We screened wastewater samples at the three Northeast Ohio Regional Sewer District water treatment plants in Cleveland for bacteriophage. The majority of phage in these samples used type IV pili as a receptor, and we wanted to exclude these from our screen. We had previously generated a ΔfliF ΔpilA double mutant strain in the lab and decided to use it to exclude both surface appendages as potential receptors. This turned out to be fortuitous, since, surprisingly, the screen identified four phage that could form plaques on the ΔfliF ΔpilA double mutant strain, but not the parental wild type P. aeruginosa PAO1. We named these Cleveland wastewater-derived phage Clew-1, -3, -6, and -10. Subsequent tests determined that it was the fliF deletion that rendered P. aeruginosa permissive for infection by these phages. All four Clew phage can plaque on a fliF deletion mutant of P. aeruginosa PAO1, but not the corresponding wild-type strain or ΔpilA mutant strain. (Fig. 1A, S1A, S1B, S2). An unrelated Pbunavirus we isolated in the same screen, which uses O-antigen as receptor (Fig. S3), was used as a control in these experiments (Our control phage, Ocp-2). The Clew bacteriophages belong to the family of Bruynogheviruses (25) and are all highly related (Fig. 1B, S1C). Morphologically, like other members of the family, they are Podoviruses (Fig. 1C).

We next examined what part of the flagellum is involved in determining sensitivity to the Clew-1 phage. Mutations affecting the MS-ring (ΔfliF) and associated proteins FliE or FliG (26) resulted in Clew-1 sensitivity. However, mutations in the ATPase complex only conferred partial sensitivity, and mutations affecting the hook or flagellar filament did not result in sensitivity, nor did a mutation that affects the type III secretion function of the flagellar basal body by impeding proton flux, flhA(R147A)(27), (Fig.S4). We therefore conclude that it is the presence of the MS-ring and not other aspects of the flagellum, such as assembly of the full flagellar structure or flagellar rotation, that control phage sensitivity.

Interestingly, we found that deletion of fleQ, which is required for transcription of flagellar genes (28), had a very minor effect on Clew-1 phage susceptibility of the wild-type bacteria, and actually decreased Clew-1 susceptibility of the ΔfliF mutant bacteria (Fig. S1E). FleQ is a c-di-GMP-responsive transcription factor that, among other things, reciprocally controls flagellar gene expression and production of biofilm-related characteristics, such as the production of the extracellular polysaccharides Psl and Pel, as well as the adhesin CdrA (28–30). We therefore examined whether manipulating c-di-GMP levels controls phage susceptibility. To this end we produced the c-di-GMP phosphodiesterase PA2133 from a plasmid (31) to artificially lower c-di-GMP levels in the ΔfliF2 deletion mutant strain. Conversely, we artificially elevated c-di-GMP levels in the wild-type by deleting the wspF gene (31). Lowering c-di-GMP levels in the ΔfliF2 mutant restored Clew-1 resistance (Fig. 1D), whereas deleting wspF rendered the parental PAO1 strain phage sensitive (Fig. 1E). Taken together, these data demonstrate that Clew-1 susceptibility is controlled by intracellular c-di-GMP levels and argue that absence of the MS-ring controls phage susceptibility through an increase in c-di-GMP.

To better understand the host factors that control susceptibility and resistance to Clew-1 infection, we carried out a pair of TnSeq experiments. In the first of these, we mutagenized the wild-type strain PAO1F with the mini-mariner transposon TnFAC (32), and the resultant mutant library was infected with phage Clew-1 at an MOI of 10 for 2 hours. The surviving bacteria were allowed to grow up after plating on an LB plate and the transposon insertion sites for the input and output pool were determined by Illumina sequencing. We identified insertion mutants that were depleted after infection (Fig. 2A). Two of the genes with the most significant depletion were fliF and fliG, consistent with our previous analysis indicating that these mutations sensitize PAO1 to Clew-1 infection. Interestingly we also noted depletion of pch and bifA insertions, both encoding phosphodiesterases that are involved in depleting c-di-GMP in the flagellated daughter cell after cell division (33–36). In fact, pch interacts with the chemotactic machinery (33), highlighting, here too, the importance of c-di-GMP in controlling Clew-1 sensitivity.

In a reciprocal experiment, we carried out the TnSeq analysis in a fliF mutant strain. This analysis identified insertions in the psl operon as the most highly enriched group of mutants after Clew-1 selection, suggesting that Psl is required for phage infection (Fig. 2B). We examined the requirement for Psl explicitly by generating pslC and pslD mutants in the PAO1F ΔfliF2 strain background. PslC is a glycosyltransferase required for Psl biosynthesis, while PslD is required for Psl export from the cell (37, 38). Deletion of either pslC or pslD rendered the fliF mutant bacteria Clew-1 resistant and sensitivity could be restored through complementation using a plasmid-borne copy of the deleted open reading frame (Fig. 2C). These data demonstrate that Psl production is required for infection of P. aeruginosa by phage Clew-1.

