Relative abundance of Dr fimbriae in the model strain AAEC191A Dr⁺ and the clinical strain IH11128

To validate AAEC191A Dr⁺ as a model for investigating the physicochemical properties of the Dr envelope, the relative abundance of Dr fimbriae was first compared with that of the clinical strain IH11128. Because plasmid pCC90, responsible for Dr fimbriae production in the model strain, carries the dra operon without the regulatory region responsible for phase variation [52], this comparison was necessary to assess the physiological relevance of the model.

Overnight cultures of AAEC191A Dr⁺ and IH11128 grown in liquid M63 medium supplemented with 1% casamino acids and 0.2% glycerol were analyzed after normalization to identical optical density (OD₆₀₀). Western blot analysis of isolated fimbrial fractions showed that IH11128 produced DraE, corresponding to the Dr fimbrial monomer, at 20  ± 5% of the level observed in AAEC191A Dr⁺ (S1A Fig). Immunofluorescence analysis of the same cultures showed that all AAEC191A Dr⁺ cells displayed Dr fimbriae-associated fluorescence, whereas only 16 ± 4% of IH11128 cells were fluorescent (S1B Fig). After correction for the fraction of fimbriae-producing cells, Dr fimbriae production in AAEC191A Dr⁺ was found to be comparable to that estimated for the fimbriae-positive subpopulation of IH11128.

Production of Dr envelope alters the macroscopic properties of bacterial cultures

The presence of Dr fimbriae was readily visible in bacterial cultures, as reflected by the clear macroscopic differences between Dr⁺ and Dr⁻ strains. The influence of Dr fimbriae on bacterial surface properties was evident during centrifugation of overnight cultures. Control Dr⁻ strains (AAEC191A, BL21(DE3), and JM101) grown to OD₆₀₀ ≈ 5 produced a clear supernatant after centrifugation, whereas the corresponding Dr⁺ cultures remained visibly turbid (OD₆₀₀ ≈ 0.5). In addition, pellets formed by Dr⁻ strains were compact and stable, whereas pellets obtained from Dr⁺ strains were loose and rapidly dispersed after supernatant removal (Fig 1A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Macroscopic manifestations of Dr fimbrial envelope formation in Escherichia coli.

(A) Representative macroscopic appearance of overnight cultures of E. coli AAEC191A, BL21(DE3) and JM101 carrying either the control plasmid pACYC177 (Dr⁻) or the Dr fimbriae expression plasmid pCC90 (Dr⁺) after centrifugation for 15 min at 1240 × g (upper panel), followed by decanting of the supernatant and 5 min incubation in an inverted position (lower panel), demonstrating decreased cohesion of Dr⁺ bacterial pellets. (B) Apparent viscosity changes monitored as the relationship between optical density (OD₆₀₀) and time for AAEC191A Dr⁺ and Dr⁻ cultures measured in a microfluidic system using 2.16 µl droplets flowing through channels with a cross-sectional area of 600 × 600 µm. The experiment was repeated five times; averaged values are shown, and twice the mean standard deviation is represented by the width of each curve. (C) Filtration time of AAEC191A Dr⁺ and Dr⁻ suspensions through a 0.45 µm cellulose filter in 100 mM potassium nitrate as a function of OD₆₀₀. Dr⁺ suspensions with 1.3-fold higher OD₆₀₀ were used to obtain comparable cell numbers to Dr⁻ suspensions. The lower point indicates the appearance of the first filter area free of suspension, whereas the upper point indicates complete disappearance of liquid from the filter surface. Data represent mean ± SD (n = 3), statistical significance was evaluated using non-parametric Kruskal-Wallis test, showing a significant effect for Dr⁺ filtration (p < 0.001), whereas no significant differences were observed for Dr⁻ filtration (p > 0.05). (D) Representative scanning electron microscopy images of bacterial layers formed after filtration of Dr⁺ and Dr⁻ suspensions with OD₆₀₀ values of 1.7 and 1.3, respectively. (E) Western blot analysis of fimbrial fractions isolated from Dr⁺ cells used in the filtration assay, with detection of DraE protein. Lane Dr25: fimbrial fraction incubated in Laemmli buffer at 25 °C before electrophoresis; lane Dr100: the same fraction incubated at 100 °C; lanes Filt1, Filt2, and Filt3: 100-fold concentrated filtrates obtained after filtration of AAEC191A Dr⁺ suspensions with OD₆₀₀ values of 1.7, 0.85, and 0.43, respectively.

https://doi.org/10.1371/journal.ppat.1014588.g001

To quantify this effect, culture viscosity was measured using a non-invasive microfluidic method allowing simultaneous determination of viscosity and optical density in a moving microdroplet. Continuous measurements performed for AAEC191A Dr⁻ and AAEC191A Dr⁺ cultures grown in Luria-Bertani (LB) broth for 24 h showed similar viscosities during early growth (1–2 cP), followed by a marked increase in the Dr⁺ culture after ~3 h (Fig 1B). Viscosity reached ~22 cP at 8 h and ~35–38 cP at 12 h in the Dr⁺ culture, compared with ~8 cP and ~12–15 cP, respectively, in the Dr⁻ control. From ~6 h onward, AAEC191A Dr⁺ cultures also showed consistently higher OD values, with a maximal difference of ~0.3 OD units at ~15 h (Fig 1B). However, direct cell counting at OD₆₀₀ = 0.65 showed 522,746 ± 3,400 cells μl⁻¹ in the Dr⁻ culture and 393,250 ± 3,200 cells μl⁻¹ in the Dr⁺ culture (n = 3, unpaired Student’s t-test, p < 0.001), indicating stronger light scattering by Dr⁺ cells.

The altered physical behavior of Dr⁺ suspensions was further confirmed in a vacuum filtration assay. Suspensions of AAEC191A Dr⁻ cells in 100 mM KNO₃ (OD₆₀₀ 0.325, 0.650, and 1.3) passed through a cellulose membrane within 5–13 ± 2 s, whereas corresponding Dr⁺ suspensions (OD₆₀₀ 0.43, 0.85, and 1.7) required substantially longer filtration times (Fig 1C). The first dry point on the membrane appeared after 12 ± 4, 106 ± 18, and 462 ± 68 s, respectively, whereas complete disappearance of the liquid required 29 ± 5, 173 ± 26, and 593 ± 69 s. Comparable results were obtained in PBS at pH 5.7, 7.0, and 8.0, indicating no significant effect of pH on filtration behavior (S2 Fig). Filtrate analysis confirmed the absence of bacterial cells or released fimbriae, while both strains formed confluent layers on the membrane surface (Figs 1D,E and S3). Thus, prolonged filtration of Dr⁺ suspensions was associated with the presence of Dr fimbriae rather than release of extracellular material.

TEM and LC-MS analysis of Dr envelope encoded by the model E. coli AAEC191A Dr⁺

Macroscopic observations and rheological measurements indicate that the Dr fimbrial envelope profoundly alters the physical behavior and surface properties of bacterial populations. To investigate the structural basis of these effects, we used transmission electron microscopy (TEM) together with a computational model informed by liquid chromatography–mass spectrometry (LC-MS) analysis. This framework provides a basis for interpreting the physicochemical properties and biological activities of Dr-fimbriated bacteria described in the following sections.

Available micrographs of Dr-fimbriated bacteria typically show only small regions of the cell surface near the membrane where the fimbriae are anchored [15]. Consequently, these images do not allow estimation of the overall volume of the Dr envelope, which is critical for understanding its contribution to bacterial surface properties. Similar to other proteinaceous adhesive systems, Dr fimbriae are highly hydrophilic and readily undergo structural disruption during TEM preparation and imaging. In TEM of the AAEC191A Dr⁺ strain, negative staining conditions were identified that stabilized the envelope, allowing visualization of its overall dimensions. Under these conditions, bacterial cells were surrounded by a pronounced electron-dense layer, corresponding to an envelope ~200–300 nm thick. Envelope density prevented resolution of individual filaments, but allowed reliable estimation of its relative size compared to the cell (S4A Fig). Bundled fimbrial structures, likely preparation artifacts, appeared only in partially disrupted envelopes (S4B Fig). TEM of AAEC191A Dr⁻ cells showed neither the electron-dense layer nor fibrillar bundles (S4C Fig).

Computational model of the Dr fimbrial envelope produced by the model strain AAEC191A Dr⁺

The experimentally determined stoichiometry of DraE and DraC proteins provided the basis for constructing a structural model of the Dr envelope. Assuming standard E. coli dimensions (radius ~500 nm and total length ~2000 nm) and that all DraC usher proteins participate in fimbrial assembly, the measured DraC copy number corresponds to approximately one fimbrial polymer per ~63 × 63 nm patch of the bacterial surface. Based on the DraE-to-DraC ratio, each polymer was estimated to contain approximately 140–150 DraE subunits.

To generate realistic polymer geometry, molecular dynamics simulations were first performed for short DraE heptamers in explicit solvent. Donor-strand complemented DraE dimers extracted from these simulations were then used as building blocks to assemble ensembles of ~150-subunit fimbrial polymers (with very low model sensitivity to the assumed variance in polymer length, see S6 Fig). These polymers were built as XY-periodic 5 × 5 surface patches using Monte Carlo sampling to reproduce plausible spatial organization and steric constraints between neighboring fimbriae. The resulting patches were then geometrically transformed into cylindrical segments corresponding to the surface curvature of a bacterial cell, yielding a realistic representation of the Dr envelope (Fig 2A; S1 Video).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Model and structural properties of the Dr bacterial envelope.

