bacterial cytoskeleton homologs, such as FtsZ. The results provide a new dimension for consideration in the design and testing of inhibitors of bacterial targets that are membrane-associated and provide additional insight into the structural characteristics of antibiotics that disrupt the membrane.
The loss of the transmembrane potential (ΔΨ) was recently reported as a negative regulator of FtsZ function.10 ΔΨ arises from the separation of different concentrations of charged ions across the bacterial membrane and can be measured using a permeable cationic dye and the Nernst equation (eqn (1)), where Ci is the concentration of the dye inside the cell and Co is the concentration of dye outside, R is the gas constant, T is temperature, Z is the number of electrons per mole of dye, and F is the Faraday constant.11
![]() | (1) |
ΔΨ is a source of potential energy that facilitates the transport of molecules and ions across the cell membrane. The establishment of the proton motive force (pmf) contributes significantly to ΔΨ and facilitates ATP production via oxidative phosphorylation.12 The relationship between ΔΨ and the chemical proton potential can be described by eqn (2), where ΔpH is the difference between the internal and external pH (pHin − pHout).12
![]() | (2) |
A recently discovered role of ΔΨ is the localization of membrane-associated proteins, including bacterial proteins that regulate division and cell shape.10 COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundCarbonyl cyanide m-chlorophenylhydrazone (CCCP, 1, Table 1) is among the best-studied small molecules that shuttle protons across the membrane (e.g. ‘protonophores’) and affects the localization of several membrane-associated proteins, including the division protein FtsA;10Fig. 1 illustrates an effect of ΔΨ on the localization of MinD. Small molecules that perturb ΔΨ may mislead the assignment of their target in vivo. Despite evidence that small molecules can influence the localization of division proteins via perturbing bacterial membranes, the effect of the relatively large number of putative inhibitors of this class of proteins on ΔΨ has not yet been studied.
Inhibitor structure | MIC |
---|---|
![]() |
10 μM |
![]() |
0.16 μM |
![]() |
2.5 mM |
![]() |
2.5 μM |
![]() |
10 μM |
![]() |
0.63 μM |
![]() |
2.5 μM |
![]() |
20 μM |
![]() | ||
Fig. 1 The loss of ΔΨ disrupts the normal localization pattern of membrane-associated proteins. The cartoon depicts the diffuse pattern of a polarly localized protein after reduction of ΔΨ. The length of the line across the membrane depicts the relative magnitude of ΔΨ. Fluorescence images below the cartoon represent MinD localization in E. coli cells with COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundDMSO treatment on the left and 1 treatment on right. The scale bar represents 2 μm. |
FtsZ is among the most widely studied bacterial cytoskeletal proteins and is a central component of the cell division machinery. The essential role of FtsZ makes it a target of interest for the development of antibiotics. Consequently, several small molecule inhibitors of FtsZ have been reported and their target identification draws upon their activity in vitro and in vivo.13–21 The in vivo activity of several of these FtsZ inhibitors is centered upon two distinct observations: (1) cell filamentation; and (2) mislocalization of FtsZ. However, these experiments do not assess whether the inhibitor influences the properties of membranes, which provides the mechanism for mislocalizing membrane-associated proteins, such as FtsZ. An example is the proposed inhibitor of FtsZ, zantrin Z1, which we refer to as ‘3Z1’ or 2.13 Although 2 delocalizes FtsZ and has a minimum inhibitory concentration (MIC) of 0.08–40 μM against a range of bacterial strains,22 treatment of Escherichia coli with this small molecule (5 μM, or 1× MIC) does not filament cells, which is the canonical phenotype of FtsZ inhibition. This result suggests that the activity of this compound is not due to inhibiting FtsZ per se. Many putative FtsZ inhibitors produce phenotypes that are unrelated to FtsZ inhibition and their physicochemical properties make them excellent candidates for interacting with phospholipid bilayers.