We next examined whether attachment of Clew-1 to P. aeruginosa is Psl-dependent. We first used efficiency of center of infection (ECOI) analysis to examine attachment. In this analysis, the phage is allowed to adhere to the bacteria for 5 minutes, before washing the bacteria to remove unattached phage. The bacteria are then diluted, mixed with top agar and a sensitive indicator bacterium (ΔfliF2), and then plated to allow for plaque formation as a biological readout of attached bacteriophage. Attachment of phage Clew-1 is Psl-dependent. Interestingly, we were able to detect Psl-dependent attachment both with wild-type and ΔfliF2 mutant bacteria (Fig. 3A), which contradicted out initial efficiency of plating experiments. We therefore reexamined phage susceptibility by monitoring phage infection in liquid media and generating lysis curves for wild-type and ΔfliF2 mutant bacteria, as well as their ΔpslC mutant derivatives (Fig. S5). The ΔfliF2 mutant strain was lysed after ∼40 minutes of infection. The wild-type bacteria displayed a significant slowing of growth upon Clew-1 infection when compared to the uninfected culture, but not clear lysis as was observed with the ΔfliF2 mutant. In both instances, deleting pslC abolished any phage-dependent effect on growth.

We hypothesized that perhaps, the difference between wild-type and ΔfliF2 mutant bacteria is due to the fraction of cells that are producing Psl and therefore permissive for phage attachment. To test this hypothesis, we labeled phage Clew-1 with the DyLight594 fluorescent dye and examined attachment directly by microscopy (Fig. 3B). We observed a statistically significant increase in the percentage of bacteria with attached bacteriophage in the ΔfliF2 mutant bacteria compared to the wild-type, arguing that the increase in c-di-GMP in the ΔfliF2 mutant increases the fraction of Clew-1 susceptible cells in the population. As anticipated, no phage was observed attached to the corresponding ΔpslC mutant (Fig. 3C).

We next examined whether phage Clew-1 can bind to Psl directly. We first determined whether we could precipitate phage Clew-1 from filter sterilized culture supernatants of a ΔfliF2 mutant using an antibody directed against Psl. The presence of the phage was determined by quantitative PCR. We were able to pull down phage Clew-1 in a Psl and antibody-dependent manner with ΔfliF2 ΔpslC culture supernatants serving as a control (Fig. 4A). Notably, we observed some Psl-dependent attachment in the absence of antibody, arguing that Psl binds non-specifically to the magnetic beads we used in our experiments. Including the anti-Psl antibody resulted in a statistically significant increase compared to this background level of attachment (Fig. 4A). We next repeated the pulldown using a partially purified fraction of cell-associated Psl to repeat the pulldown and again found Psl and antibody dependent precipitation of phage Clew-1 (Fig. S7). Finally, we examined phage binding using a biotinylated, affinity purified preparation of Psl and found that we could pull down the phage using this Psl fraction as well, arguing that Clew-1 binds Psl directly (Fig. 4B).

Since Clew-1 exploits Psl for infection and Psl is a key component of the biofilm matrix of most strains of P. aeruginosa (39, 40), we hypothesized that, perhaps, Clew-1 can infect biofilm bacteria. We used a static-biofilm model to test this hypothesis. Biofilms were established overnight in a 96-well plate. One plate was washed and fixed with ethanol to quantify the day one biofilm mass using crystal violet staining. In a second plate, established in parallel, the biofilms were washed with PBS and LB was added back, either without addition, or with 10^9 pfu of phage Clew-1 or phage Ocp-2. The plates were incubated overnight and the next day, the day 2 biofilm mass was quantified using crystal violet. A similar experiment was carried out in 5-mL culture tubes to illustrate the result is shown in Fig. 5A. The averages of 5 biological replicates in the 96-well experiment are shown in Fig. 5B. Treatment of the day one biofilm with phage Clew-1 resulted in a statistically significant decrease in biofilm mass compared to the biomass present at day 1. Phage Ocp-2 infection, on the other hand, did not result in a reduction in biofilm (Fig. 5B). Notably, phage Clew-1 was not able to reduce a biofilm formed by P. aeruginosa strain PA14, a natural psl mutant (Fig. S8).

To corroborate this result, we also conducted a converse experiment where we monitored the ability of phage Clew-1 or Ocp-2 to replicate on biofilm bacteria over a two-hour period. Here, biofilms were generated overnight in 5-mL culture tubes, the biofilms were washed with PBS and exposed to 10^5 pfu/mL phage Clew-1 or phage Ocp-2 for 2 hours. At the end of the experiment, the culture supernatants were filter sterilized and the phage were titered. Consistent with the reduction in biofilm mass seen in Fig. 5B, we found that phage Clew-1, but not Ocp-2, was able to replicate when grown on biofilm bacteria (Fig. 5C).

