Phage liquid crystalline droplets form occlusive sheaths that encapsulate and protect infectious rod-shaped bacteria

Significance In this study, we investigate how phage molecules secreted by pathogenic Pseudomonas aeruginosa bacteria drive antibiotic tolerance by forming phase-separated liquid crystalline compartments around bacterial cells. This study spans across spatial scales, combining atomic structure determination using electron cryomicroscopy with cellular electron cryotomography, optical microscopy, and biochemical reconstitution. We show that encapsulation of rod-shaped bacteria by spindle-shaped liquid crystalline droplets made of phage molecules is a process profoundly influenced by shape and size complementarity.


This PDF file includes:
Supplementary Materials and Methods Figures S1 to S10 Table S1 Legends for Movies S1 to S6 SI References Other supplementary materials for this manuscript include the following: Movies S1 to S6 capsid was fitted using real-space refinement (13) in PHENIX (14) using non-crystallographic symmetry (NCS) restraints. Density within the inner cavity of Pf4 filaments with ssDNA map could not be explained by any residue of the CoaB protein and was therefore assigned to correspond to Pf4 genomic DNA. While the ribose phosphate backbone corresponding to a linear single-stranded genome was clearly visible, the bases of the nucleotides were smeared due to averaging over the Pf4 genome. Therefore, we built a single stranded poly-adenine chain into a B-factor dampened (50 Å 2 ) cryo-EM density using Coot and refined the structure by iteratively rebuilding the model and real-space refinement in PHENIX. Comprehensive model validation of the Pf4 with ssDNA and Pf4 without ssDNA structures were performed in PHENIX (Table S1). Figures containing protein structures or cryo-EM data were prepared using USCF Chimera.

Fluorescent labelling of Pf4 phage
Purified Pf4 phage was dialysed into 10 mM sodium carbonate buffer pH 9.2 using a 10 kDa MWCO snakeskin dialysis membrane (ThermoFisher). One ml of Pf4 phage (5 mg/ml) was incubated with 100 µg A488 fluorescent dye (ThermoFisher) for 1 hour at room temperature (RT) with end-over-end agitation. The sample was then passed over two PD10 desalting columns (GE Healthcare) to separate A488-labelled Pf4 phage from free A488 dye. Pf4 ghost filaments were labelled with A568 following the same protocol.

Fluorescence Recovery after Photobleaching (FRAP) of Pf4 liquid crystalline droplets
To carry out FRAP experiments, 3 mg/ml Pf4 and 10 mg/ml hyaluronan were mixed in a 1:1 (v/v) ratio and placed in a capillary on a glass slide. The Zeiss ZEN software bleaching mode was used on the Zeiss LSM Exciter confocal microscope. A custom circular region within the field of view was selected. A laser power of 50% was used for the photobleaching step, set to occur every 20 frames beginning at frame 20, and a laser power of 1% was used for imaging of all frames between these steps. The Pf4 liquid crystalline droplets selected for FRAP experiments were in the bulk of the sample and imaged 2 hours after their formation. Frames were recorded with a time interval of 1.5 s.

Pf4 ghost production
The DNA genome of Pf4 was extracted using a modified version of a previously described protocol (1). Briefly, purified Pf4 phage (5 mg/ml) was incubated with 10 M LiCl in a 1:1 (v/v) ratio for 2 days at 46°C. The sample was subsequently diluted 1 in 10 with PBS and incubated with 10 µg/ml DNaseI (Sigma) and 1 U/ml benzonase (Sigma) for 2 hours at 37°C. The sample was then centrifuged (100,000 g, 1 hour, 4°C) and the pellet resuspended in PBS. Phage protein concentration was determined using a Nanodrop and infectivity (pfu/ml) was measured using a plaque assay as described above.

Antibiotic protection assay
An overnight culture of PAO1 ΔPA0728 was grown in LB media at 37°C, diluted 1 in 100 into LB medium and grown at 37°C to an OD600 of 0.5. Hundred µl of the resulting culture was added to a 96-well plate and grown for a further 30 minutes at 37°C. Hundred µl of the indicated Pf4 and/or polymer components were added to the culture such that final concentrations of components were: sodium alginate (4 mg/ml), Pf4 (1 mg/ml) and Pf4 ghost (1 mg/ml). Additionally, 20 µg tobramycin (Sigma) (Fig. 4A) or 20 µg gentamicin (Sigma) (Fig.   4B) or 10 µg colistin (Sigma) (Fig. 4C) was added as indicated. Cultures were grown for a further 3 hours before a 10 µl sample for each assay condition was taken, serially diluted 10fold and dilutions plated onto LB agar plates. Plates were incubated overnight at 37°C and colonies forming units (cfu) enumerated. For antibiotic titration experiments, tobramycin concentration was varied as indicated while other components of the assay remained constant as described (Fig. S7A). For Pf4 phage titration experiments, Pf4 concentration was varied as indicated in the presence of 20 µg tobramycin while other components of the assay remained constant as described above (Fig. S7B). Mean cfu/ml with standard deviation were calculated and plotted using Prism GraphPad software.

