In this method paper, we present a high-throughput screening strategy to identify chemical compounds, such as osmolytes, that have a significant impact on bacterial persistence.
Bacterial persisters are defined as a small subpopulation of phenotypic variants with the capability of tolerating high concentrations of antibiotics. They are an important health concern as they have been associated with recurrent chronic infections. Although stochastic and deterministic dynamics of stress-related mechanisms are known to play a significant role in persistence, mechanisms underlying the phenotypic switch to/from the persistence state are not completely understood. While persistence factors triggered by environmental signals (e.g., depletion of carbon, nitrogen and oxygen sources) have been extensively studied, the impacts of osmolytes on persistence are yet to be determined. Using microarrays (i.e., 96 well plates containing various chemicals), we have designed an approach to elucidate the effects of various osmolytes on Escherichia coli persistence in a high throughput manner. This approach is transformative as it can be readily adapted for other screening arrays, such as drug panels and gene knockout libraries.
Bacterial cultures contain a small subpopulation of persister cells that are temporarily tolerant to unusually high levels of antibiotics. Persister cells are genetically identical to their antibiotic-sensitive kins, and their survival has been attributed to transient growth inhibition1. Persister cells were first discovered by Gladys Hobby2 but the term was first used by Joseph Bigger when he identified them in penicillin-treated Staphylococcus pyogenes cultures3. A seminal study published by Balaban et al.4 discovered two persister types: type I variants that are primarily formed by passage through the stationary phase, and type II variants that are continuously generated during the exponential growth. Persisters are detected by clonogenic survival assays, in which culture samples are taken at various intervals during antibiotic treatments, washed, and plated on a typical growth medium to count the surviving cells that can colonize in the absence of antibiotics. The existence of persisters in a cell culture is assessed by a biphasic kill curve4,5 where the initial exponential decay indicates the death of antibiotic-sensitive cells. However, the killing trend decreases over time, eventually leading to a plateau region which represents the surviving persister cells.
Persister cells have been associated with various diseases such as tuberculosis6, cystic fibrosis7, candidiasis8 and urinary tract infections9. Almost all microorganisms tested so far were found to generate persister phenotypes, including highly pathogenic Mycobacterium tuberculosis6, Staphylococcus aureus10, Pseudomonas aeruginosa7 and Candida albicans8. Recent studies also provide evidence of the rise of multidrug-resistant mutants from persister subpopulations11,12. Substantial efforts in this field have revealed that persistence mechanisms are highly complex and diverse; both stochastic and deterministic factors associated with the SOS response13,14, reactive oxygen species (ROS)15, toxin/antitoxin (TA) systems16, autophagy or self-digestion17 and ppGpp-related stringent response18 are known to facilitate persister formation.
Despite significant progress in understanding the persistence phenotype, the effects of osmolytes on bacterial persistence have not been fully understood. Since the maintenance of optimal osmotic pressure is a necessity for cells’ growth, proper functioning and survival, an in-depth study of osmolytes could lead to potential targets for anti-persister strategies. Although laborious, high-throughput screening is a very effective approach for identifying metabolites and other chemicals that play a crucial role in the persistence phenotype19,20. In this work, we will discuss our published method19, where we have used microarrays, i.e., 96 well plates containing various osmolytes (e.g., sodium chloride, urea, sodium nitrite, sodium nitrate, potassium chloride), to identify osmolytes that significantly influence E. coli persistence.
1. Preparation of growth medium, ofloxacin solution and E. coli cell stocks
2. Propagation of cells to eliminate pre-existing persisters
3. Validating the elimination of pre-existing persister cells
4. Microarray plate screenings
5. Validating the identified conditions
Figure 1 describes our experimental protocol. The dilution/growth cycle experiments (see Protocol 2) were adapted from a study conducted by Keren et al.5 to eliminate the persisters originating from the overnight cultures. Figure 2A is a representative image of agar plates used to determine CFU levels of cell cultures before and after OFX treatment. In these experiments, cells were cultured in modified LB medium with osmolytes in half-area 96 well plates as described in step 4.2. After incubating the plate in an orbital shaker for 24 h, the persister assay was performed using a generic flat bottom 96 well plate (see step 4.3). The osmolytes and the concentration being tested here were chosen based on our previous study19, where we performed steps 4.1 and 4.3 using the PM-9 plate that includes various osmolytes at different concentrations (sodium chloride, potassium chloride, sodium sulfate, ethylene glycol, sodium formate, urea, sodium lactate, sodium phosphate, sodium benzoate, ammonium sulfate, sodium nitrite and sodium nitrate). The first column in Figure 2A shows the CFU counts of the control group. The second column represents a condition where cells were cultured in 100 mM sodium nitrate; this condition was previously found to slightly increase the persister levels19. The third column represents a condition where cells were cultured in 60 mM sodium nitrite, and this condition was previously found to significantly decrease the persister levels compared to controls19. Figure 2B is a graphical representation of the CFU data obtained from the agar plates. Figure 2C shows the persister fractions of the cell cultures tested in 96 well plates. To calculate the fractions, persister counts were normalized to the cell counts obtained before the antibiotic treatments. Figure 3 shows the biphasic kill curves and the persister fractions, respectively, for the assay cultures performed in baffled flasks. In these experiments, cells were first cultured in 25 mL of modified LB medium with the indicated osmolytes in 250 mL baffled flasks for 24 h, and then the cells were transferred to persister-assay flasks for persister enumeration as described in Protocol 5.