In order to determine distinguish whether Clew-1 infection results in decolonization of bacteria from the biofilm, or killing of biofilm bacteria, we carried out two complementary experiments. In the first experiment, we generated wild type P. aeruginosa biofilms on polystyrene microbeads grown overnight with agitation. The beads were then washed with PBS and incubated for 2 hours either in LB or LB with Clew-1. Bacteria were released from the biofilm by sonication and the input CFU (total CFU attached to beads) were compared to the output after 2h of incubation (sum of supernatant and bead-associated viable bacteria). Clew-1 resulted in a significant reduction in viable CFU, arguing that phage infection kills biofilm bacteria (Fig. 5D). In a related experiment, we used LIVE/DEAD staining of bacteria to assess bacterial viability by microscopy. We established biofilms in microwell slides overnight, then added Syto 9 and propidium iodide (PI) to each well along with phage Clew-1, or a buffer control, and incubated the slide overnight again before imaging the biofilms by confocal microscopy. Infection with Clew-1 resulted in a significant increase in the fraction of the biofilm volume that is PI-positive, indicating that the biofilm bacteria are dying (Fig. 5E).

Given the ability of phage Clew-1 to infect P. aeruginosa biofilms, we next examined whether Clew-1 could be used to treat a P. aeruginosa infection. Corneal infections by P. aeruginosa involve formation of a biofilm (8, 40). In fact, a bivalent antibody directed against Psl and the type III secretion needle-tip protein, PcrV, was found to be effective in clearing such corneal infections (8). We therefore examined the ability of Clew-1 to reduce the P. aeruginosa bacterial burden in a corneal infection model. Mice were infected with 5*10^4 cfu of wild-type P. aeruginosa strain PAO1 and given a topical application of 5*10^9 pfu Clew-1 in 5µL of PBS, or PBS alone, at 24h and 48h post-infection (Fig. 6A). After 48 hours the infection we quantified corneal opacity, a measure that correlates with the infiltration of immune cells (41–43), and GFP fluorescence (produced by the P. aeruginosa strain used in the infection) by image analysis. We also assessed the bacterial burden (colony forming units). Mice infected with PAO1 developed severe corneal disease manifest as corneal opacification in the region of bacterial growth indicated by GFP fluorescence (Figure 6B-D). Phage Clew-1 treatment was able to significantly reduce the bacterial burden, and in some instances completely eradicate the infection, as measured by GFP fluorescence and CFU (Fig. 6D, E). Corneal opacity, was similarly reduced by the phage treatment, indicating that inflammation was also starting to resolve (Fig. 6B, C).

We describe the isolation of four phage belonging to the family of Bruynogheviruses that use the P. aeruginosa exopolysaccharide Psl as a receptor. Psl is not a capsular polysaccharide, so this distinguishes the Clew phages from phages such as KP32 that infect Klebsiella pneumoniae. Moreover, these Klebsiella phages use a capsular depolymerase to break down the capsular polysaccharide (44, 45). Clew-1, on the other hand, has no such activity (Fig. S9) arguing that the role of Psl in infection is distinct from that seen in capsule-targeting bacteriophage.

Phage Clew-1 has the surprising quality that it fails to plaque on wild-type P. aeruginosa PAO1, but forms plaques on a fliF mutant. We determined that the fliF mutation generates a c-di-GMP dependent signal that up-regulates Psl production. Importantly, it increases the fraction of bacteria to which the phage can bind, resulting in efficient lysis in liquid cultures, and plaque formation in top agar. Plaque formation is likely masked in the wild-type bacteria by the fraction of cells that are not phage-susceptible. Notably, certain Bruynogheviruses are able to bind to P. aeruginosa PAO1, but not plaque (46). We now have an explanation for this observation.

The identification of Psl as phage receptor prompted us to examine the ability of phage Clew-1 to infect wild-type P. aeruginosa in a biofilm. We found that phage Clew-1, unlike the unrelated Ocp-2 phage, was able to disrupt biofilms formed by wild-type bacteria. Moreover, Clew-1 was able to actively replicate on biofilm bacteria, while phage Ocp-2 could not. Taken together, our data suggests that phage Clew-1 has specialized to replicate on P. aeruginosa growing in a biofilm. Given the prevalence of bacterial biofilms in nature, this specialization makes sense. Moreover, our observation suggests that we may have underestimated the prevalence of biofilm-tropic bacteriophage since standard isolation techniques using plaque formation of wild-type bacteria would miss phage akin to Clew-1. In fact, another bacteriophage was recently described that requires an intact psl operon for replication and can only plaque on PAO1 with elevated c-di-GMP levels (35). This bacteriophage, Knedl, belongs to the family of Iggyvirueses (47), highlighting that more biofilm-tropic bacteriophage wait to be discovered. Our data also suggest that, for P. aeruginosa, using a ΔfliF ΔpilA double mutant will allow us to enrich for biofilm-specific bacteriophage, by excluding dominant type IV pilus-dependent phage and up-regulating biofilm-specific surface structures such as Psl, Pel, and CdrA. Given the importance of biofilms in contributing to the antibiotic resistance of P. aeruginosa in infections such as the CF lung, catheters or wound infections, treatment modalities that are targeted towards biofilm bacteria are sorely needed. Indeed, phage Clew-1 shows some promise in this regard, since it was able to kill biofilm bacteria in vitro, and control P. aeruginosa infection in a mouse model of keratitis, which involves biofilm formation at the site of infection. While Clew-1 plaques can sometimes look turbid, this is likely a function of the fact that Clew-1 can only bind to a fraction of the bacterial population (80-90% in the case of PAO1 ΔfliF2, Fig. 3). We have not been able to isolate Clew-1 lysogens, which is consistent with observations using other Bruynogheviruses (18, 48, 49) and argues that Clew-1 is lytic and not temperate. While many bacteriophages are not able to infect P. aeruginosa biofilms, some phage with the ability to target biofilms have been described, including the Bruynoghevirus Delta (18). We present here a way by which phage that target P. aeruginosa biofilms can be enriched during isolation.

