Strains, Plasmids and Growth Conditions
Bacterial strains, plasmids and their relevant characteristics are listed in Supplementary Table 1. Strains were grown in lysogeny broth (LB) (10 g L-1 tryptone, 5 g L-1 yeast extract, and 4 g L-1 NaCl) as needed, 1.5 g L-1 agar was added. Final concentrations of antibiotics used when required: 25 Âµg mL-1 kanamycin (Km), 60 Âµg mL-1 tetracycline (Tet), 20 Âµg mL-1 gentamicin (Gm). Cultures with plasmids used to start experiments have always been grown with appropriate antibiotics to prevent loss of plasmids.
Construction of strains and plasmids for detection of loss of plasmids
P. putida KT2442 and P. putida MRB1 were both marked with PA10403gfpmut3-GenerationR using the Tn7 system21,22,23. pMiB4 and pMiB8 are derivatives of the IncP-1 conjugative plasmid pKJK5 which was previously complemented with kanamycin resistance and constitutively expressed here, thereq. Here we have inserted mCherry and GenerationR in drfA1 of pMiB4 and pMiB8 using the Î» Red recombination system24. The combination of these plasmids with PA10403gfpmut3-tagged P. putida KT2442 and P. putida MRB1, allowed detection of plasmid loss: cells with plasmids express plasmid-encoded mCherry, while chromosomal GFPmut3 is repressed by plasmid-encoded LacIq. Cells that become plasmid-free will start expressing GFP only, as plasmid-encoded mCherry and LacIq repressor are lost.
Loss of planktonic plasmid and biofilm detected by flow cytometry
The stability of both the conjugative plasmid and the conjugation-deficient plasmid was studied in planktonic and biofilm cultures to assess the effect of horizontal transfer on plasmid loss in the absence of plasmid-specific selection. Planktonic cultures were grown in tubes containing 5 mL of LB. From these cultures, 5 ÂµL were transferred daily to new 5 mL LB tubes. The tubes were incubated at 24 Â° C and 250 rpm. To obtain the ratio of cells without plasmid to cells carrying plasmids, samples were prepared by washing cells in 0.9% saline (NaCl: 9 g L-1) three times before analysis on a FACSAria IIIu flow cytometer (Becton Dickinson Biosciences, San Jose, CA, USA).
Biofilm experiments were performed in silicone tubes with ten-fold diluted LB medium. Silicone tubes measuring 3 Ã 1 Ã 5 mm (Mikrolab Aarhus A / S) were used for the growth of the biofilm in combination with an assortment of suitable connectors and a 205 S / CA peristaltic pump (Watson Marlow). Overnight cultures were adjusted to OD600 0.01 before inoculation. Then the bacteria were allowed to attach to the flow cell for 2 h. After that, the flow started at 2.5 ml h-1 so that unattached cells have been removed from the system. The biofilms were grown at room temperature.
Biofilms were collected by harvesting three samples from a row collecting sequential 1.5 cm sections of the row 1.0 cm downstream of the inoculation site and beyond. Each section was cut into halves and the two halves were placed in 1 ml of 0.9% saline solution. The samples were then sonicated for 5 min at 45 kHz (VWR Ultrasonic Cleaner USC-THD) and degassed for 2 min at power setting 6 before analysis on the FACS.
Loss of plasmid from biofilm and planktonic samples was detected using FACS with a 488nm (20mW) laser connected to a green fluorescence detector (530/30nm bandpass filter) and a 561 nm (50 mW) connected to the red fluorescence detector (610/20 nm bandpass filter). BD FACSDiva software v.6.1.3 was used for both operation and analysis. See Fig. 2 additional for the trigger strategy.
A competition test was performed to measure the fitness effect of the plasmid on the strains. This was done like DelaFuente et al.25 with modifications. Briefly, overnight cultures of strains with and without plasmid were washed in LB 3 times and adjusted to OD600 1. The adjusted cultures were then mixed and used to inoculate 5 Âµl in 5 ml of LB. The tubes were incubated at 24 Â° C at 250 rpm. To estimate the initial and final numbers, the adjusted cultures and mixtures were measured on the FACSAria IIIu flow cytometer in 0.9% saline (NaCl: 9 g L-1). The dilution used for flow cytometry and the sampling time were recorded allowing the calculation of the return to the pr ml event based on the input volume taken by the machine.