Fluorescence microscopy
Pf4 liquid crystalline droplets: 2 mg/ml A488-labelled Pf4 phage or A568-labelled Pf4 ghost was mixed in a 1:1 (v/v) ratio with sodium alginate (10 mg/ml) or hyaluronan (10 mg/ml) and incubated at room temperature for the indicated time period. Ten µl of the resulting sample was pipetted onto 0.7% (w/v) agar pads constructed using 15 x 16 mm Gene Frames (ThermoFisher) following the manufacturer's protocol, with a coverslip placed on top. The slide was imaged with a 100x objective using a Zeiss Axioimager M2 (Carl Zeiss) in bright field and fluorescence mode. Quantification of Pf4 liquid crystalline droplet major axis length was performed by thresholding, creating a bounding rectangle and measuring lengths using Fiji (15). Graphs were plotted using Prism GraphPad software. Images were background subtracted and figure panels prepared using Fiji.
Pf4 liquid crystalline droplets / P. aeruginosa bacteria: In order to image conditions equivalent to those utilised in the antibiotic protection assay, PAO1 ΔPA0728 was grown to an OD600 of 0.5 and incubated with A488-labelled Pf4 (final concentration 1 mg/ml) and sodium alginate 3D confocal and time-lapse microscopy of Pf4 liquid crystalline droplets / P. aeruginosa bacteria: Samples equivalent to those utilised in the protection assay were prepared as described above and z-stacks were acquired in the fluorescent and brightfield channels with an Olympus Fluoview FV1000 microscope equipped with a 100x objective. For time-lapse experiments images were acquired in the fluorescent and brightfield channels at 10 minute intervals over a 90 minute time period at room temperature using the same microscope and objective. Time-lapse images were drift corrected using the StackReg plugin in Fiji.

Propidium iodide staining of Pf4 liquid crystalline droplets / P. aeruginosa bacteria: PAO1
ΔPA0728 was grown to an OD600 of 0.5 and incubated with A488-labelled Pf4 (final concentration 1 mg/ml) and sodium alginate (final concentration 4 mg/ml) for 3 hours in the presence of 10 µg tobramycin before incubation with 0.1 mM propidium iodide for 30 minutes at room temperature. 10 µl of the sample was pipetted onto agar pads as described above and imaged with a 100x objective using a Zeiss Axioimager M2 (Carl Zeiss) in bright field and fluorescence mode. Quantitation of cell numbers and propidium iodide positive cells was performed by thresholding using Fiji. Graphs were plotted using Prism GraphPad software.
Pf4 liquid crystalline droplets/SU-8 rods: Cyanine 3 labelled SU-8 colloidal rods, fluorescent at 532 nm, were used as particles to mimic bacterial cells. These were approximately 0.5 µm in diameter and ranged between 1 and 8 µm in length. Initially dispersed in water, samples were placed in an oven for 5 minutes to dry out, before 10 mg/ml sodium alginate in PBS was added as the new solvent for the rods. Equal volumes of this solution and 3 mg/ml A488-Pf4 in PBS were combined and mixed on a vortex mixer for 5 minutes to ensure even dispersion of rods, before being placed in a capillary. This was placed onto a glass slide and imaged on the Zeiss LSM Exciter confocal microscope. A488-Pf4 tactoids were observed in one channel, and the rods were observed in the other channel (as reflected light).

Automated segmentation of images with liquid crystalline droplets surrounding bacterial cells
Brightfield channel images were Gaussian filtered, background subtracted and thresholded to identify the positions of bacterial cells. Bacterial cell shapes were found using the activecontour algorithm in Matlab (16). The regions of identified bacteria were dilated and used as seed inputs for the segmentation of the liquid crystalline droplets in the green channel using the activecontour algorithm. The segmented bacterial cells from the bright field and the liquid crystalline droplets from the green channel were used to calculate morphological parameters ( Fig. 5 and Fig. S8).

Statistical analysis
Statistical analysis was performed using Prism GraphPad software and an unpaired t-test was used to calculate p-values. Fig. S1. Purification of Pf4 phage from P. aeruginosa PAO1 biofilms (related to Fig. 1

Supplementary Movie Legends
Movie S1. Cryo-EM structure of Pf4 with ssDNA at 3.2 Å resolution (related to Fig. 1) The 3.2 Å resolution structure of Pf4 with ssDNA. Cryo-EM density (grey isosurface) and