(A) Cylindrical sections of the envelope built upon a bacterial cell, assuming a standard E. coli radius of 500 nm, at different values of the σ parameter controlling the tendency of the fimbriae to grow and extend outwards; the color scale ranges radially from 500- to 900-nm distance from the cylinder axis (0–400 nm from the outer membrane). At σ = 1, all downward moves (i.e., adding a DraE unit whose center of mass is closer to the outer membrane than the previous one) are rejected; at σ = 0.5, a downward move would be rejected with a 50% chance. In central and rightmost panels, the corresponding side and top views of the 5x5 patch are shown, with individual fimbriae colored differently to show the degree of mixing and tangling. An atomistic structure is shown in the inset to highlight the scale of the assembly. (B) Protein density profiles as a function of the distance away from the bacterial outer membrane for different values of σ; the densities are normalized to integrate to 1. Excess fimbrial surface per nm is calculated as the fraction of the original outer membrane surface provided by DraE proteins contained in a 1-nm thick slice of the fimbrial envelope; this integrates to 2.7, the ratio of total fimbrial surface to the outer membrane surface area.

https://doi.org/10.1371/journal.ppat.1014588.g002

To explore how the spatial organization of fimbrial polymers may vary within the envelope, an “effective solubility” parameter σ was introduced into the model, with values of 0.5, 0.67, 0.84, and 1.0 (Fig 2A). This parameter controls the tendency of fimbrial polymers to extend away from the outer membrane and thereby imposes different packing dynamics of DraE subunits as a function of distance from the cell surface (Fig 2A,B). Physically, σ reflects both the electrostatic repulsion between negatively charged DraE subunits and the hydrophilic tendency to swell and avoid close contact between fimbriae, which together determine the degree to which fimbrial polymers extend into the surrounding medium. For σ = 0.5, the peak density of fimbrial proteins occurs within ca. 50 nm from the membrane surface, locally creating a thick mesh whose local adsorption-capable surface area more than doubles that of the outer membrane. Under these conditions, fimbrial polymers are densely packed, forming a compact envelope primarily localized within ~100 nm of the membrane surface. In contrast, for σ = 1.0 the density of DraE proteins becomes essentially constant over a thickness of 250 nm, reduced to one third to half of the thick mesh described above. In this configuration, the Dr envelope adopts a more extended, porous and relatively uniform structure. Intermediate σ values (0.67 and 0.84) produce correspondingly intermediate packing regimes. To verify that the obtained models accurately represent real structures, we compared the projected protein densities corresponding to different values of σ with two representative density slices of a TEM micrograph and found excellent agreement with the model corresponding to σ = 0.67 (S7 Fig). In this configuration, the Dr layer increases the effective cell diameter by approximately 40–60% relative to Dr-negative cells. Nevertheless, we speculate that alternative configurations might correspond to dynamic states of the Dr fimbrial envelope arising under varying physiological or environmental conditions experienced by the bacterial cell that might modulate effective forces between individual fimbriae.

Physicochemical properties of the Dr fimbrial envelope

The Dr envelope of our AAEC191A Dr⁺ model consists of approximately 220,000–240,000 DraE protein subunits. Assuming an average accessible surface area (ASA) of 77 nm² per DraE within the polymer, this corresponds to a total envelope surface area of ~18.5 µm², compared to ~6 µm² for a non-fimbriated cell. This surface is formed by flexible fimbrial fibers ~2 nm in diameter and up to ~600 nm long. Consequently, the chemical properties of the envelope surface are largely determined by DraE.

Analysis of the solvent-accessible surface area of DraE at the level of individual amino acid residues shows similar fractions of apolar (absolute atomic charge |Q| < 0.25) and polar atoms (|Q| > 0.25) (Fig 3A). This distribution reflects the high abundance of surface-exposed Thr and Lys residues, whose side chains contain extended apolar fragments. As a result, the Dr envelope is amphipathic, with a large and chemically heterogeneous surface. Understanding its physicochemical properties therefore requires multiple complementary approaches, each revealing different aspects of its surface organization and contribution to bacterial behavior.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Dr fimbrial envelope enhances hydrophobic surface properties of Escherichia coli.

(A) Surface hydrophobicity and hydrophilicity distribution of the DraE subunit calculated from solvent-accessible surface area (SASA) within the polymeric Dr fimbrial structure at atomistic resolution. Residues were classified as hydrophobic or hydrophilic as indicated in the left panel. Atoms were classified as polar or apolar using an absolute atomic charge cut-off of 0.25. The total solvent-accessible surface area per DraE subunit was 76.76 nm². The inset shows zeta potential values of E. coli AAEC191A Dr⁺ and Dr⁻ cells measured in phosphate-buffered saline (PBS) at pH 5.7, 7.0, and 8.0. Data are presented as mean ± SD (n = 3). Statistical significance was evaluated using Welch’s corrected Student’s t-test. For all Dr⁺/Dr⁻ comparisons at each pH value, *** indicates p < 0.001. (B) Microbial adhesion to hydrocarbons (MATH) assay performed with AAEC191A Dr⁺ and Dr⁻ suspensions using xylene or hexadecane as the organic phase. The percentage of bacteria transferred from the aqueous phase to the organic phase was calculated relative to OD₅₇₀ values of control suspensions without organic solvent. Bacterial suspensions were normalized to equivalent cell numbers in PBS at pH 7.0. Data represent mean ± SD (n = 3), statistical significance was evaluated using non-parametric Kruskal-Wallis test (** p < 0.01, *** p < 0.001). (C) Representative phase-contrast images of the interface between xylene (marked X) and the aqueous phase containing AAEC191A Dr⁺ or Dr⁻ cells. The same bacterial suspensions as in panel (B) were used. Images were acquired 15 min after phase equilibration.

https://doi.org/10.1371/journal.ppat.1014588.g003

Probing the ultrastructure of the model Dr fimbrial envelope by glutaraldehyde crosslinking.

To investigate the ultrastructural organization and packing density of the Dr fimbrial envelope, we assessed intercellular crosslinking of Dr⁺ bacteria with glutaraldehyde, which reacts with ε-amino groups of lysine to form covalent bridges. According to our model, each AAEC191A Dr⁺ cell contains ~240,000 DraE subunits assembled into ~1,600 fimbriae, with each DraE carrying eight lysines, providing nearly two million potential crosslinking sites. This exceptionally high density of reactive groups is expected to favor intermolecular crosslinking between neighboring cells. To test this, bacterial suspensions were incubated with a gradient of glutaraldehyde concentrations (0.01–2%) and the formation of aggregates was monitored.

Dr⁺ cells formed aggregates across a broad concentration range, down to 0.1%, with maximal aggregation at 0.5%. Higher concentrations (1–2%) reduced aggregate formation, likely due to intrafimbrial or interfimbrial crosslinking within individual cells. Aggregate sizes ranged from small clusters to massive structures up to ~1200 µm, with dense packing observed by phase-contrast microscopy (Fig 4). Aggregate growth was enhanced by the increasing total surface area, providing more sites for intermolecular crosslinking, a “surface amplification” effect. In contrast, Dr⁻ cells showed no detectable aggregation (Fig 4).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Glutaraldehyde-induced crosslinking reveals the extended reactive architecture of the Dr fimbrial envelope.

Representative phase-contrast images of E. coli AAEC191A Dr⁺ and Dr⁻ cells suspended in PBS (pH 7.5, OD₆₀₀ = 1.0) and incubated for 15 min at 4 °C with increasing concentrations of glutaraldehyde (0.01%, 0.1%, 0.25%, 0.5%, 1%, and 2%). In the Dr⁺ column, the left image series shows the overall extent and size distribution of bacterial aggregates formed at each glutaraldehyde concentration (scale bar, 200 µm), whereas the right image series presents higher-magnification views of aggregate ultrastructure and local packing density (scale bar, 50 µm). In the Dr⁻ column, only high-magnification images are shown (scale bar, 50 µm), as no detectable aggregate formation occurred at any glutaraldehyde concentration tested.

https://doi.org/10.1371/journal.ppat.1014588.g004

These findings highlight that the Dr fimbrial envelope forms an extended, highly reactive surface layer whose three-dimensional architecture is critical for efficient intercellular crosslinking, revealing the remarkable spatial reach and connectivity of the fimbrial network surrounding the bacterial cell.

Adhesion of the model Dr-fimbriated bacteria to abiotic surfaces under static conditions

Given that the Dr fimbrial envelope forms an amphipathic and extended surface layer capable of nonspecific interactions, we examined adhesion of AAEC191A Dr⁺ bacteria to abiotic surfaces under static conditions. Glass and polystyrene, with water CAs of ~25° and ~85°, served as hydrophilic and hydrophobic substrates. Bacterial suspensions (OD₆₀₀ = 0.7) in PBS at pH 5.7, 7.0, or 8.0 were deposited onto these surfaces, incubated, and non-adherent cells were gently washed off. Surface-associated cells were quantified by phase-contrast microscopy. Dr⁺ bacteria adhered strongly to both surfaces, with ~80–110 cells/100 µm² on glass and ~105 cells/100 µm² on polystyrene. Adhesion to polystyrene was pH-independent, whereas adhesion to glass decreased by ~30% at pH 7.0 and 8.0 compared to pH 5.7. In contrast, Dr⁻ cells showed minimal adhesion, ~ 5 cells/100 µm² on glass and ~8 cells/100 µm² on polystyrene, independent of pH (Fig 5).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Dr fimbrial envelope promotes strong adhesion of Escherichia coli to hydrophilic and hydrophobic abiotic surfaces under static conditions.

Adhesion of AAEC191A Dr⁺ and Dr⁻ cells suspended in phosphate-buffered saline (PBS) at pH 5.7, 7.0, or 8.0 (OD₆₀₀ = 0.7) to glass and polystyrene surfaces under static conditions. The graphs show the number of adherent bacteria quantified after 15 min incubation followed by gentle washing with the corresponding buffer. Adhesion is expressed as the number of surface-associated cells per 100 µm². Data represent mean ± SD (n = 3), statistical significance was evaluated using non-parametric Kruskal-Wallis test (*** p < 0.001). Representative phase-contrast images show bacterial adhesion patterns on glass and polystyrene. Scale bars, 20 µm.

https://doi.org/10.1371/journal.ppat.1014588.g005

Stability of the model Dr-fimbriated bacterial adhesion to glass and polystyrene under receding-meniscus capillary forces

To estimate the strength of bacterial adhesion to abiotic surfaces, we examined the susceptibility of surface-attached cells to detachment by a receding liquid meniscus. Suspensions of AAEC191A Dr⁺ and Dr⁻ bacteria in PBS (pH 7.0, OD₆₀₀ = 0.5) were deposited onto glass and polystyrene surfaces and subsequently covered with a coverslip, forming a liquid layer approximately 10 ± 2 µm thick. Controlled evaporation of water at the edge of the coverslip induced a gradual recession of the liquid meniscus across the surface. Bacteria adhered to the substrate were thus exposed to forces generated by the moving meniscus, which effectively acted as a mechanical scraper removing weakly attached cells.

On glass, the meniscus detached 98 ± 2% of Dr⁻ cells, moving at ~14 ± 2 µm s⁻¹. In contrast, 67 ± 3% of Dr⁺ cells remained attached, and meniscus motion became discontinuous, slowing to ~3.9 ± 2 µm s⁻¹, with pronounced slowing observed where surface-bound bacteria were present (Fig 6A,C and S2 Video). Different behavior was observed on the hydrophobic polystyrene surface. Dr⁻ bacteria were completely removed by the receding meniscus, which moved at a velocity of approximately 7 ± 1 µm s⁻¹. For Dr⁺ cells, the capillary force also detached essentially 100% of bacteria from the polystyrene surface, with the meniscus moving uniformly at approximately 3.4 ± 1.5 µm s⁻¹. However, detached Dr⁺ bacteria were immediately transported within the meniscus and accumulated at the wetting arm contacting the upper glass coverslip, where they subsequently adhered with high efficiency. As a consequence, a pronounced accumulation of bacteria was observed on the glass surface behind the receding meniscus in regions where the liquid film had already evaporated (Fig 6B,D and S3 Video).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Dr fimbrial envelope stabilizes bacterial adhesion to glass during exposure to receding-meniscus forces.