In reviewing proposed FtsZ inhibitors, we identified several compounds that may influence the properties of bacterial membranes. COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundCinnamaldehyde (3) is an antimicrobial agent that has been reported to inhibit FtsZ.14 The chemically related compound COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcurcumin affects the activities of multiple disparate membrane proteins by changing lipid bilayer properties.23 For example, COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcurcumin thins bilayers and decreases their stiffness.23,24 COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundTotarol (4) has been assigned several other functions in vivo, including: (1) inhibiting multidrug efflux pumps;25 (2) inhibiting the bacterial electron transport chain;26 and (3) disrupting the physical properties of the membrane.27 Totarol has a large phospholipid–water partition coefficient (Kp = 1.8 × 104) and is sequestered in membranes.28 Sanguinarine (5) inhibits the activity of guinea pig cardiac Na+, K+-ATPase29 and causes mitochondrial depolarization in mouse melanoma cells.30 Viriditoxin (6) activates ATP hydrolysis and induces calcium sensitized swelling of rat liver mitochondria.31 A common feature of these compounds is their interaction with phospholipid membranes. Although these compounds may bind FtsZ in vitro, the mislocalization of membrane-associated proteins in vivo may be due to changes in the physicochemical properties of membranes that arise in response to the small molecules.
Several of these compounds share physicochemical characteristics with reported uncouplers of ΔpH and ΔΨ; the compounds are lipophilic weak acids that contain electron-withdrawing groups.32,33 We sought to test the hypothesis that their effect on FtsZ and other membrane-associated proteins in vivo arises from disrupting ΔΨ. We tested a panel of reported FtsZ inhibitors (Table 1) for their ability to deplete ΔΨ in B. subtilis 168 and determined whether the effect delocalized the cytoplasmic membrane-associated protein, MinD. In this paper we demonstrate that many of the small molecules identified as hits from various high-throughput screens against the bacterial protein FtsZ, depolarize the bacterial membrane and delocalize FtsZ non-specifically.
For our studies, we used MinD translationally fused to green fluorescent protein (GFP) as a model membrane-associated protein for several reasons. (1) MinD plays a role in bacterial cell division and regulation of FtsZ activity in vivo.36,37 (2) MinD contains a well-characterized terminal amphipathic helix that associates with the cytoplasmic membrane.38 (3) MinD binds preferentially to anionic phospholipids, and the local organization of these lipids is thought to influence the position of the protein in vivo.39,40 In principle, disrupting ΔΨ should cause MinD to mislocalize in vivo. To test this hypothesis, we used Bacillus subtilis as a model Gram-positive bacterium. We chose B. subtilis because all the compounds we tested are reported to target FtsZ in Gram-positive cells; very few are active against Gram-negative bacteria. We tested compounds with broad-spectrum activity against B. subtilis 168 and E. coli MG1655.
The compounds used in this study included: COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcarbonyl cyanide m-chlorophenylhydrazone (CCCP, 1), 3Z1 (2), COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcinnamaldehyde (3), COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundtotarol (4), sanguinarine (5), viriditoxin (6), and PC190723 (7) (Table 1). We included COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundcefuroxime (8) as a negative control as it is a therapeutic antibiotic that inhibits cell wall assembly in growing bacteria and should not perturb ΔΨ on the time scale of our experiments (e.g. 30 min) (Table 1). β-lactam antibiotics can depolarize bacterial membranes, but the response can take several hours depending on the antibiotic and the dosage.41
In a typical experiment, we measured the fluorescence intensity of ∼10000 cells by flow cytometry. The result of compound treatment on ΔΨ is shown in Fig. 2. We used 1 as a positive control for these experiments: treating B. subtilis 168 cells with 1 at 1× MIC for 5 min produced a ratio of λ575/λ530 of 0.045, which indicated a large decrease in ΔΨ. As a negative control, we treated B. subtilis cells with 8 at 1× MIC for 20 min and observed no decrease in ΔΨ: λ575/λ530 was 0.306 ± 0.011 for 8 compared to 0.257 ± 0.003 for an untreated control. We found that the treatment of B. subtilis cells with compounds 2–5 at 1× MIC decreased λ575/λ530 indicating a significant reduction in ΔΨ. FtsZ inhibitors 6 and 7 did not perturb λ575/λ530 in our assay and thus have no effect on ΔΨ within the time scale of our experiments. We also performed fluorescence controls with 1–8 only (i.e., no cells) to confirm there was no competitive fluorescence emitted by 1–8 (Fig. S1A and B, ESI†). As the emission of 5 spans a region of the spectrum including λ575, we performed further controls to determine the magnitude of this signal compared to the fluorescence emission from DiOC2. We determined that the fluorescence of 5 was not significant compared to the DiOC2 signal emitted from labelled cells under the conditions used (Fig. S2, ESI†).
