Pseudomonas aeruginosa PA14 biofilms produce R-bodies, extendable protein polymers with roles in host colonization and virulence

reb genes code for R-bodies: large, extendable polymers that are known for their roles in obligate endosymbioses. In the non-endosymbiotic pathogen Pseudomonas aeruginosa, reb homologues are part of a cluster found in virulent strains. Here, we demonstrate that R-bodies are produced in abundance by P. aeruginosa PA14 subpopulations during biofilm growth, identify regulators of reb gene expression, and show that reb genes are required for full colonization and virulence in host models.

genes, rebZ and fecI2, which code for a transcription factor and sigma factor, respectively, and which in a prior study were required for full virulence in a C. elegans killing assay 9 (Figure 1a). 40 Via bioinformatic analyses, we found that though reb structural gene homologues are scattered across the phylogenetic tree of the genus Pseudomonas (Supplementary Figure 2), homologues of the regulatory genes show strong conservation in strains that contain reb structural genes, both within the species P. aeruginosa (in 146/147 reb homologue-containing genomes; Supplementary Table 1) and among other pseudomonads (Figure 1a, 45 Supplementary Figure 2). Though all of these observations hinted at a potential role for Rbodies in P. aeruginosa-host associations, R-body production by P. aeruginosa biofilms and during host colonization had not been investigated.
To examine whether PA14 produces R-bodies, we applied a modified R-body enrichment 50 protocol 11 to biofilms formed by the WT and by ∆pel, a mutant that is unable to produce the major PA14 exopolysaccharide and yields biofilms that are more amenable to disruption. We performed protein mass spectrometry (MS) and scanning-electron microscopy (SEM) on the isolated SDS-insoluble fractions (Figure 1b). Products of the reb cluster were detected by MS.
Three products-PA14_27680, RebP1, and RebP2--were detected at similar levels and were 55 among the 12 most abundant proteins in our sample (Figure 1c). RebP1 and RebP2 have 40-42% and 34-40% homology, respectively, to RebB and RebA from C. taeniospiralis, which constitute the main structural components of R-bodies in that organism 12 . PA14_27680 is conserved in all pseudomonads that contain at least one reb structural gene, suggesting that it may encode another R-body structural component. One other uncharacterized protein coded for 60 by the reb gene cluster, PA14_27675, was also detected in the SDS-insoluble fractions (Supplementary File 2 and Figure 1c). Consistent with the MS results, our SEM analyses revealed the presence of R-bodies in preparations from WT PA14 biofilms (Figure 1d). We detected R-bodies in various states of extension ranging from hypercoiled (~ 450 nm long) to 4 fully extended (~ 5.5 µm long) (Figure 1d, e). These structures were not observed in a deletion 65 mutant lacking the rebP1P2 operon (∆rebP1P2) (Supplementary Figure 3). Taken together, these data indicate that R-bodies are produced by PA14 biofilms.
Based on the conservation of rebZ and fecI2 in strains containing R-body structural gene homologues (Figure 1a; Supplementary File 1; Supplementary Figure 1), we reasoned that 70 one or both of these genes control expression of rebP1P2. To test this, we generated constructs that report expression from the rebP1P2 promoter (PrebP1) as red (mScarlet) or green (GFP) fluorescence and inserted these into a neutral site on the chromosome in WT PA14 and regulatory gene mutants. We observed PrebP1 activity in WT colony biofilms that was dependent on the presence of both rebZ and fecI2 (Figure 1f biofilm development leads to the formation of steep chemical gradients that can affect gene expression patterns along the z-axis 13 , we wanted to visualize PrebP1 activity across depth. We employed a thin sectioning protocol to prepare vertical sections taken from the center region of three-day-old colony biofilms 14 . Confocal microscopy of PrebP1-mScarlet biofilm sections revealed that, strikingly, cells expressing rebP1P2 are present in thin striations such that both 80 bright and dark cells are situated at the same depth. For comparison, we prepared sections of biofilms in which gfp expression was driven by a constitutive promoter (PPA1/04/03-gfp) and observed fluorescence that is more consistent between cells at the same depth (Supplementary Figure 3). Our results suggest that PrebP1-driven expression is stochastic and that expression status is heritable during vertical growth in the the biofilm. 85 The fact that homologues of the C. taeniospiralis R-body structural genes are found in diverse free-living bacteria suggests that their roles extend beyond endosymbiotic interactions 7 Figure 1). P. aeruginosa forms biofilms and acts as an extracellular pathogen during colonization of a range of hosts including plant and nematode models 15,16 . To examine 90 5 whether rebP1P2 is expressed during interactions with hosts, we employed the PrebP1-mScarlet reporter strain in Arabidopsis thaliana seedling and Caenorhabditis elegans gut colonization assays. In both hosts, as in biofilms, rebP1P2 expression patterns were distinct from those exhibited by a promoter driving constitutive expression of mScarlet (PPA1/04/03-mScarlet). While constitutive mScarlet production was uniform and fuzzy across colonized regions of hosts 95 (Figure 2a, b), rebP1P2 expression was punctate and restricted to a subset of PA14 cells (Figure 2a, b). To rule out the possibility that constitutive mScarlet production represented nonspecific, background fluorescence, we expressed mScarlet without a promoter (MCS-mScarlet) in these hosts. Indeed, promoterless mScarlet expression was not observed in either A. thaliana or C. elegans (Figure 2a, b). The distinct rebP1P2 expression patterns we observed in the 100 colony biofilm and host contexts may indicate that there is an advantage to restricting R-body production to a subset of cells in the population.