(A, B) Schematic representation of bacterial behavior during passage of a receding liquid meniscus on glass (A) and untreated polystyrene (B) surfaces. Region (a) indicates the area exposed to air after meniscus passage, region (b) the liquid-filled region ahead of the moving meniscus, and (m) the meniscus position. Black arrows indicate the direction of meniscus movement. (C, D) Representative phase-contrast images showing meniscus-driven displacement of E. coli AAEC191A Dr⁺ and Dr⁻ cells on glass (C) and untreated polystyrene (D). Bacterial suspensions in phosphate-buffered saline (PBS, pH 7.0, OD₆₀₀ = 0.5) were confined beneath a coverslip, forming a liquid layer of 10 ± 2 µm thickness. Images were recorded during evaporation-induced meniscus recession at 25 °C. Time after onset of recording is indicated in each panel. Scale bars, 50 µm. Meniscus motion was recorded at 5 frames per second for 1 min.

https://doi.org/10.1371/journal.ppat.1014588.g006

Similar results were obtained for bacterial suspensions prepared in PBS buffers at pH 5.7 and 8.0. Together, these observations indicate that adhesion of Dr-fimbriated bacteria to the hydrophilic glass surface is substantially more stable under the applied experimental conditions than adhesion to the hydrophobic polystyrene surface. Using a physicochemical framework describing capillary forces generated during liquid recession in thin liquid films, the characteristic force acting on a bacterium with an effective radius of approximately 0.5 µm can be estimated to be on the order of several tens of nanonewtons, depending on the local geometry of the receding meniscus (see S2 Rawgel). The observation that a substantial fraction of Dr⁺ bacteria remained attached to the glass surface after passage of the meniscus indicates that the adhesion strength of Dr-fimbriated cells to glass is comparable to this force scale. The pronounced reduction in meniscus velocity observed on surfaces populated with Dr⁺ bacteria is consistent with enhanced pinning of the three-phase contact line by strongly adhered cells, which increases the energy required for the receding interface to advance across the surface.

Adhesion of the model Dr-fimbriated bacteria to glass and polystyrene under laminar shear flow

The capillary forces generated by a receding meniscus, reaching tens of nanonewtons, far exceed the hydrodynamic drag forces acting on bacteria under laminar flow, which in physiological environments such as the urinary tract are typically only several to tens of piconewtons. To assess adhesion under more physiologically relevant shear, AAEC191A Dr⁺ and Dr⁻ strains were analyzed on glass and polystyrene in a flow chamber under laminar shear stress from 0.01 to 6.6 pN µm⁻². Bacteria suspended in PBS at pH 5.7, 7.0, or 8.0 were first allowed to attach under no-flow conditions, and then subjected to gradually increasing shear while the number of surface-adhered cells was monitored (Fig 7A; see also Video 6 and Video 7 in the Zenodo repository [53]).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Influence of the Dr fimbrial envelope on bacterial adhesion to glass and polystyrene under laminar shear flow at different pH values.

(A) Cell number-normalized suspensions of E. coli AAEC191A Dr⁻ and Dr⁺ bacteria in PBS at pH 5.7, 7.0, or 8.0 were introduced into flow chambers assembled on glass or untreated polystyrene surfaces and incubated for 10 min under static conditions to allow initial adhesion. Flow was then applied in sequential 1-min intervals generating increasing shear stress (0.01–6.6 pN µm⁻²). The graphs show the number of adherent bacteria during stepwise increase of shear stress. The right panels indicate the percentage of initially attached bacteria remaining on the surface at the end of the flow experiment. (B) Accumulation of cell number-normalized Dr⁻ and Dr⁺ bacterial suspensions in PBS at pH 5.7 during 15 min of continuous flow at constant shear stress values of 0.01, 0.05, or 0.28 pN µm⁻² on glass and untreated polystyrene surfaces. Error bars in (A) and (B) represent 95% confidence intervals calculated from three independent biological experiments. Statistical significance was evaluated using non-parametric Kruskal-Wallis test (* p < 0.05, ** p < 0.01).

https://doi.org/10.1371/journal.ppat.1014588.g007

On glass at pH 5.7, Dr⁺ bacteria adhered efficiently during the no-flow phase (364 ± 40 cells), with slight additional accumulation at low shear (0.01–0.05 pN µm⁻²). Detachment was minimal up to 2.8 pN µm⁻², with 50 ± 4% remaining at 6.6 pN µm⁻². In contrast, Dr⁻ cells adhered poorly (98 ± 26), with substantial detachment already at 0.28 pN µm⁻² and only 3% remaining at 6.6 pN µm⁻². At pH 7.0 and 8.0, initial Dr⁺ adhesion was lower (98 ± 58 and 59 ± 3 cells), with no increase in bacterial number at 0.01 pN µm⁻², detachment began around 1.1 pN µm⁻², and 23 ± 8% and 31 ± 6% of cells remained at 6.6 pN µm⁻². For the non-fimbriated strain, the initial number of adhered bacteria was similar to that of Dr⁺ cells at these pH values, but detachment began already at 0.05 pN µm⁻², leaving only 0 ± 0% and 3 ± 2% of cells at the end of the experiment for pH 7.0 and 8.0, respectively. On polystyrene, initial attachment was similar for both strains across pH 5.7–8.0. Despite this, Dr⁺ bacteria resisted detachment more effectively: at 6.6 pN µm⁻², 20 ± 8%, 17 ± 4%, and 17 ± 2% of Dr⁺ cells remained at pH 5.7, 7.0, and 8.0, respectively, compared with only 2 ± 0%, 6 ± 1%, and 6 ± 2% for Dr⁻ cells.

Time-dependent accumulation under continuous flow at low shear (0.01, 0.05, 0.28 pN µm⁻², pH 5.7) showed that Dr⁺ cells on glass accumulated to 236 ± 50, 183 ± 39, and 13 ± 4 cells, whereas Dr⁻ cells reached only 25 ± 7, 8 ± 3, and 4 ± 2. On polystyrene, accumulation was observed only at the lowest shear (0.01 pN µm⁻²), with 60 ± 17 Dr⁺ and 59 ± 17 Dr⁻ cells (Fig 7B; see also Video 8 and Video 9 in the Zenodo repository [53]).

Overall, the Dr fimbrial envelope markedly enhances adhesion stability under laminar flow, particularly on hydrophilic glass, where cells also accumulate under low shear. Adhesion to glass shows clear pH dependence, whereas polystyrene binding remains largely unaffected.

Adhesion of the model Dr-fimbriated bacteria to kidney stone minerals: hydroxyapatite and calcium oxalate

The Dr fimbrial envelope forms a strongly hydrophilic surface due to the high density of polar residues exposed on DraE subunits, which is expected to promote interactions with polar mineral phases present in urinary tract stones. To model potential interactions of uropathogenic E. coli with kidney stone components, adhesion of AAEC191A Dr⁺ and Dr⁻ strains to hydroxyapatite (HA) and CaOx was examined.

Adhesion was quantified as the percentage decrease in OD₆₀₀ of the initial bacterial suspension after mixing with a defined amount of HA or CaOx (Figs 8A, 9A), and qualitatively visualized by DAPI staining of bacteria bound to mineral sediments (Figs 8B, 9B, and S8, S9). To assess binding strength, sediments with attached bacteria were washed three times with fresh buffer by intensive pipetting and reanalyzed microscopically. At pH 5.7, 87 ± 2% of Dr-fimbriated bacteria were transferred from suspension to HA, consistent with intense fluorescence of DAPI-stained cells attached to HA crystals (Fig 8). At pH 7.0 and 8.0, adhesion decreased sharply to 10 ± 4% and 1 ± 8%, respectively (Fig 8A). In contrast, the non-fimbriated strain showed only weak HA association across the tested pH range, with 17 ± 3%, 3 ± 6%, and 11 ± 8% binding at pH 5.7, 7.0, and 8.0, respectively. Washing revealed progressive loss of permanently associated Dr⁺ cells with increasing pH, whereas almost no Dr⁻ cells remained attached after washing (Figs 8B and S8).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Influence of the Dr fimbrial envelope on bacterial adhesion to hydroxyapatite as a function of pH.

(A) Percentage of E. coli AAEC191A Dr⁻ and Dr⁺ bacteria transferred from suspension to HA, calculated from the decrease in OD₆₀₀ relative to control suspensions without HA. Bacterial suspensions normalized for cell number were prepared in PBS at pH 5.7, 7.0, or 8.0. Data represent mean ± SD (n = 3), statistical analysis was performed using the non-parametric Kruskal-Wallis test. For both strains, differences between adherence values on HA at all analyzed pH conditions were highly significant (p < 0.001), except for the Dr⁻ strain between pH 5.7 and pH 8.0, where p < 0.01 was obtained. (B) Representative phase-contrast images of HA crystals and corresponding fluorescence images of DAPI-stained bacteria attached to the crystal surface at pH 5.7. Binding shows samples after DAPI staining followed by a single gentle wash to remove excess fluorophore. Detachment shows bacteria remaining associated with HA after three washing cycles combined with vigorous pipetting to assess binding strength. Corresponding images obtained at pH 7.0 and 8.0 are shown in S8 Fig.

https://doi.org/10.1371/journal.ppat.1014588.g008

Download:

• PNG
  larger image
• TIFF
  original image

Fig 9. Influence of the Dr fimbrial envelope on bacterial adhesion to calcium oxalate as a function of pH.

(A) Percentage of E. coli AAEC191A Dr⁻ and Dr⁺ bacteria transferred from suspension to CaOx monohydrate, calculated from the decrease in OD₆₀₀ relative to control suspensions without CaOx. Bacterial suspensions normalized for cell number were prepared in PBS at pH 5.7, 7.0, or 8.0. Data represent mean ± SD (n = 3), statistical analysis was performed using the non-parametric Kruskal-Wallis test. For both strains, adherence on CaOx differed significantly across all analyzed pH conditions (p < 0.001). (B) Representative phase-contrast images of CaOx crystals and corresponding fluorescence images of DAPI-stained bacteria attached to the crystal surface at pH 5.7. Binding shows samples after DAPI staining followed by a single gentle wash to remove excess fluorophore. Detachment shows bacteria remaining associated with CaOx after three washing cycles combined with vigorous pipetting to assess binding strength. Corresponding images obtained at pH 7.0 and 8.0 are shown in S9 Fig.

https://doi.org/10.1371/journal.ppat.1014588.g009

A different pattern was observed for CaOx. Dr-fimbriated bacteria showed nearly complete binding at all tested pH values (Fig 9A), although washing revealed a marked reduction in binding strength with increasing pH, with almost no Dr⁺ cells remaining attached at pH 8.0 (Figs 9B and S9). The non-fimbriated strain also interacted with CaOx, but less efficiently, with binding levels of 88 ± 3%, 31 ± 3%, and 25 ± 3% at pH 5.7, 7.0, and 8.0, respectively. In all cases, washing nearly completely removed Dr⁻ bacteria from CaOx sediments (Figs 9B and S9).