Figure 1: The experimental procedure. Cells from a frozen cell stock were grown overnight (12 h) in fresh LB medium. At 12 h, the overnight culture was diluted (1:100) in 25 mL of modified LB medium and grown until OD600=0.5. This propagation step was repeated twice. After the final propagation step, the exponential phase cells (at OD600=0.5) were diluted (1:100) in fresh modified LB medium in a 250 mL baffled flask and a 50 mL centrifuge tube, respectively. The cell suspension in the 250 mL baffled flask was treated with 5 μg/mL of OFX to quantify persisters. The cell suspension in the 50 mL centrifuge tube was transferred to microarrays and incubated for 24 h in an orbital shaker at 250 rpm and 37 °C. The cells from the microarrays were then transferred to persister-assay plates to quantify persisters. This figure was created using biorender.com. This figure has been modified from our previous publication19. Please click here to view a larger version of this figure.
Figure 2: Microarray experiments. (A) Cells were first grown in modified LB medium with or without indicated osmolytes in a half-area 96 well plate for 24 h. The cells were then transferred to a generic flat-bottom 96 well plate and treated with 5 μg/mL of OFX for 6 h. Before and after 6 h treatment, cells were serially diluted in PBS, plated on agar plates and incubated at 37 °C for 16 h. Each condition has 8 technical replicates. (B) CFU measurements were performed to assess the effects of osmolytes on cell viability and persistence, respectively. The straight line indicates the limit of detection (600 CFUs). (C) The graph represents the persister fractions of the cell cultures, calculated by taking the ratio of the CFU counts after and before OFX treatment. The persister fraction of the cell culture that has 60 mM sodium nitrite was not calculated as its persister level is below the limit of detection. Each data point was denoted by mean value ± standard deviation, calculated from 8 technical replicates. Please click here to view a larger version of this figure.
Figure 3: Validation experiments. (A) Exponential-phase cells were transferred to 25 mL of fresh modified LB medium with indicated osmolytes in 250 mL baffled flasks and cultured for 24 h. Then, the cells were diluted (1:100) in 250 mL baffled flasks containing 25 mL of modified LB medium and treated with 5 μg/mL OFX for 6 h. The CFU counts in the assay cultures were monitored hourly to generate biphasic kill curves. * indicates the condition that significantly affects the OFX persister levels compared to no-osmolyte controls (two-tailed unequal variance t-test, p<0.05). (B) The graph represents the persister fractions of the cultures. Each data point was denoted by mean value ± standard deviation, calculated from 3 biological replicates. This figure has been modified from our previous publication19. Please click here to view a larger version of this figure.
The high throughput persister assay described here was developed to elucidate the effects of various chemicals on E. coli persistence. In addition to commercial PM plates, microarrays can be constructed manually as described in step 4.2. Moreover, the protocol presented here is flexible and can be used to screen other microarrays, such as drug panels and cell libraries, that are in 96 well plate formats. The experimental conditions including the growth phase, inoculation rate and medium can be adjusted to test these libraries. For instance, if one wants to screen a cell library, such as Keio E. coli knockout collection23, they can first transfer the strains from the library to 96 well plates that include fresh media, using a multichannel pipette. Once the cells reach the desired growth phase (e.g., exponential or stationary phase), the cells are then transferred to persister-assay plates as described in step 4.3 to enumerate the persister levels of the knockout strains. Similarly, this strategy can be used to screen the E. coli Promoter collection24 (a library of fluorescent reporter strains in 96 well plate formats) to identify promoters that are activated by antibiotics. These promoters can be readily detected by measuring the fluorescence signals of strains in persister-assay plates during the antibiotic treatment.
