Biophysical basis of filamentous phage tactoid-mediated antibiotic tolerance in P. aeruginosa

Abstract

Inoviruses are filamentous phages infecting numerous prokaryotic phyla. Inoviruses can self-assemble into mesoscale structures with liquid-crystalline order, termed tactoids, which protect bacterial cells in Pseudomonas aeruginosa biofilms from antibiotics. Here, we investigate the structural, biophysical, and protective properties of tactoids formed by the P. aeruginosa phage Pf4 and Escherichia coli phage fd. A cryo-EM structure of the capsid from fd revealed distinct biochemical properties compared to Pf4. Fd and Pf4 formed tactoids with different morphologies that arise from differing phage geometries and packing densities, which in turn gave rise to different tactoid emergent properties. Finally, we showed that tactoids formed by either phage protect rod-shaped bacteria from antibiotic treatment, and that direct association with a tactoid is required for protection, demonstrating the formation of a diffusion barrier by the tactoid. This study provides insights into how filamentous molecules protect bacteria from extraneous substances in biofilms and in host-associated infections.

Fig. 1. Cryo-EM structure of the fd bacteriophage capsid at 3.2 Å-resolution.

a Side view of the fd capsid (ribbon depiction) with a single pVIII subunit highlighted in purple. b Top (perspective) view of the capsid shows a 24 Å-wide lumen. The vertical interactions of capsid proteins results in 10 protofilament-like stacks observed from the top. c Atomic model of pVIII with the cryo-EM density shown as an isosurface at 6 σ away from the mean value. d Hydrophobic interaction networks within the capsid, with interacting hydrophobic residues marked. e Four lysine residues near the C-terminus extend into the capsid lumen.

Fig. 2. Comparison of the cryo-EM structures of the class I fd phage (this study) and the class II Pf4 phage (PDB 6TUP [10.2210/pdb6tup/pdb]).

a Orthographic top view of fd and Pf4 capsids shown as ribbon diagrams. Flexible residues at the fd N-termini have been modelled to allow comparison with Pf4, where all residues are ordered in the cryo-EM structure. b Clustal Omega sequence alignment of the major capsid protein of fd versus Pf4. Basic residues extending into the lumen are marked in blue. Helical regions are marked above, with fd in blue and Pf4 in salmon. c Electrostatic surface of the capsid. Disordered residues in fd (1–5) were modelled (*) to allow for a more accurate comparison of electrostatic surfaces. d Sliced side view of the capsid lumen depicting electrostatic charge distribution at the luminal surface.

Fig. 3. Atomistic molecular dynamics simulation analysis of fd and Pf4 capsids.

a Weighted ionic density of chloride ions (computed using VMD volmap) averaged over all trajectory frames. Density of chloride ions is shown in red. Dark red indicates higher ionic density. Fd capsid (left), shown in blue, depicts high levels of chloride ions in the capsid lumen (sliced at the midpoint of the filament). Pf4 capsid (right), shown in salmon, has comparatively lower levels of chloride ions in the capsid lumen (sliced at the midpoint of the filament). Both systems were simulated in 0.15 M NaCl. b Quantification of ion number at different positions in a cross section of the phage, starting from the centre of the capsid for fd (left) and Pf4 (right). Fd protein atoms are coloured blue, Pf4 coloured salmon, chloride ions red and sodium ions yellow. Protein atoms are shown for clarity. The recruitment of chloride ions to the interior of the capsid can be seen as a peak at approximately 10 Å (indicated by arrows), with a higher peak observed for fd as compared to Pf4. Mean of four repeats (simulations) is plotted for each system in bold colour, with the standard error of the mean (SEM) plotted in grey. See Supplementary Fig. 5 for additional analyses. Source data for graphs are provided as a Source Data file.

Fig. 4. Comparison of fd and Pf4 tactoid morphology and filament packing.

a, b Representative light microscopy images of tactoids formed by Alexa-488-labelled fd (cyan) and Pf4 phages (magenta). c Bar chart showing the average area of individual tactoids as assessed by light microscopy followed by segmentation of tactoids. Values shown are the mean of three independent experiments and error bars represent standard deviation. d, e Tactoid morphology as observed via cryo-ET. Inset: Zoom-in and Fourier Transform. Images are representative of 3 tomograms and 10 tomograms respectively. f Schematic of the coarse-grained model wherein phages are modelled as hard rods and the phage tactoids are modelled as tactoids with bulk (elastic) and surface energetic contributions (see Eqs. (2–3)). g Tactoid aspect ratio as a function of tactoid volume. Tactoids, as visualised via light microscopy, follow the scaling law R/r α V^-1/5 as shown through lines of best fit to R/r=CV^-1/5, where C is the fitting parameter (C_Pf4 = 5.88 ± 0.11 μm^3/5 and C_fd = 3.72 ± 0.09 μm^3/5). Source data for graphs are provided as a Source Data file.