Overall, the Dr fimbrial envelope markedly enhances bacterial association with mineral phases relevant to kidney stone formation, particularly under mildly acidic conditions.

Growth of the model Dr-fimbriated and non-fimbriated bacteria in artificial urine containing hydroxyapatite and calcium oxalate

Given the strong interaction of Dr-fimbriated bacteria with HA and CaOx observed in the adhesion assays, we next examined whether the presence of these minerals influences bacterial growth in artificial urine (AU).

The experiment was performed in AU adjusted to pH 5.7. At pH 7.0 and 8.0, intensive spontaneous precipitation of various crystalline phases, mainly carbonates and phosphates, was observed, which made it difficult to attribute the observed effects of bacterial growth to specific mineral types. In contrast, at pH 5.7, also within the physiological range of human urine, the spontaneous formation of crystals was strongly reduced. This allowed the establishment of a controlled model of kidney stone components by supplementing AU with defined crystals of CaOx or HA. As a potential carbon source, AU was supplemented with 0.2% L-lactate, a metabolite commonly present in the urinary tract [6]. During 24 h of cultivation under these conditions, the optical density of the bacterial cultures increased from an initial OD₆₀₀ of 0.025 to approximately 0.1.

After 2 h of cultivation, approximately 80% of Dr-fimbriated bacteria were associated with CaOx crystals. The binding exhibited a highly specific spatial pattern dominated by interactions occurring through the bacterial cell pole, resulting in a characteristic radial arrangement of bacteria surrounding the central CaOx crystal. The number of attached Dr⁺ cells was sufficiently large that their cumulative motility caused visible oscillations and occasional displacement of CaOx crystals along the bottom surface of the cultivation chamber (S4 Video). In contrast, the Dr⁻ strain did not display any specific interactions with CaOx and remained uniformly distributed throughout the medium. Interestingly, Dr-fimbriated bacteria did not exhibit detectable interactions with HA crystals under these conditions, similarly to the non-fimbriated strain (S5 Video).

After 24 h of cultivation, bacteria were stained with DAPI, then planktonic cells were removed by washing, and those associated with the polystyrene bottom surface and with CaOx or HA crystals were analyzed using fluorescence and phase-contrast microscopy. In cultures supplemented with CaOx, very large aggregates of Dr-fimbriated bacteria associated with the crystal surfaces were observed (Figs 10 and S10). In contrast, the control Dr⁻ strain did not display an increased level of association with CaOx compared with the number of cells attached to the polystyrene surface of the culture vessel. Furthermore, for both strains grown in AU in the presence or absence of CaOx or HA crystals, no bacterial clusters indicative of biofilm formation were detected on the bottoms of the culture wells (Fig 10).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 10. Preferential accumulation of Dr⁺ bacteria on calcium oxalate during growth in artificial urine.

Representative phase-contrast (PC), fluorescence (DAPI), and merged (PC+DAPI) images showing E. coli AAEC191A Dr⁺ and Dr⁻ bacteria after 24 h growth in AU (pH 5.7), with control samples consisting of AU alone (without HA or CaOx), and experimental samples containing HA or CaOx at a final concentration of 0.05% (w/v). Overnight cultures were diluted in AU to OD₆₀₀ = 0.025 and incubated statically at 37 °C in 24-well polystyrene plates. After incubation, bacteria associated with crystal surfaces were stained with DAPI, then planktonic cells were gently removed and crystals remaining at the bottom of the wells were rinsed once with AU. Phase-contrast images show crystal morphology, fluorescence images show DAPI-labeled bacteria, and merged images present the spatial relationship between bacteria and crystal surfaces. Scale bars, 20 µm. The graphs on the right show the percentage of bacteria transferred from suspension to HA or CaOx during incubation in AU under agitation, calculated from the decrease in OD₆₀₀ relative to control suspensions without crystals. Data represent mean ± SD, statistical significance was evaluated using non-parametric Kruskal-Wallis test (*** p < 0.001).

https://doi.org/10.1371/journal.ppat.1014588.g010

Notably, even after 24 h of cultivation, no increased association of Dr-fimbriated bacteria with HA crystals was observed relative to the background level, similarly to the behavior of the Dr⁻ strain (Figs 10 and S11). The absence of detectable binding of Dr-fimbriated bacteria to HA crystals in AU at pH 5.7 was unexpected in the context of the strong interaction observed previously in PBS at the same pH (Fig 8). To verify this observation, an additional experiment was performed in which Dr⁺ and Dr⁻ bacteria suspended in AU (pH 5.7) were mixed with HA or CaOx crystals under agitation, and the decrease in OD₆₀₀ of the supernatant was monitored as a measure of bacterial binding to the mineral phase. In mixtures containing CaOx, approximately 82 ± 3% of Dr-fimbriated bacteria and about 7 ± 2% of non-fimbriated bacteria were removed from suspension due to binding to the crystals. In contrast, no detectable binding of either bacterial strain to HA was observed in the analogous experiment (Fig 10). These results indicate that the lack of interaction between bacteria and HA in AU, compared with PBS at the same pH, is likely caused by components present in AU. Chemical analysis showed that citrate ions present in AU cause gradual decomposition of HA crystals (S12 Fig).

Together, these results confirm the strong and selective interaction of Dr-fimbriated bacteria with CaOx and indicate that components of AU can profoundly modify interactions with HA.

Phage susceptibility reveals size-dependent shielding by the model Dr fimbrial envelope

Autoaggregation of bacterial cells is typically mediated by specific proteins exposed on the cell surface and leads to visible sedimentation under stationary conditions. The E. coli BL21(DE3) and JM101 strains exhibited strong autoaggregation, resulting in a 50% reduction in optical density after approximately 2 and 6 h, respectively (S13 Fig). In contrast, AAEC191A showed no detectable sedimentation, and its optical density remained unchanged even after 24 h. Introduction of Dr fimbriae into BL21(DE3) and JM101 completely abolished autoaggregation, causing both strains to behave similarly to AAEC191A (S13 Fig). These observations indicate that the Dr envelope physically masks surface determinants responsible for cell–cell interactions and suggest that, beyond promoting adhesion to biotic and abiotic surfaces, it may also function as a steric barrier limiting access of large external particles to the bacterial surface.

To test this possibility, susceptibility to bacteriophage infection was examined using three lytic E. coli phages differing in particle size and receptor specificity: P1 vir, λ vir and T7. Among them, P1 vir represented the largest particle (250–300 nm total length, 65–80 nm capsid diameter) [54], λ vir displayed intermediate dimensions (150–170 nm total length, head diameter 55–60 nm, flexible tail 135–150 nm) [55], whereas T7 was the smallest (60–70 nm diameter) [56]. In addition, λ vir recognizes the LamB maltose porin [57], while P1 vir and T7 utilize lipopolysaccharide as surface receptors [56,58].

Efficiency of plating assays performed on AAEC191A and JM101 revealed a strong protective effect of the Dr envelope against infection by P1 vir and λ vir. The strongest reduction was observed for P1 vir, where the efficiency of plating (EOP) ratio calculated for Dr-positive relative to Dr-negative cells reached 0.02 ± 0.01 in JM101 and 0.116 ± 0.05 in AAEC191A. For λ vir, the corresponding values were 0.23 ± 0.05 and 0.10 ± 0.008, respectively. In contrast, T7 infection remained unaffected, with EOP ratios close to unity (1.1 ± 0.24 for JM101 and 1.04 ± 0.05 for AAEC191A), indicating that only larger phage particles were efficiently excluded by the Dr envelope (Figs 11A and S14, S2 Table in S1 File).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 11. Dr fimbrial envelope limits bacteriophage access to the bacterial surface.

(A) Representative plaque assays showing lysis induced by P1 vir, λ vir and T7 on lawns of E. coli AAEC191A Dr⁺ and Dr⁻ bacteria. Ten-fold serial dilutions of bacteriophage lysates (10⁰–10⁻⁸) were spotted onto LB top agar containing bacterial cells. Images show contrast-enhanced versions highlighting plaque boundaries. (B) Permeability of the fimbrial layer to model large-scale assemblies. The plots show probe clearance probability, defined as the probability to avoid clashes with the fimbrial mesh, as a function of the Z-distance from the outer membrane (OM). Probabilities were calculated for spherical probes of different diameter, as well as for 3 models of phages (T7, λ, P1), and were averaged over different probe placement and probe rotation (for non-spherical probes). Plots are shown for 4 different mesh geometries, corresponding to different values of the sigma parameter.

https://doi.org/10.1371/journal.ppat.1014588.g011

To determine whether reduced plaque formation resulted from impaired adsorption, phage binding assays were performed for the AAEC191A strain pair. The most pronounced difference was observed for P1 vir: 79.5 ± 9.2% of phage particles remained unbound in the Dr-positive strain, compared with only 2.2 ± 0.3% in the Dr-negative variant. For λ vir, the corresponding fractions of unadsorbed particles were 69 ± 8.5% and 26 ± 1.4%, respectively. In contrast, adsorption of T7 remained nearly identical in both variants, with approximately 30% of particles remaining unbound (S3 Table in S1 File). These results demonstrate that the Dr envelope strongly limits adsorption of larger phages, whereas access of T7 to the cell surface remains unaffected. Equivalent adsorption measurements were attempted for JM101 strains; however, the obtained values showed very high variability and were therefore not considered reliable for quantitative interpretation.

Because λ vir utilizes LamB as its receptor, LC–MS analysis of overnight culture pellets of AAEC191A Dr⁺ and AAEC191A Dr⁻  was performed to verify whether Dr fimbriae production alters LamB abundance (S15 Fig and S6 Table in S1 File). No differences in maltose outer membrane channel protein level were detected, indicating that reduced susceptibility to λ vir results from steric shielding rather than receptor downregulation.