In our experiments, we used a modified LB (lacking NaCl) to avoid any additional effect that could arise from NaCl, considering that the microarrays already have various osmolytes. Although NaCl in regular LB medium is known to be good for preserving the membrane integrity of cells, it has been reported that NaCl at 0-1% range has a minimal effect on cell growth25. Moreover, results from propidium iodide (PI) staining and persisters assays in our previous study have shown that the absence of NaCl does not significantly affect the membrane integrity and the persistence of E. coli cells19. This modification was made specifically to address the concerns in our study and can be changed based on the nature of the research.
We have also adapted a method developed by Keren et al.5 to eliminate pre-existing persisters in our cell cultures prior to inoculation into the microarrays. Type I persisters are known to be generated during the stationary phase; therefore, the direct inoculation of the cells from overnight cultures into microarray plates would transfer a significant number of persisters, which may hinder the effects of osmolytes. With the dilution/growth cycle experiments (see Protocol 2), we were able to significantly reduce these preexisting persisters arising from the overnight cultures19. We note that we have cultured the cells in microarray plates for 24 h to be able to count all persisters, including type I variants that are formed during the stationary phase in the presence of osmolytes. However, these adapted techniques can always be modified depending on the nature of the study.
We treated the cells in microarrays with 5 µg/mL OFX to enumerate the persisters. Since the washing procedure to remove the antibiotic from 96 samples would be very labor-intensive, cells after the treatment were serially diluted in PBS without washing (see step 4.2). This procedure diluted the OFX concentration more than 30-fold. The diluted cell suspensions were then plated on antibiotic-free agar plates where the OFX was further dispersed. Our preliminary studies where we repeated these experiments with and without washing have verified that this serial-dilution method did not affect the persister levels19.
During the multi-channel pipetting in 96 well plates, one should be carefully mixing the cell suspensions to maintain a homogenous cell distribution. To do this, depending on the working volume, it would be beneficial to note the minimum number of pipetting that is needed to make the solution homogeneous. For this purpose, we conducted a simple control experiment where we monitored the pipetting (up and down) by dispersing a dye molecule in PBS and medium under the conditions studied here. Additionally, if working with more complex environments such as pH buffers, we would suggest measuring the pH of the cell culture before and during the incubation in a shaker as the cell culture tends to be very alkaline (pH ≈ 8) over time. This can give an idea of the quality as well as the pH range of the buffer used. Additionally, in order to obtain countable CFUs on agar plates from the high-throughput assays, certain conditions, such as the cell inoculation rate, age of cultures, and serial-dilution parameters (PBS volume, dilution rates and the number of dilution steps, etc.) should be optimized before testing the microarrays. Finally, the results obtained from the microarrays should be further validated in flasks as the volume of the culture, surface area and aeration in a 96 well plate could have additional effects on the observed results.
The authors have nothing to disclose.
We would like to thank the members of Orman Lab for their valuable inputs during this study. This study was funded by the NIH/NIAID K22AI125468 career transition award and a University of Houston startup grant.
14-ml test tube | Fisher Scientific | 14-959-1B | |
E. coli strain MG1655 | Princeton University | Obtained from Brynildsen lab | |
Flat-bottom 96-well plate | USA Scientific | 5665-5161 | |
Gas permeable sealing membrane | VWR | 102097-058 | Sterilized by gamma irradiation and free of cytotoxins |
Half-area flat-bottom 96-well plate | VWR | 82050-062 | |
LB agar | Fisher Scientific | BP1425-2 | Molecular genetics grade |
Ofloxacin salt | VWR | 103466-232 | HPLC ≥97.5 |
Phenotype microarray (PM-9 and PM-10) | Biolog | N/A | PM-9 and PM-10 plates contained various osmolytes and buffers respectively |
Round-bottom 96-well plate | USA Scientific | 5665-0161 | |
Sodium chloride | Fisher Scientific | S271-500 | Certified ACS grade |
Sodium nitrate | Fisher Scientific | AC424345000 | ACS reagent grade |
Sodium nitrite | Fisher Scientific | AAA186680B | 98% purity |
Square petri dish | Fisher Scientific | FB0875711A | |
Tryptone | Fisher Scientific | BP1421-500 | Molecular genetics grade |
Varioskan lux multi mode microplate reader | Thermo Fisher Scientific | VLBL00D0 | Used for optical density measurement at 600 nm |
Yeast extract | Fisher Scientific | BP1422-100 | Molecular genetics grade |