Microfluidic Giant Polymer Vesicles Equipped with Biopores for High-Throughput Screening of Bacteria

Abstract

Understanding the mechanisms of antibiotic resistance is critical for the development of new therapeutics. Traditional methods for testing bacteria are often limited in their efficiency and reusability. Single bacterial cells can be studied at high throughput using double emulsions, although the lack of control over the oil shell permeability and limited access to the droplet interior present serious drawbacks. Here, a straightforward strategy for studying bacteria-encapsulating double emulsion-templated giant unilamellar vesicles (GUVs) is introduced. This microfluidic approach serves to simultaneously load bacteria inside synthetic GUVs and to permeabilize their membrane with the pore-forming peptide melittin. This enables antibiotic delivery or the influx of fresh medium into the GUV lumen for highly parallel cultivation and antimicrobial efficacy testing. Polymer-based GUVs proved to be efficient culture and analysis microvessels, as microfluidics allow easy selection and encapsulation of bacteria and rapid modification of culture conditions for antibiotic development. Further, a method for in situ profiling of biofilms within GUVs for high-throughput screening is demonstrated. Conceivably, synthetic GUVs equipped with biopores can serve as a foundation for the high-throughput screening of bacterial colony interactions during biofilm formation and for investigating the effect of antibiotics on biofilms.

Keywords: GUVs; antibiotics; bacteria; high-throughput screening; polymers; vesicles.

Figure 1

Workflow of bacteria encapsulation in double emulsion templated GUVs; bacteria are grown in a medium and injected into the Si‐glass microfluidic chip via the inner aqueous phase, yielding double emulsions that mature into GUVs by solvent evaporation. Pores mediate the supply of encapsulated bacteria with fresh buffer and nutrients for growth and enable entry of test compounds targeting growth. GUVs can be analyzed using high‐throughput methods, such as flow cytometry. Matrix‐assisted laser desorption/ionization time‐of‐flight mass spectroscopy (MALDI‐ToF MS) allows for the analysis of the inner components of the GUVs.

Figure 2

a) Schematic representation of the production of double emulsions using a microfluidic six‐way junction. Scale bar corresponds to 50 µm. b) Transmission micrograph of a single vesicle (arrow) dewetting from a double emulsion to a GUV over 20 min. c) Fluorescent micrograph of GUVs with the PDMS‐b‐PMOXA membrane stained with BODIPY 630/650. Inset represents a size histogram and regressed normal distribution (red line) of 187 measured GUVs with a diameter of 31.3 ± 0.5 µm. Scale bars, 30 µm.

Figure 3

a) Fluorescence micrograph of GFP‐expressing E. coli (green) inside GUVs stained with BODIPY 630/650. b) GUV occupancy and fraction of GUVs with aggregates depending on the concentration of the bacterial suspension added to the double emulsion microfluidics, as determined by fluorescence microscopy. Data represented as mean ± SD. c) Determination of bacteria per GUV based on fluorescence microscopy image analysis (n = 30–60 GUVs per condition). One‐way ANOVA was used for comparison: p > 0.05 (n.s.), p < 0.05 (*), p < 0.005 (**), p < 0.0005 (***), and n.s. = not significant, Tukey's post hoc test. d+e) Representative fluorescence micrographs of GFP‐expressing E. coli (green) encapsulated within PDMS‐b‐PMOXA GUVs stained with BODIPY 630/650 (red) at input OD₆₀₀ of d) 0.1 (≈ 8.0 × 10⁷ cells mL⁻¹) and e) 1.5 (≈ 1.2 × 10⁹ cells mL⁻¹). f) Flow cytometric analysis of GUVs with encapsulated GFP‐expressing E. coli depending on input bacterial concentration. Scale bars, 30 µm.

Figure 4

Growth behavior of wildtype E. coli and B. subtilis in different media. Bacteria were encapsulated at a starting OD₆₀₀ = 0.4. Representative CLSM images of a‐f) SYTO 9‐stained E. coli (green) in GUVs (blue) in LB (a+d), TB (b+e) and MSgg (c+f) medium before a–c) and after d—f) incubation at 37 °C for 36 h. g–l) SYTO 9‐stained B. subtilis (green) in LB (g+j), TB (h+k), and MSgg (i+l) medium before g–i) and after j–l) incubation at 37 °C for 36 h. Scale bars, 30 µm.