Capsule-targeting bacteriophage tend to use LPS as a secondary receptor for tight binding to the target cell. While Clew-1 does not seem to require either A-or B-band LPS for infection (Fig. S3), we did hit several genes involved in the biosynthesis of outer core LPS in our TnSeq analysis (Table S3). We tested whether one of these genes, ssg (PA5001), is required for Clew-1 infection and found that this is the case (Fig. S3). How binding to Psl and LPS relate to the infection of P. aeruginosa by Clew-1 will be examined in a future study.

Another interesting aspect of the work described herein is the relationship between the presence of the MS-ring (FliF) and its associated proteins FliE and FliG with c-di-GMP levels. While it has been noted previously that flagellar mutations lead to increases in c-di-GMP levels and increased production of Psl upon surface contact (50), our results differ somewhat in that phage susceptibility was primarily the result of loss of the MS-ring and associated proteins (FliEFG), not, for example, the flagellar filament (FliC). This difference may be due to differences between planktonically grown (as in our study) and surface-attached bacteria. Surface contact leads to up-regulation of c-di-GMP through surface sensing by the Wsp chemosensory system (31, 51). Attached bacteria divide asymmetrically, with c-di-GMP levels decreasing in the flagellated daughter cell (33–35, 52). This asymmetry requires the phosphodiesterase Pch, which has been reported to bind to the chemosensory protein CheA (33, 34). A second phosphodiesterase, BifA, is also required for maintaining c-di-GMP homeostasis and developing an asymmetric program of cell division upon attachment to surfaces (35, 36). In our TnSeq experiment we found that insertions in flagellar genes, such as fliF and fliG, but also insertions in pch and bifA resulted in Clew-1 sensitivity. Whether the strong Clew-1 sensitivity associated with deletion of fliF, fliE, or fliG in our data, relative to deletions in other flagellar components, relates to a pivotal role of the MS-ring in controlling the activity of Pch and/or BifA is unclear, but worth further investigation. However, our work, along with the work of the Jenal group (35), suggests that phage such as Clew-1 or Knedl could be a useful tool for interrogating c-di-GMP signaling pathways in P. aeruginosa.

In summary, we have described here the isolation of a group of bacteriophages that target P. aeruginosa biofilms by using the exopolysaccharide Psl as a receptor. Consistent with the critical role of Psl as part of the P. aeruginosa biofilm matrix, we demonstrate that phage Clew-1 can replicate on biofilm bacteria and control P. aeruginosa in a mouse model of keratitis. Moreover, we have described a generalizable method that allows for the enrichment of biofilm-tropic bacteriophage, which is important due to their potential utility in combating biofilm infections that are notoriously recalcitrant to antibiotic therapy.

Bacterial strains were grown in LB (10g/L tryptone, 5g/L yeast extract, 5g/L NaCl) at 37°C unless indicated otherwise. Bacterial strains and plasmids used in this study are listed in table S1. Mutations were introduced into the genome of P. aeruginosa by allelic exchange. Briefly, flanks defining the mutation were amplified from the P. aeruginosa genome and cloned into plasmid pEXG2 by Gibson cloning. The primers used for the amplifications were designed using AmplifX 2.1.1 by Nicolas Jullien (Aix-Marseille Univ, CNRS, INP, Inst Neurophysiopathol, Marseille, France - https://inp.univ-amu.fr/en/amplifx-manage-test-and-design-your-primers-for-pcr) and are noted in table S2. Plasmid pEXG2 harboring the mutation construct was transformed into E. coli strain SM10 and mated at 37°C into P. aeruginosa by mixing the donor and recipient strains on an LB plate. The mating mixture was then plated on an LB plate with 30µg/mL gentamicin and 5µg/mL triclosan and grown overnight at 37°C (selecting against the E. coli donor strain). Cointegrates were restruck and subsequently grown in LB lacking salt until the culture was barely turbid. The bacteria were then plated on a sucrose plate (5% sucrose, 10g/L tryptone, 5g/L yeast extract) and incubated overnight at 30°C. Sucrose resistant colonies were tested for gentamicin sensitivity and the presence of the mutation was tested by PCR.

Complementing plasmids were generated by amplifying the open reading frame and using Gibson assembly (53) to clone it into pPSV37. The plasmids were then transformed into P. aeruginosa by electroporation.

For motility assays, individual bacterial colonies were used to inoculate motility agar plates (0.3% agar, LB plates) and incubated at 37°C for ∼8h before imaging the plate.