As an additional control, M13 bacteriophage infection was examined using E. coli JM101, which constitutively produces F pili and is therefore naturally susceptible to M13 infection. Because M13 enters bacterial cells through F pili rather than outer membrane receptors [59], its infection pathway should remain independent of envelope-mediated shielding. Consistently, both Dr-positive and Dr-negative variants of JM101 exhibited identical susceptibility, with an EOP ratio of 1.03 ± 0.06 (S14 Fig and S2 Table in S1 File).

The structural model of the Dr envelope predicts that steric clashes with the fimbrial mesh will yield the observed sieve-like behavior for viral-size particles (larger than 20–30 nm) but remain irrelevant for low-molecular-weight compounds (Fig 11B). To verify this, susceptibility of AAEC191A Dr-positive and Dr-negative strains to five antibiotics was tested using minimum inhibitory concentration (MIC) and disc diffusion assays. Neither MIC values determined in LB broth nor inhibition zone diameters on solid medium differed significantly between both variants, indicating unrestricted penetration of small antimicrobial molecules through the Dr envelope (S16 Fig).

Together, these observations demonstrate that the Dr envelope functions as a size-selective molecular filter, efficiently restricting access of large biological particles while remaining permeable to small diffusible compounds.

Validation of Dr capsule-dependent surface properties in the clinical UPEC strain IH11128

Because AAEC191A represents a simplified model for investigating Dr capsule-dependent surface properties, we next asked whether the principal phenotypes identified in this model could also be detected in the clinical UPEC strain IH11128. For this purpose, IH11128 was compared with its isogenic Dr-deficient transposon mutant DR14.

Physicochemical properties of the Dr capsule are retained in the clinical strain.

Immunofluorescence analysis confirmed the heterogeneous expression of Dr fimbriae in IH11128. Approximately 20% of bacterial cells displayed surface-associated Dr fluorescence, whereas no fluorescent cells were detected in the DR14 mutant (Fig 12A).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 12. Validation of Dr-dependent surface properties in the clinical UPEC strain IH11128.

(A) Representative phase-contrast (PC) and fluorescence (IF Anti-Dr) images of IH11128 and the isogenic Dr-deficient mutant DR14 following immunofluorescent labeling with rabbit polyclonal anti-Dr antibodies and TRITC-conjugated secondary antibodies. Fluorescence identifies the Dr-positive subpopulation present in the phase-variable IH11128 culture, whereas no specific labeling is observed for DR14. Scale bars, 20 µm. (B) Filtration time of IH11128 and DR14 suspensions through a 0.45 µm cellulose filter in 100 mM potassium nitrate as a function of OD₆₀₀ (0.35, 0.70, and 1.40). The lower point indicates the appearance of the first filter area free of suspension, whereas the upper point indicates complete disappearance of liquid from the filter surface. Data represent mean ± SD (n = 3). Statistical significance was evaluated using the non-parametric Kruskal–Wallis test, showing a significant effect for IH11128 filtration (p < 0.001), whereas no significant differences were observed for DR14 filtration (p > 0.05). (C) Microbial adhesion to hydrocarbons (MATH) assay performed with IH11128 and DR14 suspensions using xylene as the organic phase. The percentage of bacteria transferred from the aqueous phase to the organic phase was calculated relative to OD₅₇₀ values of control suspensions without organic solvent. Bacterial suspensions were adjusted to OD₆₀₀ = 1.3 in PBS (pH 7.0). Data represent mean ± SD (n = 3), statistical significance was evaluated using the non-parametric Kruskal–Wallis test (*** p < 0.001). (D) Phase-contrast images showing antibody-mediated immunoagglutination of IH11128 and DR14 following incubation with rabbit polyclonal anti-Dr antibodies at the indicated dilutions (1:10, 1:25, 1:50, and 1:500). Large bacterial aggregates were observed only for IH11128, whereas DR14 remained uniformly dispersed at all antibody dilutions. Scale bars, 20 µm.

https://doi.org/10.1371/journal.ppat.1014588.g012

Despite the relatively small Dr-positive subpopulation, IH11128 exhibited pronounced physicochemical differences from DR14. During vacuum filtration, IH11128 suspensions passed through cellulose membranes markedly more slowly than DR14 over the entire range of optical densities tested. At OD₆₀₀ = 1.4, complete filtration required 487 ± 105 s for IH11128 compared with 31 ± 1 s for DR14, while at OD₆₀₀ = 0.7 the corresponding values were 155 ± 18 s and 15 ± 0 s, respectively (Fig 12B). The MATH assay likewise demonstrated altered surface properties. IH11128 showed significantly greater partitioning into xylene (12.9 ± 0.6%) than DR14 (2.7 ± 0.3%), whereas neither strain partitioned detectably into hexadecane (Fig 12C). Although the increase was smaller than that observed for the constitutively expressing AAEC191A Dr⁺ model, it demonstrated that phase-variable production of Dr fimbriae measurably modifies the surface properties of the clinical strain.

Unlike the constitutively Dr-positive AAEC191A strain, glutaraldehyde treatment did not induce detectable aggregation of IH11128, most likely because only a minority of cells expressed Dr fimbriae. However, immunoagglutination with polyclonal anti-Dr antibodies readily produced large bacterial aggregates at serum dilutions of 1:10, 1:25, and 1:50, whereas no aggregation was observed for DR14 or with antibody diluted 1:500 (Fig 12D).

Dr-producing cells exhibit enhanced mechanical stability of adhesion.

To determine whether the adhesion phenotypes observed in the AAEC191A model were retained in the clinical isolate, adhesion experiments were performed in AU. Following withdrawal of a receding liquid meniscus, only scattered IH11128 cells remained attached to the glass surface. Immunofluorescence demonstrated that virtually all retained bacteria belonged to the Dr-positive subpopulation, whereas Dr-expressing cells represented only a minor fraction of bacteria present before meniscus withdrawal (Fig 13A,B). In contrast, DR14 cells were removed almost completely from the surface and no attached bacteria remained after meniscus passage (Fig 13C). Laminar-flow experiments produced similar results. On polystyrene, IH11128 and DR14 displayed comparable initial adhesion (478 ± 21 and 466 ± 23 cells per field, respectively) and similar resistance to detachment up to 2.8 pN μm⁻². At higher shear stresses, IH11128 remained significantly more resistant, with 59 ± 8% of initially attached bacteria remaining at 8.8 pN μm⁻², compared with only 14 ± 7% for DR14 (Fig 13D). A comparable trend was observed on glass. Initial attachment reached 597 ± 30 bacteria per field for IH11128 and 504 ± 47 for DR14. Differences became evident above 4.4 pN μm⁻², and at 8.8 pN μm⁻² 48 ± 7% of IH11128 bacteria remained attached compared with only 9 ± 4% of DR14 cells (Fig 13D).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 13. Validation of Dr-dependent adhesion under urinary conditions in the clinical UPEC strain IH11128.

(A) Representative phase-contrast (PC), immunofluorescence (IF anti-Dr), and merged (PC + IF) images of the phase-variable clinical strain IH11128 during the meniscus retraction assay performed on glass in AU. The phase-contrast image shows the region exposed after meniscus withdrawal (a), the receding meniscus (m), and the bacterial suspension remaining ahead of the meniscus (b). Immunofluorescence identifies the Dr-positive subpopulation labeled with rabbit polyclonal anti-Dr antibodies and TRITC-conjugated secondary antibodies, while the merged image shows the spatial relationship between Dr-producing bacteria and the meniscus. Scale bars, 20 µm. (B) Representative phase-contrast, immunofluorescence, and merged images of IH11128 bacteria remaining attached to the glass surface after complete meniscus withdrawal. Immunofluorescence demonstrates that the retained bacteria belong almost exclusively to the Dr-positive subpopulation. Scale bars, 20 µm. (C) Representative phase-contrast and immunofluorescence images of the Dr-deficient mutant DR14 after meniscus withdrawal under identical conditions. No bacteria remained attached to the glass surface after passage of the meniscus, and no specific fluorescence signal was detected. Scale bars, 20 µm. (D) IH11128 and DR14 suspensions in AU (OD₆₀₀ = 0.35) were introduced into flow chambers assembled on glass or untreated polystyrene surfaces and incubated for 10 min under static conditions to allow initial adhesion. Flow was then applied in sequential 1-min intervals generating increasing shear stresses of 0.01, 0.05, 0.28, 0.6, 1.1, 2.8, 4.4, 6.6, and 8.8 pN µm⁻². The graphs show the number of adherent bacteria during stepwise increase of shear stress. The right panels indicate the percentage of initially attached bacteria remaining on the surface at the end of the flow experiment. Data represent mean ± SD. Statistical significance was evaluated using the non-parametric Kruskal–Wallis test (*** p < 0.001).

https://doi.org/10.1371/journal.ppat.1014588.g013

Together, these observations demonstrate that the mechanically stable adhesion phenotype identified in the AAEC191A model is also present in the clinical UPEC strain and is associated specifically with the Dr-producing bacterial subpopulation.

Dr fimbriae mediate selective association of clinical UPEC with calcium oxalate.

The interaction of IH11128 with CaOx was examined under the same AU conditions. Approximately 31 ± 2% of IH11128 bacteria became associated with CaOx crystals, whereas only 1 ± 1% of DR14 cells were removed from suspension (Fig 14A). Phase-contrast and SYTO 9 fluorescence microscopy confirmed extensive accumulation of IH11128 cells on crystal surfaces, while only occasional bacteria remained free in suspension (Fig 14B). Preincubation of IH11128 with polyclonal anti-Dr serum (1:10) reduced crystal-associated bacteria by approximately 90%, indicating that binding depends directly on Dr fimbriae (Fig 14C). In contrast, DR14 showed only sporadic interaction with CaOx, and anti-Dr antibodies produced no detectable effect (Fig 14D,E).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 14. Dr fimbriae mediate specific binding of the clinical UPEC strain IH11128 to calcium oxalate crystals.