Figure 5

Flow cytometric analysis of GFP‐expressing E. coli encapsulated at OD₆₀₀ = 0.4 in GUVs with and without melittin permeabilization (n = 3000–6000 GUVs) and incubated overnight at 37 °C. Violin plots of a) GFP fluorescence and b) side scattering (SSC) evaluated per GUV. Two‐way ANOVA was used for comparison: p > 0.05 (n.s.), p < 0.05 (*), p < 0.005 (**), p < 0.0005 (***), and n.s. = not significant, Tukey's post hoc test. c) Schematic of the bacteria‐encapsulating vesicles. Bacteria are encapsulated within polymeric GUVs and through pores, analytes, and fresh medium can diffuse into the lumen of the GUV. Flow cytometry analysis of B. subtilis growth inside GUVs after 72 h d) without melittin pores and e) with 10 µм melittin added to the outer aqueous phase (n = 5300–8000 GUVs). f) Histogram of SYTO 9 fluorescence from flow cytometry of B. subtilis GUVs permeabilized with 10 or 50 µм melittin exposed to 50 µg mL⁻¹ kanamycin from the outside after production and GUVs containing B. subtilis with 50 µg mL⁻¹ kanamycin inside (n = 5300–8000 GUVs). Significance levels: p > 0.05 (n.s.), p < 0.05 (*), p < 0.005 (**), and p < 0.0005 (***).

Figure 6

a) Schematic representation of bacterial biofilm colony growth within GUVs. B. subtilis are encapsulated in PDMS‐b‐PMOXA GUVs and grown for an extended duration to allow for matrix and biofilm formation. Sessile bacteria deposit matrix at the inner face of the vesicle membrane, leading to an increased antibiotic resistance. B. subtilis‐encapsulating GUVs b) before and c) after incubation in MSgg for 72 h at 30 °C. The nucleic acid stain SYTO 9 (green) and FilmTracer SYPRO Ruby Biofilm Matrix Stain (red) were used to stain the bacteria and biofilm matrix, respectively, and the GUV surface was functionalized with DBCO‐Cy5 (blue) that was linked to an azide end group on the polymer membrane. Scale bars, 10 µm. d) Violin plots showing the ratio of SYPRO Ruby biofilm matrix stain to SYTO 9 nucleic acid over time measured by flow cytometry (n = 2000–10000 GUVs). e) Measured polymer lateral membrane diffusion coefficients (D (µm²s⁻¹)) for GUVs incubated for 72 h with and without B. subtilis. Measurements were performed and averaged for 10 GUVs per condition. f) Pearson correlation matrix of MALDI‐ToF‐MS comparing empty GUVs, GUVs containing biofilms, and a native biofilm grown on MSgg agar plates. Interaction plot showing the population mean of the g) SYTO 9 and h) PI fluorescence intensity of B. subtilis‐containing GUVs exposed to different kanamycin concentrations. Kanamycin is administered 2 h (control) or 48 h (biofilm) after production of GUVs and incubation at 30 °C (n = 1500–10000 GUVs). Two‐way ANOVA was used for comparison in d, g, and h: p > 0.05 (n.s.), p < 0.05 (*), p < 0.005 (**), p < 0.0005 (***), and n.s. = not significant, Tukey's post hoc test. Two‐tailed t‐test was used in e.

Figure 7

a) Schematic illustration of antibiotics testing in bacteria‐loaded GUVs. GFP (top row) and PI (bottom row) fluorescence profiles of bacteria‐GUVs exposed to different concentrations of b) ampicillin, c) kanamycin, and d) vancomycin. Observed fluorescence was fitted to a four‐parametric logistic regression (red). Data are normalized and mean and 95% confidence interval is shown (black) (n = 3). One‐way ANOVA was used for comparison in d, g, and h: p > 0.05 (n.s.), p < 0.05 (*), p < 0.005 (**), p < 0.0005 (***), and n.s. = not significant, Tukey's post hoc test.