Loss of plasmid in alginate beads
Alginate beads were prepared according to the protocol published by Sonderholm et al.26, with modifications27. The beads were prepared using alginate extracted from brown seaweed Laminaria hyperborea (Protanal LF10 / 60; FMC Biopolymer, Drammen, Norway). A total of 100 ml of alginate solution was prepared by suspending 4.9 g of Protanal LF10 / 60 in 70 ml of MilliQ water at 50 Â° C until dissolved, then adding 30 ml of MilliQ water. The alginate solution was then autoclaved at 120 Â° C for 20 minutes. For the preparation of the beads, 7.5 ml of the alginate solution was mixed with 2.5 ml of washed cell suspension and adjusted to OD of precultured planktonic cultures. Cell cultures were added to the alginate solution, aiming for example ~ 108 ml cells-1. Droplets of the alginate-bacteria suspension were dispensed via a 21 gauge needle placed 3 cm above the surface of a stirred 0.25 M CaCl.2 solution at room temperature. The droplets were formed at constant speed using a syringe pump. A total of 2 ml of alginate-bacteria suspension was distributed in droplets giving beads of approx. 2.4mm size. The beads were allowed to harden for 5 min in the stirred solution before being transferred to a new 250 ml vial with 100 ml of 0.25 M CaCl.2 hardening. The beads were allowed to harden completely for 30 min while shaking at 150 rpm at room temperature. After curing, the beads were transferred to a new 250 ml vial with 100 ml of 0.9% NaCl wash solution to remove weakly attached cells from the surface of the beads. The beads were washed in the wash solution for 30 min while shaking at 150 rpm at room temperature. After washing, the beads were transferred to new 250 ml vials with 100 ml 10% LB and incubated overnight at 24 Â° C with horizontal shaking at 150 rpm.
Individual beads were selected from the vials and divided into two equal halves using a razor blade. One half was placed with the cut side down in a clear WillCo dishÂ® Glass bottom dish (HBST-12, Willco Wells BV, Ã dish / glass: 35/12 mm, Glass thickness: 0.17 mm / # 1.5). The bead was covered with 130 ÂµL of 0.9% NaCl as a fountain solution to keep the bead moist. Visualization was performed with a confocal laser scanning microscope (CLSM) (LSM 800, Zeiss) with an EC Plan-Neofluar10x / 0.30 M27. Z stacks were recorded using Axiocam 503 mono and excitation at 488 nm for GFP and 561 nm for mCherry. The free open source software ImageJ (National Institute of Health, USA) was used to define the threshold using Otsu’s algorithm for the background signal before quantification and analysis of the distribution in R statistical language.
Plasmid loss in biofilm examined by CLSM
Biofilms were grown in flow cells following the same procedure described in Olsen et al.28. The flow cells were 1 Âµ-slide VI0.4 (Ibidi) to allow visualization of the development of the biofilm. The flow cell system was filled with five-fold diluted LB medium and the chambers were inoculated with overnight cultures adjusted to OD.600 of 0.05. After inoculation, the bacteria were allowed to settle for 1 h before starting the pump at a flow rate of 2.5 mL h-1 at room temperature. The biofilms were analyzed in situ on an inverted CLSM (LSM 800, Zeiss) with a Plan-Apochromat 63x / 1.4 DIC M27 oil. Z-stacks of different spots were recorded after 48 h. The free open source software ImageJ (National Institute of Health, USA) was again used to define the threshold of the background signal with the Yen and Moments algorithm for both channels before quantification and distribution analysis in the statistical language R.
Statistical analysis of FACS data
Samples of planktonic cultures were analyzed by FACS in four biological replicates, which were used to calculate the mean and standard deviation. Samples of biofilm cultures with strains WT and MRB1 with the conjugative plasmid were analyzed by FACS in four biological replicas and three technical replicas and MRB1 with the conjugation-deficient plasmid in three biological replicas and three technical replicas, for which the measurements were averaged. The differences in the trends in plasmid loss were compared with linear models and the trends package function emmeans29.
Image analysis of flow cell biofilms
For each image, the positions in the z direction were changed by subtracting the mean z position weighted by the number of pixels in each layer; thus, ensuring that a result z position 0 indicates the layer with the most biomass, making different images comparable. The numbers of pixels from different images in each biological replica were averaged. For each biological replica, a loess regression was adjusted to the proportion of plasmid loss. For statistical analysis of the data presented in Figure 2a, the ratio of cells without plasmid to cells carrying plasmid was calculated for each layer of the images in the z direction. Then the averages were calculated based on the top and bottom ratios of the biofilms. The top and bottom have been marked as corrected z above or below 0, respectively, so that each contains half the biomass. The images are available on Zenodo (https://doi.org/10.5281/zenodo.5493705).
Image analysis of alginate beads
The images were the first median smoothed with a 3 Ã 3 Ã 3 filter (smoothIMG function). Then the centers of the alginate beads were calculated as the middle position of all pixels in the image (the center of mass function). Subsequently, the clusters were detected by grouping adjacent pixels, including diagonals (tufts function30), and the distances from the aggregates to the center of the bead were calculated. Aggregates less than 50 cubic microns were removed and aggregates visibly outside the alginate beads were removed. The distances to the center were normalized for each bead by dividing by the maximum distance. The images are available on Zenodo (https://doi.org/10.5281/zenodo.5493705).
Summary of the report
Further information on the research design can be found in the Nature Research Report Summary linked to this article.