(A) Percentage of IH11128 and DR14 bacteria transferred from suspension to CaOx during incubation in AU (pH 5.7) under agitation, calculated from the decrease in OD₆₀₀ relative to control suspensions without crystals. Overnight cultures were resuspended in AU to OD₆₀₀ = 0.8 and incubated with 1.7% (w/v) CaOx crystals for 40 min with gentle inversion followed by 30 min sedimentation. Data represent mean ± SD. Statistical significance was evaluated using the non-parametric Kruskal–Wallis test (*** p < 0.001). (B, D) Representative phase-contrast (PC), SYTO 9 fluorescence, and merged (PC+SYTO 9) images of IH11128 (B) and the Dr-deficient mutant DR14 (D) following incubation with CaOx crystals in AU. Phase-contrast images show crystal morphology, fluorescence images show SYTO 9-labeled bacteria, and merged images present the spatial relationship between bacteria and crystal surfaces. (C, E) Representative images of IH11128 (C) and DR14 (E) following pre-incubation with rabbit polyclonal anti-Dr antibodies (1:10 dilution) before incubation with CaOx. Imaging was performed as described for panels B and D. Scale bars in panels B–E, 100 µm. (F, G) Representative phase-contrast (PC), anti-Dr immunofluorescence (TRITC), and merged (PC+IF) images of IH11128 (F) and DR14 (G) incubated with CaOx crystals. Following incubation, samples were immunolabeled with rabbit polyclonal anti-Dr antibodies and TRITC-conjugated goat anti-rabbit secondary antibodies. TRITC fluorescence was detected exclusively on IH11128 bacteria associated with CaOx crystals, whereas no specific fluorescence signal was observed for DR14. Scale bars, 10 µm.

https://doi.org/10.1371/journal.ppat.1014588.g014

Immunofluorescence analysis further demonstrated that virtually all bacteria attached to CaOx crystals expressed Dr fimbriae (Fig 14F), whereas no fluorescence was detected for DR14 (Fig 14G). Considering that only approximately 20% of IH11128 cells expressed Dr fimbriae under these growth conditions, these observations indicate highly efficient and selective recruitment of the Dr-positive subpopulation to CaOx crystals.

Bacterial strains, culture media, and experimental buffers

The E. coli strain IH11128 (O75:K5:H-) is a clinical uropathogenic isolate recovered from a patient with pyelonephritis that expresses Dr fimbriae [94,95]. The isogenic draC insertion mutant DR14 was generated from IH11128 by insertional mutagenesis using a suicide vector, as previously described. Inactivation of draC, required for Dr fimbrial assembly, abolishes the surface expression of Dr fimbriae [96]. E. coli AAEC191A is a derivative of the K-12 laboratory strain carrying a deletion of the entire fim operon (Δfim), generated by allelic exchange, which abolishes expression of type 1 fimbriae [97]. E. coli JM101 (ATCC 33876) and BL21(DE3) (Novagen) were used together with AAEC191A for heterologous production of Dr fimbriae encoded by plasmid pCC90. Plasmid pCC90, a derivative of pACYC177, carries the dra operon in which the promoter region and regulatory genes upstream of draB were deleted. The operon was cloned under control of the constitutive kanamycin resistance gene promoter [52]. The parental plasmid pACYC177 (New England Biolabs) was used to construct non-fimbriated control strains. Throughout the manuscript, strains carrying pCC90 and pACYC177 are referred to as Dr⁺ and Dr⁻, respectively.

Unless stated otherwise, all strains were maintained on LB broth with agar. Liquid cultures of the laboratory strains E. coli AAEC191A, BL21(DE3), and JM101 carrying either the Dr fimbriae-encoding plasmid pCC90 (Dr⁺) or the control plasmid pACYC177 (Dr⁻) were routinely grown in LB broth supplemented with ampicillin (100 μg/ml; Sigma-Aldrich). Liquid cultures of the clinical UPEC strain IH11128 were routinely grown in M63 minimal medium supplemented with 1% (w/v) casamino acids hydrolysate and 0.2% (w/v) glycerol, whereas the isogenic Dr-deficient mutant DR14 was grown under the same conditions with the addition of ampicillin (10 μg/ml). Unless indicated otherwise, pH-dependent experiments were performed with AAEC191A Dr⁺ and Dr⁻ bacteria suspended in PBS buffers of defined composition: pH 5.7 (130 mM NaCl, 0.1774 mM Na₂HPO₄, 19.82 mM NaH₂PO₄), pH 7.0 (130 mM NaCl, 10.36 mM Na₂HPO₄, 9.636 mM NaH₂PO₄), and pH 8.0 (130 mM NaCl, 18.3 mM Na₂HPO₄, 1.698 mM NaH₂PO₄). AU used in adhesion and crystal interaction assays contained NaCl (110 mM), Na₂SO₄·10H₂O (10 mM), KH₂PO₄ (7 mM), K₂HPO₄ (7 mM), lactic acid (1.1 mM), citric acid (3 mM), NaHCO₃ (2 mM), urea (170 mM), uric acid (0.4 mM), creatinine (7 mM), NH₄Cl (25 mM), MgSO₄·7H₂O (2 mM), CaCl₂·2H₂O (1.663 mM), and FeSO₄ (0.005 mM). The solution was adjusted to pH 5.7 and sterilized by filtration through a 0.22 µm membrane filter.

Bacterial concentration in cultures and suspensions was routinely estimated by OD₆₀₀ measurements. For laboratory strains carrying either the Dr fimbriae-encoding plasmid pCC90 (Dr⁺) or the control plasmid pACYC177 (Dr⁻), Dr fimbriae production increased light scattering, resulting in higher OD₆₀₀ values at identical cell numbers. Therefore, equivalent cell numbers were obtained by using a 1.3-fold higher OD₆₀₀ for Dr⁺ than for Dr⁻ suspensions. This correction factor was established by direct cell counting in a Petroff chamber. In contrast, the clinical strain IH11128 contains only a subpopulation of Dr-expressing bacteria, resulting in no appreciable difference in OD₆₀₀ relative to its isogenic Dr-deficient mutant DR14. Therefore, identical OD₆₀₀ values were used for IH11128 and DR14 throughout the study.

Because BL21(DE3) and JM101 showed strong autoaggregation and sedimentation, they were excluded from experiments where spontaneous aggregation could affect interpretation of Dr fimbrial effects, including viscosity measurements, vacuum filtration, zeta potential determination, MATH, CA analysis, adhesion to abiotic surfaces, phage adsorption assays, and LC–MS analysis.

Antisera and antibodies

Rabbit polyclonal anti-Dr antiserum (Immunolab, Gdynia, Poland) was generated by immunization with purified native Dr fimbriae isolated from E. coli AAEC191A carrying plasmid pCC90. The antibody titer, determined by ELISA against purified Dr fimbriae, was 1:11,500. This antiserum was used as the primary antibody in immunofluorescence, immunoaggregation, and Western blotting experiments. For Western blot analysis, horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (Goat Anti-Rabbit IgG, HRP-conjugate; Merck, Cat. #12–348) was used. For immunofluorescence, either rhodamine-conjugated goat anti-rabbit IgG (Goat Anti-Rabbit IgG, H & L Chain Specific Rhodamine Conjugate–TRITC; Calbiochem, Cat. #40 1321) or fluorescein-conjugated goat anti-rabbit IgG (Goat Anti-Rabbit IgG, H & L Chain Specific Fluorescein Conjugate–FITC; Calbiochem, Cat. #40 1314) was used as the secondary antibody.

Isolation of Dr fimbriae from cell surfaces

An overnight culture of E. coli AAEC191A Dr⁺ adjusted to OD₆₀₀ = 3.4 was used for fimbriae isolation. One milliliter of culture was collected, centrifuged, and the pellet resuspended in 500 µl PBS (pH 7.4). Samples were incubated at 65 °C for 3 h to release surface-associated fimbriae, followed by centrifugation at 1,500 × g for 5 min. Supernatants containing detached fimbrial fractions were collected and analyzed by LC–MS. For zeta potential measurements, fimbrial fractions were concentrated using Amicon Ultra-15 Ultracel-100K (Merck) and further purified by size exclusion chromatography on a Superdex 200 Increase 10/300 GL column (Cytiva) using the appropriate running buffer.

Detection of Dr fimbriae with Western blotting

For all experiments involving Dr⁺ strains, the presence of Dr fimbriae was routinely verified by Western blotting. Bacterial suspensions were mixed with Laemmli buffer and either heated at 100 °C or incubated at room temperature for 60 min before separation by SDS–PAGE in 15% (w/v) polyacrylamide gels. Heating caused complete depolymerization of Dr fimbriae into DraE monomers, detected as a ~ 14 kDa band, whereas non-heated samples preserved the polymeric structure and remained at the top of the stacking gel because of their high molecular weight. Parallel analysis of heated and non-heated samples confirmed that DraE was present as polymerized fimbriae. Proteins were transferred to nitrocellulose membranes, which were blocked for 1 h at room temperature in PBS containing 1% (w/v) bovine serum albumin (BSA). Membranes were then incubated for 1 h at room temperature with rabbit polyclonal anti-Dr primary antibodies diluted 1:1,000 in PBS containing 1% BSA, followed by incubation for 1 h at room temperature with HRP-conjugated goat anti-rabbit IgG secondary antibodies (1:1,000 dilution in PBS containing 1% BSA).

Relative quantification of Dr fimbriae in model AAEC191A Dr⁺ and IH11128 strains

To compare Dr fimbriae production between AAEC191A Dr⁺ and IH11128, both strains were cultured for 16 h at 37 °C with shaking in M63 medium supplemented with 1% casamino acids and 0.2% glycerol. Cells were harvested by centrifugation (5000 × g) and resuspended to OD₆₀₀ = 2.5 before isolation of fimbrial fractions, which were analyzed by Western blotting. The relative DraE content in fimbrial fractions heated at 100 °C was determined densitometrically using a VersaDoc imaging system and Quantity One (Bio-Rad Laboratories). AAEC191A Dr⁺ was set as 100%, and DraE levels in IH11128 were calculated relative to this reference. To estimate the fraction of IH11128 cells actively producing Dr fimbriae, overnight cultures were harvested, resuspended in 1 ml PBS to an OD₆₀₀ of 0.7, and analyzed by immunofluorescence microscopy. Bacterial suspensions were incubated for 1 h at room temperature with rabbit polyclonal anti-Dr primary antibodies (1:50 dilution in 1 ml PBS containing 1% BSA), followed by three washes with 1 ml PBS. Cells were subsequently incubated for 1 h at room temperature with FITC- or TRITC-conjugated goat anti-rabbit IgG secondary antibodies (1:100 dilution in 1 ml PBS containing 1% BSA), washed three times with 1 ml PBS, and examined by phase-contrast and fluorescence microscopy. Phase-contrast images were used to determine the total number of bacterial cells, whereas fluorescence images identified the Dr fimbriae-positive subpopulation. The percentage of Dr-positive cells was determined by counting bacteria in three randomly selected microscopic fields, each measuring 700 × 700 µm. The percentage of fimbriae-producing cells was used to correct DraE abundance values obtained by Western blotting. Three independent biological replicates, each including three technical replicates, were performed.

Culture viscosity measurements

Two E. coli strains, AAEC191A Dr⁻ and AAEC191A Dr⁺, were used for viscosity measurements. Cells were grown in LB broth at 37 °C to OD₆₀₀ ≈ 0.5 and diluted tenfold with fresh medium before introduction into the microfluidic system.