(Supplementary
In C. taeniospiralis and Azorhizobium caulinodans, R-body production has been linked to hostassociated phenotypes that confer toxicity 4, 17,18 . We hypothesized that R-bodies contribute to 105 PA14 host colonization and/or pathogenicity and therefore performed established assays to test such contributions of PA14 R-body production in our two host systems. In A. thaliana, we quantified the extent to which PA14 WT and mutant strains colonized seedlings after inoculation and five days of plant growth. For the C. elegans virulence assay, we quantified percent survival for up to four days after synchronized populations of worms were exposed to PA14 WT and 110 mutant strains. When compared to WT PA14, ∆rebP1P2 displayed attenuated host colonization In many hosts and various infection sites in humans, biofilm formation is critical to bacterial 115 colonization and virulence and requires production of matrix 19 . Matrix secretion is a defining 6 feature of biofilms; the matrix functions as the "glue" that holds bacterial cells together and can facilitate attachment to the host 20 . To test whether matrix production is affected by a defect in R-body production, we examined biofilm development using a standardized colony morphology assay, in which the dye Congo red is provided in the medium and binds to biofilm matrix. We 120 found that matrix production and biofilm development were unaltered in the ∆rebP1P2 mutant ( Figure 2c). Furthermore, ∆rebP1P2 showed no fitness disadvantage when grown in mixedstrain colony biofilms with the WT parent (Figure 2d). We conclude the A. thaliana colonization and C. elegans killing defects observed in the ∆rebP1P2 mutant are not mediated via effects on biofilm formation and that the benefit of R-body production is specific to the host-associated 125 lifestyle.
We have shown here the production of R-bodies by PA14 cells growing in biofilms and demonstrated a role for these structures during host infection. Expression of the PA14 R-body structural genes is stochastic and can be observed both in independent biofilms and during 130 association with hosts. However, an advantage to R-body production is detectable only during host colonization, suggesting that R-body function is specific to infection. Caedibacter R-body extension is triggered by low pH and occurs in phagosomes of naive paramecia, prompting Caedibacter lysis and the release of other toxins within Caedibacter cells and thereby killing the eukaryotic predator 4,5 . We speculate that PA14 R-body extension is similarly triggered in a 135 host-specific manner and that cells in the R-body-producing subpopulation act as "kamikazes" to release other bacterial products that facilitate colonization and enhance virulence. Our findings identify R-bodies as novel P. aeruginosa virulence factors and raise the possibility that R-body production contributes to the behaviors of diverse free-living bacteria during interactions with other organisms.  Construction of PA14 reporter strains. The desired promoter region (500 bp upstream of the rebP1P2 operon for PrebP1P2-mScarlet; the synthetic lac promoter for PPA1/03/04-mScarlet) was 160 amplified using the primers listed in Supplementary Table 1 and ligated into pLD3208, upstream of mScarlet. Resulting plasmids were integrated into a neutral site in the PA14 genome using biparental conjugation and the plasmid backbone was resolved out via FLP-FRT recombination.
Reporter constructs containing mScarlet were isolated using selective media and confirmed by

PCR. 165
Colony biofilm morphology assays. Ten µL of liquid subcultures were spotted on 60 mL of colony morphology medium (1% tryptone, 1% agar, 40 µg/mL Congo red dye, 20 µg/mL 8 Coomassie blue dye) and incubated for up to five days in the dark at 25˚C with >90% humidity (Percival CU-22L) and imaged daily with an Expression 11000XL scanner (Epson). Images 170 shown are representative of at least six biological replicates.
Competition assays. Fluorescent, mScarlet-expressing and non-fluorescent cells were mixed together in a 1:1 ratio from subcultures and grown as mixed-strain biofilms as described above.
After three days, biofilms were resuspended in 1 mL of 1% tryptone and homogenized in a bead 175 mill homogenizer (Omni Bead Ruptor 12) for 99 s on the "high" setting. Serial dilutions of homogenized cells were plated onto 1% tryptone + 1.5% agar plates for colony forming unit (CFU) determination. Fluorescent CFUs were determined by imaging with a Typhoon FLA7000 fluorescent scanner (GE Healthcare).