![]() | ||
Fig. 2 Measurement of the perturbation of ΔΨ of B. subtilis 168 cells by compounds 1–8. We determined the relative magnitude of ΔΨ of B. subtilis 168 cells by measuring the λ575/λ530 fluorescence emission intensity ratio of DiOC2 after excitation at λ488. A large value of λ575/λ530 represents a high ΔΨ. Labels ‘C’, ‘D’, and ‘M’ stand for control (untreated), COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundDMSO, and COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundmethanol, respectively. Error bars represent two standard deviations from the mean (n = 3). |
We pretreated B. subtilis cells with compounds 1–8 at 1× MIC for 5 min, dosed cells with PI (100 μM), incubated for 30 min, and measured the fluorescence intensity at λemission = 620 nm. Fig. 3 summarizes the resulting fluorescence intensity of cells. Using the Kruskal–Wallis test, we found no significant difference between the solvent controls, 7, and 8. Pairwise comparison of solvent controls with 1–6 gave p-values of <0.001. We found that treating cells with 1 or 2 reduced their labelling with PI (compared to solvent control samples), suggesting that these compounds do not perturb membrane permeability. Instead, 1 and 2 decreased ΔΨ and eliminated the transport of the cationic fluorophore PI by the pmf. Treating B. subtilis 168 cells with 3–6 increased their permeability to PI. For 3–5, we observed a decrease in ΔΨ, suggesting that these compounds change global membrane properties or perturb proteins that form pores in the membrane.
![]() | ||
Fig. 3 Effect of 1–8 on the integrity of B. subtilis 168 cell membranes. We measured the permeability of the cell membrane to PI by labeling B. subtilis 168 cells after treatment with compounds. We measured PI fluorescence intensity at λ620. Large values of fluorescence intensity represent an increasing membrane permeability to PI. Labels ‘E’, ‘D’, and ‘M’ stand for COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundethanol, COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundDMSO, and COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundmethanol, respectively. Whisker plots display the median (center of the box), 25 to 75% of the population in the box, and 5 to 95% of the population between the outer whiskers. |
As a positive control, we treated B. subtilis cells with 70% COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundethanol, rinsed cells with 1× PBS to remove COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundethanol, labelled cells with PI, and measured their fluorescence. The membrane permeability of these cells increased and they were extensively labelled with PI. As a control, we measured the fluorescence properties of compounds (i.e. no cells; Fig. S3, ESI†). As the emission of 5 spans a region of the spectrum including λ620, we performed additional controls to determine if the signal would perturb the results of PI labelling. Unfortunately, the fluorescence of 5 contributed significantly in the cellular PI assay, which may account for the increase in fluorescence intensity observed (Fig. S4, ESI†).
![]() | ||
Fig. 4 Measurement of the uncoupling activity of 1 and 2 in C2C12 myoblasts. We measured the OCR of C2C12 myoblasts before and after injection of 1, 2, or COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundDMSO (the solvent control). The addition of protonophore 1 increased the OCR of C2C12 myoblasts. The addition of 2 also resulted in a rise in OCR, consistent with a depletion of the pmf. The maximum OCR was achieved at 0.6 μM of 1 and 30 μM of 2. Error bars represent two standard deviations of the mean (n = 8). |
Although the myoblast experiments in C2C12 cells did not inform us of the mechanism by which 1, 2, and 4 reduce ΔpH, which may be due to influencing ion transport, membrane permeability, or via other mechanisms, the data demonstrates that compounds 2 and 4 are less effective at reducing ΔΨ in eukaryotes than in prokaryotes. 2 required significantly higher concentrations—compared to its MIC against B. subtilis—to elicit a response from C2C12 cells. An analog of 2—COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compound2,2′-methylenebis(4-chlorophenol), also referred to as COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compounddichlorophen—is an anthelmintic agent and has been used to treat fungal infections. COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundDichlorophen is tolerated in rats and has a 50% lethal dose (LD50, in mg kg−1 of body weight) of 1506 (95% confidence interval [CI] 1310–1760) for males and 1683 (95% CI 1402–1986) for females.43 Evaluation of 2–6 in animal toxicity models would be an important step to assess their therapeutic potential.