Viscosity and optical density were measured simultaneously using a droplet-based microfluidic method analyzing individual bacterial suspension droplets flowing through a microchannel [98]. The channel had a square cross-section of 600 × 600 µm and two optical sensors placed 1 m apart, allowing determination of droplet transit time and absorbance at 600 nm. A constant pressure difference of ~25 mbar maintained flow of the continuous phase, consisting of hexadecane with 1.0% Span-80 to minimize interfacial tension effects. The system was kept at 37 °C, and droplet volume was 2.16 µl. During measurements, droplets were repeatedly passed between sensors to obtain reproducible transit time and absorbance values. Viscosity was calculated from transit time using prior calibration with water–glycerol mixtures of known viscosity; shorter transit times indicated lower viscosity and longer times higher hydrodynamic resistance. Five independent biological replicates were performed.

Vacuum filtration of bacterial suspensions

Overnight cultures of E. coli AAEC191A Dr⁺, AAEC191A Dr⁻, IH11128, and DR14 were harvested by centrifugation and washed once with PBS (pH 5.7, 7.0, or 8.0) or 0.1 M potassium nitrate. For experiments performed with the AAEC191A model, non-fimbriated bacteria were adjusted to OD₆₀₀ = 0.65 and Dr-fimbriated bacteria to OD₆₀₀ = 0.85 in PBS. In 0.1 M potassium nitrate, AAEC191A Dr⁻ cells were adjusted to OD₆₀₀ = 0.32, 0.65, and 1.3, whereas AAEC191A Dr⁺ cells were adjusted to OD₆₀₀ = 0.43, 0.85, and 1.7. Clinical strains IH11128 and DR14 were analysed only in 0.1 M potassium nitrate and were adjusted to OD₆₀₀ = 0.35, 0.70, and 1.40. For each condition, 10 ml of suspension was filtered through 0.45 µm cellulose membrane filters (47 mm diameter) under constant vacuum at room temperature. Filtration time was recorded in two stages: (i) from the start of filtration to the appearance of the first central dry area on the filter surface, and (ii) until complete disappearance of liquid from the filter surface. Three independent biological replicates, each including three technical replicates, were performed. Results were statistically evaluated using the non-parametric Kruskal-Wallis test.

Electron microscopy

The ultrastructure of bacterial cells was analyzed by transmission and scanning electron microscopy. For TEM, overnight cultures of E. coli AAEC191A Dr⁺ and Dr⁻ grown in LB broth were harvested by centrifugation and resuspended in PBS (pH 7.4) to OD₆₀₀ = 0.1. A 5 µl aliquot of diluted suspension was applied to carbon-coated grids (GF Microsystems), negatively stained with 1.5% uranyl acetate (Sigma-Aldrich), and examined using a Tecnai G2 Spirit BioTWIN (FEI Company) operated at 120 kV. For SEM, morphology was examined using a Quanta FEG250 equipped with a secondary electron ET detector. Samples were coated with a 10 nm gold layer to prevent electrostatic charging, and images were acquired at 10 kV. For both TEM and SEM analyses, three independent biological replicates were performed, each comprising three technical replicates.

LC–MS/MS analysis

Bacterial samples for proteomic LC–MS/MS analysis were prepared using a standardized workflow optimized for quantitative protein recovery. E. coli AAEC191A Dr⁺ and Dr⁻ were grown in LB broth at 37 °C with agitation to stationary phase (OD₆₀₀ 3.3–3.5). Cell numbers were determined using a Petroff chamber (Marienfeld, Germany).

For total cell proteome analysis, 180–200 µl of culture was harvested by centrifugation and washed once with ice-cold 1 × PBS to remove medium residues. Metabolism was quenched by adding two volumes of pre-chilled methanol:ethanol (1:1, v/v; −20 °C), followed by brief vortexing and immediate freezing on dry ice. Samples were thawed on ice and dried using a SpeedVac (Eppendorf, Germany) at 30 °C for 4 h. Dried pellets were resuspended in 1 ml PBS, and protein concentration was measured at 280 nm. Samples were adjusted to 10–40 µg/µl. Fimbrial fractions were prepared as described above. For denaturation and reduction, 30 µl of protein solution (~20 µg/µl) was mixed with 30 µl water, 240 µl 8 M guanidine hydrochloride, 72 µl 300 mM Tris-HCl (pH 8.0) containing 12 mM CaCl₂, and 12 µl 100 mM tris(2-carboxyethyl)phosphine. Samples were incubated at 37 °C for 30 min, then alkylated with 24 µl 300 mM iodoacetamide for 30 min in the dark at 37 °C. For enzymatic digestion, 34 µl of reduced and alkylated protein solution was mixed with 106 µl water, 10 µl 500 mM Tris-HCl (pH 8.0), and 10 µl 0.5 µg/µl Trypsin Platinum (Promega), then incubated overnight (16 h) at 37 °C. Digestion was stopped by adding 5 µl formic acid. Peptide samples were stored at −80 °C until analysis.

NanoLC–MS/MS analysis was performed using an ACQUITY M-Class (Waters) coupled to a ZenoTOF 7600 (SCIEX, Canada) with a nano-ESI source. Five microliters of peptide digest were injected onto a C18 trap column and separated on a reverse-phase C18 analytical column (75 µm × 100 mm, 2 µm particles). Mobile phase A was 0.1% formic acid in water, and mobile phase B was 0.1% formic acid in methanol. Flow rate was 350 nl/min. Mass spectrometric acquisition was performed in positive ion mode over m/z 400–3000 using full-scan acquisition with DIA (SWATH) and IDA MS/MS. Instrument settings were: ion source voltage 3800 V, collision energy 10 eV, curtain gas 35 psi, nebulizing gas 12 psi, source temperature 190 °C, and declustering potential 100 V. Raw data were processed with SCIEX OS and Spectronaut 18 (Biognosys). Quantitation was based on extracted ion chromatograms (XICs) with a ± 0.02 Da extraction window.

Absolute quantification of DraB, DraC, DraD, and DraE proteins was performed using synthetic peptide standards (Lipopharm, Gdańsk, Poland) corresponding to unique tryptic peptides: SPAPFVVTPPLFR (DraB), IEEYPALFR (DraC), TGGDGWSPVK (DraD), and FFLISDNNR (DraE) (S4 Table in S1 File). Peptides (>95% purity) were dissolved in LC-MS-grade water containing 5% methanol and added to digested biological samples to generate calibration curves (S5 Table in S1 File, S18 Fig). Doubly charged precursor ions ([M + 2H]²⁺) were quantified at m/z values 714.4059 (DraB), 569.3006 (DraC), 502.2458 (DraD), and 563.2881 (DraE). Extracted ion chromatograms were generated using a ± 0.01 Da mass window. Calibration curves were obtained from technical triplicates and used for absolute protein quantification. Retention times used for peptide identification were 27.92 min (DraB), 24.34 min (DraC), 19.34 min (DraD), and 24.27 min (DraE). Analytical sensitivity was estimated using the standard calibration-based equations for the limit of detection (LOD) and limit of quantification (LOQ): LOD = 3.3 × σ/S and LOQ = 10 × σ/S, where S denotes the slope of the calibration curve and σ denotes the standard deviation of the response estimated from the regression model. The amount-based LOD and LOQ values were subsequently converted to particles per cell using the quantitative normalization applied in the study. DraB showed the lowest calculated thresholds, with an LOD of 22.49 particles per cell and an LOQ of 68.12 particles per cell. DraD showed intermediate sensitivity (LOD 33.98, LOQ 103.03 particles per cell), whereas DraE and DraC exhibited the highest calculated thresholds, both with LOD values of approximately 63 particles per cell and LOQ values of approximately 191 particles per cell.

Three independent biological replicates, each including three technical replicates, were performed. Results were statistically evaluated using the unpaired two-tailed Student’s t-test.

Structural modeling of the Dr fimbrial envelope

A structural model of the Dr fimbrial envelope produced by E. coli Dr⁺ was constructed using the known structure of self-complemented AfaE as a template and the established DraE donor strand complementation dimerization mode [20]. A linear 7-mer polymer composed of seven identical DraE chains was generated using MODELLER [99].

Polymer conformational sampling was performed in three independent 500-ns equilibrium simulations using GROMACS with the amber99SB-ILDN force field. Simulations were carried out in an elongated 10 × 10 × 36 nm simulation box containing 113071 TIP3P water molecules and 0.15 M KCl [100,101], representing a short fimbrial segment embedded in a dense fimbrial layer. From the combined trajectories, 450 conformational snapshots were extracted and divided into five overlapping dimer geometries each, excluding the terminal N-terminal dimer, resulting in a pool of 2250 DraE dimer conformations. These dimer geometries were subsequently used to assemble grid models of the fimbrial layer consisting of 5 × 5 units, corresponding to experimentally determined DraC usher surface density (approximately 1 per 3970 nm²) and the DraE/DraC ratio, approximated as 150 DraE subunits per fimbrial chain.

Assembly of the fimbrial layer was performed using a Monte Carlo procedure. An initial dimer was positioned at Z = 0 with random lateral displacement in the XY plane, and subsequent extension steps were generated by aligning randomly selected dimers from the simulation-derived pool onto the terminal DraE subunit using MDTraj [102]. Proposed extensions were rejected if they produced steric clashes, penetrated the membrane region (center of mass below 2 nm), or introduced downward-directed growth with excessive probability. Periodic boundary conditions in the XY plane were applied during clash detection. To reconstruct the full envelope geometry, the resulting 315 × 315 nm patch was transformed into a cylindrical segment with a radius of 500 nm, corresponding to the average radius of an E. coli cell. Nine successive 36° rotations around the Y axis were applied to generate a complete cylindrical surface, followed by axial replication using VMD [103]. Additional geometric manipulations were performed using Gromologist [104].

Protein density profiles and steric clash probabilities were calculated using custom Python scripts. For large-scale assembly, a mesoscale representation of the fimbrial layer was generated by reducing each 139-residue DraE unit to 10 representative Cα atoms selected by k-medoid clustering, yielding final cylinder segments containing 377500 particles, corresponding to approximately 78 million atoms. To obtain smoothed density profiles, cylinder segments were projected on the xy plane, offset to align the profile with the edge of the cylinder representing the bacterium, and normalized to a maximum of 1 at that point. The corresponding density profiles from TEM micrographs were calculated using ImageJ.