180
Purification of SDS-insoluble biofilm fraction. Twenty, five-day-old PA14 ∆pelB-G colony biofilms were resuspended in 80 mL of sterile phosphate-buffered saline (PBS) and sonicated for 2 x 10 s on ice with the microtip of a Sonifier 250 (Branson). SDS and β-mercaptoethanol were added to final concentrations of 2% and 5%, respectively. After nutating on a Nutating Mixer (Fisher Scientific) at room temperature (RT; ~ 25˚C) for 60 min, the sample was aliquoted 185 into 1.5-mL microfuge tubes and the insoluble fraction was spun down at 16,873 x g for 5 min.
The supernatant was discarded and pellets were pooled together.

Mass spectrometry of SDS-insoluble proteins. The SDS-insoluble pellet was washed three
times with Optima water (Fisher Scientific), solubilized in 100 µL of 98% formic acid at room 190 temperature for 1 h. spin-vacuum dried for 1.5 h, washed with 50% methanol and water, and then dissolved in 2% RapiGest (Waters Corp.) with 6 mM DTT. Samples were then sonicated, boiled, cooled, cysteines reduced and alkylated and proteins digested with trypsin as described previously 23 . Liquid Chromatography mass/ spectrometry (120 min runs) was performed with the Synapt G2 (Waters Corp.) in positive resolution/ion mobility mode with proteins identified 195 with ProteinLynx Global Server (PLGS) as described previously 24 . Proteins were also identified and semi-quantitatively measured with a Q Exactive HF (Orbitrap, Thermo Scientific) with Mascot V. 2.5 as described previously 25 . Identification of the most abundant proteins in these preparations by the Q Exactive HF were confirmed by the qualitative orthogonal method (Synapt G2 ion mobility/PLGS analysis). The reference proteome UP000000653 for Pseudomonas 200 aeruginosa strain PA14 from UniProt (Release 2015_10, 11/10/2015) was used for all database searches. All raw MS data files will be publicly available at the MassIVE data repository Fluorescence visualization in P. aeruginosa colony biofilms. For whole-colony fluorescence imaging, colony biofilms were grown as described for morphology assays but on growth medium without dyes. After three days, bright field images and fluorescence images were visualized with a Zeiss Axio Zoom.V16 fluorescence stereo zoom microscope. Thin sections of colony biofilms 215 (n ≥ 3 biological replicates) were prepared as described previously 14 . Differential interference contrast (DIC) and fluorescent confocal images were captured using an LSM800 confocal microscope (Zeiss). homogenized in 1 mL of PBS in a bead mill homogenizer on the "high" setting for 99 sec. Serial dilutions of homogenized tissue were plated onto LB agar plates containing 4 µg/ml tetracycline and grown overnight at 37°C to select for P. aeruginosa, and CFU counts were quantified.

A. thaliana colonization assays. Arabidopsis thaliana ecotype Columbia (Col
C. elegans pathogenicity assays. Ten µL of overnight PA14 cultures were spotted onto NGM 235 agar plates 29 and incubated at 24 h at 37˚C, followed by 48 h at 25˚C. Thirty to 35 larval stage 4 (L4) worms were picked onto the PA14-seeded plates and incubated at 25˚C. For visualization of PA14-infected worms, worms were exposed to PA14 for three days, immobilized in 10 mM levamisole in water, mounted on a 2% agarose pad on a glass slide, and imaged at 40x on a Zeiss Axio Imager D1 epifluorescence microscope with an AxioCam Mrm. For pathogenicity killing 240 assays, which were modified from Tan 1999, live/dead worms were counted for up to four days after plating onto PA14-seeded plates. unc-44(e362) worms, which exhibit body movement deficits, were used instead of wild-type to prevent worms from crawling off the plates. Supplementary Figures 1 and 2   The role of gacA in C. elegans pathogenicity has been described previously 30 and gacA::Tn therefore serves as a control for defective killing in this assay. Error bars denote standard deviation and p-values were calculated using unpaired, two-tailed t-tests. n ≥ 4 biological replicates from three experiments.