High-throughput detection of potential bacteriocin producers in a large strain library using live fluorescent biosensors

The global increase in antibiotic resistances demands for additional efforts to identify novel antimicrobials such as bacteriocins. These antimicrobial peptides of bacterial origin are already used widely in food preservation and promising alternatives for antibiotics in animal feed and some clinical setting. Identification of novel antimicrobials is facilitated by appropriate high throughput screening (HTS) methods. Previously, we have described a rapid, simple and cost-efficient assay based on live biosensor bacteria for detection of antimicrobial compounds that act on membrane integrity using the ratiometric pH-dependent fluorescent protein pHluorin2 (pHin2). Here, we use these biosensors to develop an integrated pipeline for high-throughput identification of bacteriocin producers and their biosynthetic gene clusters. We extend the existing portfolio of biosensors by generating pHin2 expressing strains of Escherichia coli, Bacillus cereus, Staphylococcus epidermidis, and methicillin-resistant Staphylococcus aureus. These strains were characterized, and control experiments were performed to assess heterogeneity of these biosensors in response to known bacteriocins and develop a robust HTS system. To allow detection of compounds that inhibit target bacteria by inhibiting growth without disturbing membrane integrity, the HTS system was extended with a growth-dependent readout. Using this HTS system, we screened supernatants of a total of 395 strains of a collection of lactic acid bacteria. After two rounds of screening 19 strains of the collection were identified that produced antimicrobial activity against Listeria innocua and Listeria monocytogenes. Genomes of confirmed hits were sequenced and annotated. In silico analysis revealed that the identified strains encode between one and six biosynthetic gene clusters (BGCs) for bacteriocins. Our results suggest that pHin2 biosensors provides a flexible, cheap, fast, robust and easy to handle HTS system for identification of potential bacteriocins and their BGCs in large strain collections.

(A) Results of the pHin2 assay as readout.(B) The same MTP plates analyzed for pHin2 FI ratios assay were then used for a growth-dependent readout by adding 100 µl of sterile BHI and measurement of OD600 at t = 0 h and t = 4 h of incubation at 37 °C with aeration.All values are mean of n = 3 technical replicates for each supernatant with standard deviation.Values were normalized pHluorin ratios (expressed as % of intact cells) or normalized ΔAbs600 (% growth) with untreated biosensors (neg.controls) set as upper boundary (100%) and nisin-treated (10 µg mL -1 ; pHin2 assay) or ampicillin-treated (100 µg mL -1 ; Abs600) set as baseline (0%).

Figure S1 :
Figure S1: pHin2 assays with supernatants of different control strain using L. innocua LMG2785/pNZ-pHin2 Lm .biosensors.L. innocua LMG2785/pNZ-pHin2 Lm was resuspended in LMBO and incubated with supernatants (SN) of L. sakei A1609, Pediococcus acidilactici A1610, or L. lactis B1629 obtained after overnight growth in MRS medium.Sterile MRS with or without nisin (10 µg mL -1 ) used as controls.Bacteria were then analyzed for pHin2 fluorescence and results are expressed as ratios of fluorescence intensity (RFU ratio, emission at 520 nm) after excitation at 400 and 480.Values are mean ± standard deviation of n = 2-3 replicates (i.e.independent cultures of the biosensor).

Figure S2 :
Figure S2: HPLC analysis confirms presence of lactate in supernatants of L. lactis B1629.Supernatants (SN) were obtained after cultivation for 16 h in MRS medium.As controls pure lactate (50 mM in H2O) or sterile MRS medium were analyzed.HPLC was performed on a Agilent 1200 series apparatus (Agilent Technologies, Santa Clara, CA, United States) with a Refractive Index Signal Detector (RID) and signals were recorded as normalized refractory index units (nRIU).The dashed vertical line indicates the peaks for lactate at t = 10.8 min of retention.Lactate concentrations in SN was estimated to be >100 mM based on peak heights (25 000 nRIU for SN vs. 10 000 nRIU for 50 mM lactate standard).

Figure S3 :
Figure S3: Validation of pHin2-and growth-dependent readouts of E. coli MG1655/pNZ-pHin2Lm .E. coli MG1655/pNZ-pHin2 Lm was resuspended in LMBO and treatment with either CTAB to disrupt membrane integrity or ampicillin (100 µg mL -1 ) to inhibit growth without membrane damage.(A) Bacteria were then analyzed for pHin2 fluorescence and results are expressed as ratios of fluorescence intensity (RFU ratio, emission at 520 nm) after excitation at 400 and 480.(B) The same MTP plates were then used for a growth-dependent readout by adding 100 µl of sterile BHI and measurement of Abs600 at t = 0 h and t = 4 h of incubation at 37˘ °C with aeration.Results are shown as changes in Abs600 between the two measurements (ΔAbs600).All values are mean ± standard deviation of n = 3 replicates (i.e.independent cultures of the biosensor).

Figure S4 :
Figure S4: HTS of the supernatant library against the Gram-negative sensor strain E. coli MG1655/pNZ-pHin2 Lm .(A)Results of the pHin2 assay as readout.(B) The same MTP plates analyzed for pHin2 FI ratios assay were then used for a growth-dependent readout by adding 100 µl of sterile BHI and measurement of OD600 at t = 0 h and t = 4 h of incubation at 37 °C with aeration.All values are mean of n = 3 technical replicates for each supernatant with standard deviation.Values were normalized pHluorin ratios (expressed as % of intact cells) or normalized ΔAbs600 (% growth) with untreated biosensors (neg.controls) set as upper boundary (100%) and nisin-treated (10 µg mL -1 ; pHin2 assay) or ampicillin-treated (100 µg mL -1 ; Abs600) set as baseline (0%).