Fig. 5. Comparison of fd and Pf4 tactoid association with bacterial cells.

a, b Representative images showing association of fd and Pf4 tactoids with P. aeruginosa cells observed in light microscopy. Transmitted light channel shows bacteria (yellow) and fluorescence channel shows fd (cyan) and Pf4 (magenta) tactoids, respectively. c Histogram of pairwise orientational differences between bacterial cells and associated tactoids from semi-automated segmentation of images. Values reported are the mean and standard deviation from three independent experiments. Left panel fd + P. aeruginosa (n = 144), right panel Pf4 + P. aeruginosa (n = 163), significantly different with ***P_value = 0.0010. d Percentage of P. aeruginosa cells associated with tactoids (n = 205 for fd + P. aeruginosa and n = 216 for Pf4 + P. aeruginosa). Differences in association are not statistically significant, P_value = 0.1735. Values shown are the mean of three independent experiments and error bars represent standard deviation. e, f Corresponding representative images showing association of fd and Pf4 tactoids with E. coli. Transmitted light channel shows bacteria (yellow) and fluorescence channel shows fd (cyan) and Pf4 (magenta) tactoids, respectively. g Histogram of pairwise orientational differences between bacterial cells and associated tactoids from semi-automated segmentation of images. Values reported are the mean and standard deviation from three independent experiments. Left panel: fd + E. coli (n = 244), right panel: Pf4 + E. coli (n = 137), no significant difference, P_value = 0.1190. Pf4 + P. aeruginosa versus Pf4 + E. coli, *P_value = 0.0110. Fd + P. aeruginosa versus fd + E. coli, no significant difference, P_value = 0.3623. Pf4 + P. aeruginosa versus fd + E. coli, *** P_value < 0.0001. Fd + P. aeruginosa versus Pf4 + E. coli, no significant difference, P_value = 0.5404. h Percentage of E. coli cells associated with tactoids (n = 344 for fd + E. coli and n = 188 for Pf4 + E. coli). Differences in association are not statistically significant, P_value = 0.5519. Values shown are the mean of three independent experiments and error bars represent standard deviation. All p-values were calculated using an unpaired t test. Source data for graphs are provided as a Source Data file.

Fig. 6. Association with tactoids protects bacterial cells from antibiotic uptake, suggesting the formation of a diffusion barrier.

Bar graphs showing viability of cells in colony-forming units (cfu) per ml (y-axis) in the presence of different reagents (x-axis) from three independent experiments for (a) P. aeruginosa and (b) E. coli against tobramycin treatment. Both Pf4 and fd tactoids protect P. aeruginosa to a significantly greater extent than sodium alginate alone (P_value = 0.0249 and 0.0384 respectively). Fd tactoids protected significantly less well than Pf4 tactoids (P_value = 0.036). Both Pf4 and fd tactoids protect E. coli to a significantly greater extent than sodium alginate alone (P_value = 0.0271 and 0.0401 respectively). No significant difference (n.s.) was observed between the level of protection provided by fd and Pf4 tactoids (P_value = 0.0830). c–j Fluorescence and light microscopy images of Alexa 488-labelled fd, Pf4 and Pf4 ghost phage tactoids incubated with Texas Red-labelled gentamicin (GTTR) for 4 h. Shown are Alexa488 phage signal (Cyan – fd, Magenta – Pf4) and pseudocoloured cells (yellow) as determined through brightfield light microscopy (left), and Texas Red signal corresponding to uptake of the fluorescently labelled antibiotic GTTR by cells (right). Numbering indicates site of cells in corresponding images. Images are representative of 30 images taken over three biological replicates (Control, n = 149 cells, fd associated, n = 128, fd non-associated, n = 128, Pf4 associated, n = 145, Pf4 non-associated, n = 69, Pf4 ghost associated, n = 102 and Pf4 ghost non-associated, n = 100). k Plot quantifying GTTR uptake after 4 h respectively in conditions indicated on the x-axis. Bar shows the mean of three independent experiments and error bars represent standard deviation. Association with both Pf4 and fd tactoids results in significantly reduced antibiotic uptake as compared to a control with no phage (fd P_value < 0.0001, Pf4 P_value < 0.0001), and as compared to cells in the same sample that are not tactoid-associated (fd P_value < 0.0001, Pf4 P_value < 0.0001). No significant difference (n.s.) in antibiotic uptake was observed between Pf4 and Pf4 ghost tactoids (P_value = 0.9040). All p-values were calculated using an unpaired t test. Source data for graphs are provided as a Source Data file.

Fig. 7. Schematic depicting biophysical nature of phage tactoid-mediated antibiotic tolerance of bacteria.

The inoviruses, fd and Pf4, form tactoids with liquid-crystalline order in the presence of biopolymer with distinct morphologies dictated by phage size. Both fd and Pf4 phage tactoids associate with P. aeruginosa and E. coli K12, which have different cell surface chemistries. The material properties of the tactoids govern the association with bacteria independent of their cell surface properties. Phage tactoid association with bacteria results in a diffusion barrier that leads to increased antibiotic tolerance, which correlates with the level of bacterial cell encapsulation by the tactoid.