We tested 2–6 and found their effect on cells to be similar to the solvent controls (Fig. S7A and B, ESI†): namely, treatment of B. subtilis cells with 2–6 had no significant effect on ΔpH. Only cells treated with 1 at its MIC value dissipated the artificial ΔpH. The results suggest that 2–6 are not catalytic protonophores in contrast to 1, which transports protons across the membrane. Although the artificial ΔpH was not equilibrated by 2–6, it remains possible that the ΔpH and ΔΨ of these cells decreased and was not detectable in our experiments. Another limitation may be the significant difference in the amount of membrane present in the ΔpH experiments (∼3 × 1010 cells per mL) compared to MIC or DiOC2 and PI labelling experiments (∼5 × 105 or ∼1 × 108 cells per mL). The activity of 2–6 may require a threshold lipid-to-inhibitor ratio before the effect is observed. The addition of 2 to B. subtilis cells at a concentration that increased the OCR in C2C12 myoblasts (30 μM, ∼200 times greater than its MIC) equilibrated the artificial ΔpH (Fig. S6B, ESI†). Our observation that a high concentration of 2 is required to observe the relaxation of the ΔpH may suggest a relationship between the concentration of lipids and the activity of this compound in vivo.
![]() | ||
Fig. 5 Localization of MinD in B. subtilis cells treated with 1–8. Grey bars and white bars represent cells with MinD localization and without MinD localization, respectively. The addition of 1–6 reduced the localization of MinD. Localization of MinD to the poles and the midcell of dividing cells remained normal in the DMSO solvent control. P-values from a Fisher's exact test comparing COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundDMSO with treatments are represented as *** for p < 0.001 and ns as not significant. The p-values for 7 and 8 are 0.3778 and 0.1644, respectively. The number of cells analyzed in each treatment is listed as ‘n’. |
![]() | ||
Fig. 6 Distribution of MinD in E. coli cells following treatment with 1 or 2. We normalized the cell length to average the distribution of fluorescence. The cells were organized with the brightest half of the cell oriented on the left. The normalized cell length is shown in 100 divisions, with the first and last 20% of cell length labeled in blue. The remainder of the cell is divided in half by dark and light grey segments. The number of cells assayed for each treatment is as follows: COMPOUND LINKS Read more about this on ChemSpider Download mol file of compoundDMSO (n = 123), 1 (n = 111), and 2 (n = 147). The addition of 1 resulted in diffuse MinD localization throughout the cell, as opposed to the time-averaged polar localization observed in the DMSO solvent control. Addition of 2 also resulted in a significant reduction in the dynamic localization of MinD to the poles; the fluorescence of the resulting cells was disperse. |
We performed MIC measurements using the macro-dilution technique according to the NCCLS guidelines.48 To create a two-fold dilution series for the macrodilution technique, we added each compound to the first culture tube (4 mL total volume) at the highest concentration. We diluted 2 mL of this culture into an equal volume of inoculated media (a two-fold dilution). The final volume for each culture was 2 mL. We prepared solvent controls and sterility controls using the same concentration of solvent as the tubes containing the highest concentration of antibiotic. We determined the macrodilution MIC endpoints in triplicate by identifying the lowest concentration of compound that completely inhibited growth by visual inspection.
We grew E. coli MG1655 pFX9 (Plac-gfp-minD-minE) to early exponential phase (λ = 600, 0.3–0.4) in LB with incubation at 30 °C and 200 rpm shaking. We induced GFP-MinD and MinE production by adding IPTG to the media to a final concentration of 50 μM and incubating for 75 min. We treated induced cells with a 1× MIC concentration of 1, 2, or COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundDMSO solvent for 20 min before imaging. We pipetted suspensions of treated cells on 1% (w/v) agarose pads infused with 1× PBS containing 1 or 2 at their 1× MIC value (or COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundDMSO). We imaged cells from at least three separate induction experiments for 1, 2, and COMPOUND LINKS
Read more about this on ChemSpider
Download mol file of compoundDMSO. After transferring the cells to agarose pads, we imaged cells within 10 min.
Footnote |
† Electronic supplementary information (ESI) available: Supplemental methods and figures can be found in the supplemental information. See DOI: 10.1039/c2md20127e |
This journal is © The Royal Society of Chemistry 2013 |