Bacteriophage were isolated from filter sterilized waste-water samples from one of three Northeast Ohio Regional Sewer District wastewater treatment plants (Westerly, Southerly, or Easterly Sewage Treatment plant). Phage were isolated by picking individual plaques isolated on P. aeruginosa PAO1 ΔfliF ΔpilA and testing them for their ability to form plaques on the isolation strain or wild-type PAO1. Phage genome DNA was isolated as previously described (54). Genomic DNA was sequenced by Illumina sequencing and assembled using Spades version 3.15.3 (55). Genomes were annotated using PhageScope (56). Genomes for Clew-1, Clew-3, Clew-6, and Clew-10, as well as Ocp-2, have been deposited in Genbank [accession# PQ790658.1, PQ790659.1, PQ790660.1, PQ790661.1, and PQ790662.1].

Bacteriophage were purified by CsCl gradient based on a published protocol (57). Briefly, a 500mL culture of PAO1F ΔfliF2 was grown in LB to an OD₆₀₀ of ∼0.2 and inoculated with phage Clew-1 or Ocp-2 at an MOI of 0.025. After about 4h of culture, the bacteria were pelleted (12,000 x g, 15 min, 4 °C) and the supernatant was filtered through a 0.2µM filter. The supernatant was treated with DNAse and RNAse (1µg/mL each) overnight at 4°C. The following day, the phage were pelleted by centrifugation (overnight, 7,000 x g, 4 °C), the supernatant discarded and the pellets were resuspended in 1mL of SM buffer (50mM Tris.Cl (pH 7.5), 100mM NaCl, 8mM MgSO₄) without BSA each (∼2h at 4°C). The concentrated phage prep was then spun at 12,000xg for 10 minutes to pellet remaining cell debris. At this point, 0.75g of CsCl/mL was added to the cleared supernatant and the mixture was spun for 20h at 4°C at 32000 rpm in Beckman Optima MAX-TL ultracentrifuge using an MLS-50 rotor to establish the gradient. The band with the phage was removed with a syringe and 20 gauge needle and transferred to a 3.5kDa cut-off dialysis cassette (Slide-A-Lyzer, Thermo). The phage prep was dialyzed overnight against SM, then 2x for 3h against SM, then overnight against PBS and 1x for 4h against PBS. The phage prep was tested for titer and, in the case of Clew-1, the ability to plaque on a ΔfliF2 mutant, but not wild-type PAO1F.

The negative stain experiment was done as described previously (58). Briefly, a 3 μl phage Clew-1 sample (0.1-0.5 mg/ml) was loaded onto glow-discharge carbon coated grid for 60 s at room temperature and blotted with filter paper. The grid was touched with a water droplet and then blotted with filter paper. This process was repeated twice. The grid was then touched with a drop of 0.75% uranyl formate and blotted with filter paper. A second drop of 0.75% uranyl formate was applied to touch the grid for 30 s, blotted with filter paper and then air dried before data collection. The images were taken by Tecnai T20 (FEI Company) equipped with a Gatan 4K x 4K CCD camera at 80,000 x magnification.

To test phage plating efficiency, bacterial strains were back-diluted 1:200 from overnight cultures and grown to early log phase (OD600 ∼0.3). At this point, 50µL of culture were mixed with 3mL top agar (10g/L tryptone, 5g/L yeast extract, 5g/L NaCl, 0.6% agar) and plated on an LB agar plate. Once solidified, 10-fold serial dilutions of the phage in SMB buffer (50mM Tris.Cl (pH 7.5), 100mM NaCl, 8mM MgSO₄, 0.1% bovine serum albumin). were spotted onto the agar using a multichannel pipette (3µL spots). The spots were allowed to dry and the plates incubated overnight at 37°.

P. aeruginosa strains were grown to mid-logarithmic phase in LB supplemented with 5mM MgCl2 and 0.1mM MnCl2 (LBMM) concentrated and resuspended at a concentration of 10^9 cfu/mL in LBMM. 100µL bacterial suspensions were infected at an MOI of 1 with phage Clew-1 (2µL, 5*10^10 pfu/mL) for 5 mins at 37°C, then pelleted (3’ 10k RPM), washed 2x with 1mL LB, and resuspended in 100µL LB. The infected cells were serially diluted 10x, then 10µL of diluted, infected bacteria (10^-4 for WT and ΔfliF2; 10^-1 for ΔpslC and ΔfliF2 ΔpslC) were mixed with 50µL of the mid-log PAO1F ΔfliF2 culture and mixed with 2.5mL top agar, plated on an LB plate and incubated overnight at 37·C. The following day, plaques were counted to enumerate the cell-associated bacteria (59).

Strain PAO1F or PAO1F ΔfliF were mutagenized with transposon TnFac (32), a mini-mariner transposon conferring gentamicin resistance. A pool 3*10^6 (PAO1F) or 6*10^6 (PAO1F ΔfliF) insertion mutants was grown overnight, then diluted 1:200 in fresh LB and grown to an OD600 of 0.2. At this point the bacteria were infected at an MOI of 10 with phage Clew-1 and incubated for 2h to allow infection and killing of susceptible bacteria. Bacteria from 1mL culture were then pelleted, resuspended in 100µL LB with 5mM EGTA, and plated on a 3 LB plates with 30µg/mL gentamicin. The next day, surviving bacteria that had grown up where pooled and chromosomal DNA from the input and output pools were isolated using the GenElute™ Bacterial Genomic DNA Kit (Millipore-Sigma). Library preparation followed a published protocol (60). Genomic DNA was sheared to ∼300bp using a Covaris focused ultrasonicator. The sheared DNA was repaired using the NEBNext End Repair Module (New England Biolabs) and subsequently tailed with a polly-dC tail using Terminal Transferase (New England Biolabs). Tailed chromosomal DNA fragments were amplified in two consecutive steps, using primers Mar1x and olj376 for the first round and Mar2-InSeq paired with a TdT_Index primer for the second round, based on the published protocol (60). The libraries were sequenced using an Illumina MiniSeq system using the transposon-specific primer MarSeq2. Reads with the correct Tn end sequence were mapped and tallied per site and per gene using previously described scripts ((60) and https://github.com/lg9/Tn-seq). The data (hits and # of reads for each gene) are listed for each strain and condition in Table S3.

Bacteriophage Clew-1 was isolated from 500mL of culture and purified using a CsCl gradient, following a protocol published by the Center for Phage Technology at Texas A&M University. After dialysis overnight dialysis of the phage into SM buffer, the phage was dialyzed 3 more times against PBS (2x for 3h and once overnight). The purified phage was titered by efficiency of plating analysis and labeled with a Dylight594 NHS-ester (Invigtrogen) at a concentration of 0.2mM, overnight in the dark. After labeling, the residual dye was removed by gel filtration using a Performa DTR gel filtration cartridge (EdgeBio) that had been equilibrated with PBS. The labeled phage preparation was titered to ensure that the phage concentration was unchanged and that the phage had not lost infectivity.

To assess phage attachment, wild type PAO1F, PAO1F ΔfliF2, or PAO1F ΔfliF2 ΔpslC harboring plasmid pP25-GFPo, which directs the constitutive production of GFP, were grown in LB to an OD₆₀₀ of ∼0.3-0.4, normalized to an OD₆₀₀ of 03, and 0.5mL of the culture were infected for 10 minutes at 37°C with DyLight594-labeled Clew-1 phage at an MOI of 5. At this point, the infected bacteria were fixed with 1.6% paraformaldehyde [final concentration], incubated in the dark for 10 minutes, then the remaining paraformaldehyde was quenched through the addition of 200µL of 1M glycine (10 minutes at RT). The bacteria were washed 3x with 500µL of SM buffer and resuspended in 30µL SM buffer. 4µL were spotted onto an agarose pad, covered with a coverslip and imaged using a Nikon Eclipse 90i fluorescence microscope. Images were adjusted for contrast and false-colored using the Acorn software package (Flying Meat Software), and cell-associated bacteriophage were counted in ImageJ.

Wild-type P. aeruginosa was grown for 18h in M63 minimal medium ([NH₄]₂SO₄, 2 g/l; KH₂PO₄, 13.6 g/l; FeCl₃, 0.5 mg/l, pH 7) supplemented with 0.5% Casamino acids (BD), 1 mM MgCl₂, and 0.2% glucose. Bacterial cells were removed by centrifugation, the supernatant lyophilized, and Psl isolated by affinity chromatography.

The affinity column was prepared by resuspending 0.286 g of CNBr activated Sepharose (Purchased from GE Healthcare; cat#17-0430-01) in 1 M HCl (1 mL). It was subsequently filtered and washed with 1 M HCl (60 mL) and coupling buffer (1.5 mL; 0.1 M NaHCO3, 0.5 M NaCl, pH = 9). The activated Sepharose was added to a solution of Cam-003 (61) in coupling buffer (0.5 mL; 10 mg/mL) and was incubated for two hours at room temperature. The solvent was then removed by filtration, and the beads were washed with coupling buffer (3 × 1 mL). After removal of the solvent the sepharose was incubated with blocking buffer (2 mL; 0.1 M Tris, 0.5 M NaCl, pH = 8.5) for 2h at ambient temperature. The beads were washed with wash buffer (4 mL) and coupling buffer (4 mL) for four cycles until the OD280 of the wash was <0.01. The derivatized beads were loaded onto a column and after washing with 5 column volumes of PBS-buffer (pH=7.4) the affinity column was ready to use.

Crude Psl (100 mg) was dialyzed (Thermo Scientific SnakeSkinTM Dialysis Tubing 3K MWCO) for three days and six exchanges of water and then concentrated to a final volume of 1 mL (10 mg/mL). It was loaded onto the affinity column and washed with PBS-buffer (4 mL) in order to remove all not-retained material. Next, the captured Psl was eluted with glycine buffer (4 mL; 100 mM glycine × HCl, pH=2.7). The glycine fraction was dialyzed (3K MWCO) for three days and six exchanges of water and after lyophilization, pure Psl (80 µg) was obtained.

The solution was lyophilized, and the residue was fractionated by gel permeation chromatography on a Bio-Gel P-2 column (90 × 1.5 cm), eluted with 10 mM NH₄HCO₃. The collected fractions contained different size of Psl material: dimer (two repeating units), trimer (three repeating units) and high molecular weight polysaccharide. The high molecular weight polysaccharide fraction was used in our experiments.

Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry experiments were performed using Bruker ultrafleXtreme (Bruker Daltonics) mass spectrometer. All spectra were recorded in reflector positive-ion mode and the acquisition mass range was 200–6000 Da. Samples were prepared by mixing on the target 0.5 μL sample solutions with 0.5 μL aqueous 10% 2,5-dihydroxybenzoic acid as matrix solution.

For experiments in which binding of Clew-1 to Psl in culture supernatants was tested, PAO1F ΔfliF2 or PAO1F ΔfliF2 ΔpslC were grown to mid-logarithmic phase, then the bacteria were pelleted and the supernatants filter sterilized using a 0.2µM filter. Culture supernatants were mixed with 1µL of a rabbit, anti-Psl antibody (37) as well as 10^7 pfu of phage Clew-1. The mixture was incubated on ice for 1h, then 10µL of magnetic protein A beads (BioRad), washed 2x with SMB + 0.05% Triton X-100 (SMBT) were added to the mixture and incubated for an additional 30 minutes on ice. The magnetic beads were collected, washed 3x with SMBT and resuspended in 100µl of SMBT. Presence of Clew-1 in input and output samples was determined by quantitative PCR using primers designed to amplify the tail fiber gene, gp12.

Experiments using partially purified, cell-associated Psl were carried out in SM buffer. 100µL of SM-buffer were mixed with 10^7 pfu of phage Clew-1, as well as 1µg of a partially purified, deproteinated fraction of cell-associated Psl (62) and incubated for 1h on ice. All subsequent steps were the same as for the culture supernatants, above. Samples were resuspended in 100µL SMBT before quantifying Clew-1 levels.

Experiments using affinity purified, biotinylated Psl were carried out in SM. Here too, 10^7 pfu Clew-1 were incubated with 1µg of affinity purified, biotinylated Psl. The samples were either incubated with streptavidin-coated Dynabeads (M280, Invitrogen) to precipitate the biotinylated Psl (or with magnetic protein A beads as a specificity control). Otherwise, the experiments were carried out as for the partially purified, cell-associated Psl fraction, above.

Static biofilm experiments were carried out based on a published protocol (63). P. aeruginosa PAO1F was grown to mid-logarithmic phase in LB and then diluted to an OD600 of 0.05 and used to inoculate either 5mL polystyrene tubes (1mL) or 6 wells in a polystyrene 96-well plate (150µL). The cultures were incubated overnight at 37°C in a humidified incubator with a 5% CO2 atmosphere. The following day, 1 set of biofilm samples was washed three times with PBS, for 20 minutes fixed with 95% ethanol, and subsequently air dried after removing the ethanol. The remaining biofilm samples were washed 2x with PBS and reconstituted with pre-warmed LB (1.2mL in 5mL tube biofilms, 200µL. in 96-well plates), or LB harboring either 10^9 pfu of phage Clew-1 or phage Ocp-2. The biofilm samples were again incubated overnight at 37°C in a humidified incubator with a 5% CO2 atmosphere, and subsequently washed and fixed as the control samples, above. The fixed and dried biofilms were stained with a 0.1% solution of crystal violet in water for 30 minutes, the staining solution was removed, and the biofilms were washed 2x with mili-Q water and rinsed twice with deionized water before drying the stained biofilm samples. The stained biofilms in the 5mL tubes were photographed against a white background. The stained biofilms in the 96-well plates were incubated for 20 minutes in 200µL 30% acetic acid to solubilize the crystal violet stain, which was subsequently quantified by spectrophotometry (absorbance at 590nm).

For each technical replicate (3 per condition), one 6mm natural bead (Precision Plastic Ball Company) was placed in the well of a 24-well plate, and covered with 1mL of PAO1F in LB at an OD600 of 0.05. The plate was sealed with parafilm and shaken overnight (150 rpm, 37°C). The following day all beads were washed by moving the bead to 1mL sterile PBS with a forceps. Three beads were placed individually into microcentrifuge tubes with 1mL PBS with 5mM MgCl₂, sonicated for 10 minutes in a water bath sonicator, and titered on LB plates to determine the input. The remaining beads were transferred into fresh wells of the 24-well plate with either 1mL LB or 1mL LB with 10^9 pfu Clew-1 and incubated with shaking (37°C, 150rpm) for another 2 hours. At this point, the supernatant bacteria were titered, the beads were washed 1x with PBS and placed individually into microcentrifuge tubes with 1mL PBS with 5mM MgCl₂ and sonicated and titered, as above. The bacterial titers of the supernatant and bead-associated bacteria were added to obtain the total CFU in the well and compared to the input CFU (bead titer at T0).

In some experiments, the phage treated samples were treated with a virucide before tittering (64). Here 250µL of 15% wt/vol, filter-sterilized black Tea was mixed with 730µL Milli-Q H2O and mixed with 20µL of a freshly made, 200mM FeSO₄ solution just before use (4mM FeSO4 final). The virucide was mixed with the sample, 1:1 and incubated for 10 minutes at room temperature before tittering the bacteria. However, this did not significantly alter the outcome of the experiment, and the data presented are a combination of all experiments, with and without virucide.

Biofilms were generated in 8-well chamber slides (ibidi) by inoculating the well with 200µL of mid-log PAO1F (diluted to an OD₆₀₀ 0.05) and incubating the slide overnight in a humidified incubator (5% CO₂). The next day, the staining solution was added to each well (1µL of a 1:10 dilution of each component in DMSO + 8µL LB) as well as 40 µL LB (no phage control) or 40µL LB with 10^9 pfu Clew-1. The concentrations were based on a published protocol for using LIVE/DEAD staining to test antibiotic susceptibility of biofilms (65).The slides were incubated overnight and imaged using a 60x objective on a Nikon AX R laser scanning confocal microscope. The Z-stacks were analyzed and the propidium iodide positive fraction of the total biofilm volume was determined using the BiofilmQ software package (66).

C57BL/6 mice were purchased from Jackson Laboratories. The mice were housed in pathogen free conditions in microisolator cages and were treated according to institutional guidelines following approval by the University of California IACUC.

Overnight cultures of P. aeruginosa PAO1F/pP25-GFPo were grown to log phase (OD₆₀₀ of 0.2) in LB broth, then washed and resuspended in PBS at 2.5×10⁷ bacteria/ml. 7-12 weeks old C57BL/6 mice were anesthetized with ketamine/xylazine solution, the corneal epithelium was abraded with three parallel scratches using a sterile 26-gauge needle, and 2 μL of a suspension of bacteria were added topically (approximately 5×10⁴ cfu per eye). At 24h and 48h, the mice were anesthetized and treated with 5*10^9 pfu CsCl purified phage Clew-1 in PBS, or PBS alone. At 72h the mice were euthanized, and corneas were imaged by brightfield microscopy to detect opacification, or by fluorescence microscopy to detect GFP-expressing bacteria. Fluorescent intensity images were quantified using Image J software (NIH). To determine the bacterial load, whole eyes were homogenized in PBS using a TissueLyser II (Qiagen, 30 Hz for 3 minutes), and homogenates were serially diluted plated on LB agar plates for quantification of colony forming units (CFU) by manual counting. CFU were also determined at 2h to confirm the inoculum.

Strains PAO1F, PAO1F ΔfliF2, PAO1F ΔpslC, and PAO1F ΔfliF2 ΔpslC were grown to mid-logarithmic phase in LB, then diluted to a concentration of 10^8 cfu/mL. For growth curve measurements (OD600), 3 technical replicates were set up in a 96-well plate for each strain/condition. 100µL of culture were mixed with 10µL PBS or 10µL with 10^8 pfu Clew-1 and incubated at 37°C in an Agilent Cytation 5 Imaging Plate Reader with a heated chamber and orbital rotation between OD600 measurements. OD600 readings were taken every 5 minutes.

Strains PAO1F ΔfliF2 and PAO1F ΔfliF2 ΔpslC were grown to mid-logarithmic phase (OD600 ∼0.5), the bacteria pelleted by centrifugation and the culture supernatant was sterilized using a 0.22µM syringe filter. 0.5mL supernatant samples were incubated for 1h at 37°C with or without 10^7 pfu Clew-1 and subsequently diluted three times at a 1:3 ratio. 2µL of the undiluted culture supernatants and of each dilution were spotted onto a nitrocellulose filter and allowed to air-dry. The filter was then blocked with 5% non-fat milk in TBS-T (20mM Tris.Cl, 150mM NaCl, 0.1% Tween-20) for 30 minutes, washed 2x with TBS-T and incubated with the primary anti-Psl antibody (diluted 1:3000) in TBS-T overnight at 4°C. The following day, the blot was washed 3x with TBS-T, then incubated with secondary antibody (horse-radish peroxidase conjugated goat anti-rabbit antibody, Sigma) diluted 1:10000 in TBS-T for ∼3h at room temperature. The blot was then washed 3x with TBS-T and developed using the Advansta WesternBright Sirius HRP substrate and imaged on a GE ImageQuant LAS4000 imager.

The evolutionary relationship between Clew bacteriophage and other Bruynogheviruses was carried out using the Maximum Likelihood method and JTT matrix-based model (67). The tree with the highest log likelihood is shown. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the JTT model, and then selecting the topology with superior log likelihood value. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. This analysis involved 12 amino acid sequences. There was a total of 485 positions in the final dataset. Evolutionary analyses were conducted in MEGA11(68, 69). The genome comparison between Luz24 and the Clew phages was visualized using EasyFig (70).