Microbial adhesion to hydrocarbons (MATH)

Overnight cultures of E. coli AAEC191A Dr⁺, AAEC191A Dr⁻, IH11128, and DR14 were suspended in PBS (pH 7.0). AAEC191A Dr⁺, IH11128, and DR14 were adjusted to OD₆₀₀ = 1.3, whereas AAEC191A Dr⁻ was adjusted to OD₆₀₀ = 1.0 to obtain comparable bacterial cell numbers. For each measurement, 3 ml of bacterial suspension was mixed with 1 ml of organic phase (xylene or hexadecane). Control samples without organic phase were prepared in parallel. Samples were vortexed for 90 s and incubated 15 min at room temperature for phase separation. One milliliter of the aqueous phase was collected, and absorbance at 570 nm was measured. The percentage of adhesion, reflecting surface hydrophobicity, was calculated as:

where A₀ is the absorbance of the control sample, and A₁ is that of the aqueous phase after mixing.

For microscopic observation, 5 µl of bacterial suspension prepared as described for the MATH assay were placed into a 5 mm × 2 mm well on a glass slide, which was then filled with the respective organic phase and sealed with a coverslip. After 15 min, the water–organic interface was observed using an inverted phase-contrast microscope.

Three independent biological replicates, each including three technical replicates, were performed. Results were statistically evaluated using the non-parametric Kruskal-Wallis test.

Determination of wetting contact angle and ΔG_BWB

Bacterial lawns of E. coli AAEC191A Dr⁺ and Dr⁻ were prepared on cellulose membrane filters as in the vacuum filtration section. Briefly, 10 ml of suspensions in 100 mM potassium nitrate were filtered, using OD₆₀₀ = 0.85 for Dr⁺ and 0.65 for Dr⁻ cells. The confluent multilayers were washed with 5 ml deionized water, dried at room temperature for 50 min, and analyzed within 30 min. Complete coverage of the filters was confirmed by scanning electron microscopy.

Contact angles were measured using a Ramé-hart Model 90-U3-PRO system with DROPimage Pro. Deionized water, formamide, and diiodomethane (Sigma-Aldrich) served as probe liquids. Droplets were applied as freely falling drops from a height of 1 cm using a Gilmont GS-1200 Micrometer Syringe. Contact angles were determined with DROPimage Pro, with ten independent measurements per sample and liquid. Three independent biological replicates, each including three technical replicates, were performed.

Surface free energy parameters were calculated using the Good–van Oss–Chaudhury model [105]. The hydrophobicity of bacterial surfaces was expressed as the free energy of interaction between bacterial surfaces immersed in water according to the following equation:

where: ∆𝐺_BWB – value of free energy of interaction between bacteria when immersed in water, B – bacteria, W – water, 𝛾^𝐿𝑊, 𝛾⁺, 𝛾⁻ – Lifshitz-van der Waals (dispersive), acid (electron acceptor), and base (electron donor) components of surface free energy. The values of 𝛾^𝐿𝑊, 𝛾⁺, 𝛾⁻ for water were taken from reference [106].

Zeta potential measurements

Overnight cultures of E. coli AAEC191A Dr⁺ and Dr⁻ were harvested and suspended in PBS at pH 5.7, 7.0, and 8.0 to OD₆₀₀ = 0.5 and 0.4, respectively. Measurements were performed using a Zetasizer Nano ZS (Malvern Instruments, UK) based on laser Doppler electrophoresis with a 633 nm red laser. Samples were loaded into DTS1060 folded capillary cells and equilibrated for 120 s at 25 °C. Zeta potentials were calculated using the Smoluchowski approximation. Temperature was maintained at 25 °C using the Peltier-controlled holder. Dispersant parameters for calculations were: viscosity 0.89 cP, refractive index 1.33, and dielectric constant 78.5. Instrument control and data analysis were done with Malvern Dispersion Technology Software. Three independent biological replicates, each including three technical replicates, were performed. Six independent measurements were conducted per sample, each with at least 15 sub-runs and 120 s intervals. Results were statistically evaluated using the Welch’s corrected Student’s t-test.

Effect of glutaraldehyde and antibody-mediated cross-linking of bacterial cells

Overnight cultures of E. coli AAEC191A Dr⁺, AAEC191A Dr⁻, IH11128, and DR14 were harvested and resuspended in 1 ml PBS (pH 7.5) containing 0.05% Tween 20 to OD₆₀₀ = 1.0. Glutaraldehyde (25% stock solution) was added to final concentrations of 2%, 1%, 0.5%, 0.25%, 0.1%, and 0.01%. Samples were incubated for 15 min at 4 °C to allow cross-linking. Reactions were quenched by adding 300 µl PBS containing 50 mM glycine and incubated for an additional 15 min at room temperature. The degree of cross-linking was evaluated by phase-contrast microscopy by determining the lowest glutaraldehyde concentration inducing visible bacterial aggregates and the maximum aggregate size observed at each concentration.

To assess Dr-specific immunoagglutination, IH11128 and DR14 cells were resuspended in 1 ml PBS (pH 7.5) supplemented with 0.1% BSA to OD₆₀₀ = 1.0. Rabbit polyclonal anti-Dr antibodies were added at final dilutions of 1:10, 1:25, 1:50, and 1:500. Samples were incubated for 1 h at room temperature with gentle mixing, and bacterial aggregation was evaluated by phase-contrast microscopy using an inverted microscope.

Three independent biological replicates, each including three technical replicates, were performed.

Assessment of bacterial adhesion to glass and polystyrene surfaces under static conditions

Overnight cultures of E. coli AAEC191A Dr⁺ and Dr⁻ were harvested and resuspended in PBS at pH 5.7, 7.0, or 8.0 to OD₆₀₀ = 0.7. Aliquots of 40 µl were deposited onto glass or polystyrene surfaces and incubated for 15 min at room temperature under static conditions. Non-adherent cells were removed by gently rinsing with 1 ml of the corresponding buffer. Attached bacteria were visualized by phase-contrast microscopy and quantified using the ‘Count and Measure’ module in cellSens Dimension 4.1. For each slide, bacteria were counted in 10 fields, each covering an area of 100 µm². Three independent biological replicates, each including three technical replicates, were performed. Results were statistically evaluated using the non-parametric Kruskal-Wallis test.

Meniscus retraction assay on glass and polystyrene surfaces

Overnight cultures of E. coli AAEC191A Dr⁺ and Dr⁻ were harvested by centrifugation and resuspended in PBS (pH 7.0) to an OD₆₀₀ of 0.5. Aliquots of 10 µl of bacterial suspension were deposited onto glass slides or untreated polystyrene surfaces and covered with coverslips. Samples were maintained at 25 °C and constant humidity. Meniscus movement was induced by gradual evaporation of water from the edge of the coverslip, while the bacterial liquid layer remained at 10 ± 2 µm. Meniscus motion was recorded using an inverted Olympus IX73 microscope equipped with a UPlanFL N 20 × /0.30 objective and a Hamamatsu Photonics ORCA-Flash2.8 camera, providing a field of view of 696.2 × 522.2 µm and a spatial resolution of 0.363 µm per pixel. Images were acquired at 5 frames s⁻¹ for 60 s on glass and 80 s on polystyrene. Meniscus velocity was determined from the distance traveled during image acquisition. Video files were converted to MP4 using Kdenlive. Bacteria remaining after meniscus passage were quantified by counting cells in 10 fields, each covering an area of 100 µm², before and after meniscus movement within the same regions using the ‘Count and Measure’ module in cellSens Dimension 4.1.

For experiments performed with the clinical strains, only glass surfaces were used. To identify the Dr-positive subpopulation, IH11128 and DR14 cells (1 ml PBS pH 7.4, OD₆₀₀ = 1.0) were incubated for 1 h at room temperature with gentle mixing in the presence of rabbit polyclonal anti-Dr antibodies diluted 1:200. The high antibody dilution was selected to minimize interference with Dr-mediated adhesion while allowing reliable identification of Dr-producing cells by fluorescence microscopy. Cells were washed once with 1 ml PBS and subsequently incubated for 1 h with TRITC-conjugated goat anti-rabbit secondary antibodies diluted 1:200. After a single wash with AU, bacterial suspensions were resuspended in AU to OD₆₀₀ = 0.5, and the meniscus retraction assay was performed on glass as described above. The use of AU reproduced the urinary conditions employed in the remaining validation experiments with the clinical strains. Following meniscus passage, bacteria remaining attached to the glass surface were examined by both phase-contrast and fluorescence microscopy, allowing discrimination between Dr-positive and Dr-negative cells within the IH11128 population.

Three independent biological replicates, each including three technical replicates, were performed.

Adhesion to glass and polystyrene surfaces under laminar flow conditions

Adhesion of overnight cultures of E. coli AAEC191A Dr⁺ and Dr⁻ to glass and untreated polystyrene surfaces was analyzed under laminar flow-induced shear stress. Bacterial suspensions were prepared in PBS at pH 5.7, 7.0, or 8.0 to OD₆₀₀ values of 0.4 for Dr⁺ and 0.3 for Dr⁻ cells. Suspensions were introduced into a parallel plate flow chamber (GlycoTech) with a silicone gasket forming a flow cell 2.5 mm wide, 10 mm long, and 0.25 mm high. The chamber was attached to the surface by vacuum, and flow was generated using a syringe pump (Harvard Apparatus).

Resistance of surface-bound bacteria to detachment by shear stress. Bacterial suspensions in the respective buffers were introduced into the flow chamber and incubated without flow for 10 min to allow adhesion. Subsequently, sequential one-minute flow intervals generating increasing shear stresses of 0.01, 0.05, 0.28, 0.6, 1.1, 2.8, 4.4, and 6.6 pN µm⁻² were applied to assess detachment of surface-bound bacteria.

For experiments performed with the clinical strains, IH11128 and DR14 were suspended in AU to OD₆₀₀ = 0.35. Detachment assays were performed on both glass and untreated polystyrene surfaces following the same protocol as described above, except that an additional shear stress of 8.8 pN µm⁻² was applied at the end of the detachment series.

Accumulation of bacteria under shear stress. To investigate adhesion kinetics under flow, bacterial suspensions were passed through the chamber for 15 min at constant shear stresses of 0.01, 0.05, or 0.28 pN µm⁻². This experiment was performed only for bacteria in PBS at pH 5.7, as adhesion under flow at pH 7.0 and 8.0 was negligible.

Time-lapse videos were recorded using the same imaging setup as in the experiment analyzing bacterial detachment by the moving meniscus. The acquisition rate was adjusted to match the flow velocity, ranging from 2 frames s⁻¹ at a shear stress of 0.01 pN µm⁻² to 50 frames s⁻¹ at shear stress values above 6.6 pN µm⁻². For both detachment and accumulation analyses, three independent biological replicates were performed, each comprising three technical replicates. Statistical analysis of bacterial attachment and accumulation was performed as described in reference [25].

Adhesion of bacteria to calcium salt crystals

Overnight cultures of E. coli AAEC191A Dr⁺ and Dr⁻ were harvested and resuspended in PBS adjusted to pH 5.7, 7.0, or 8.0 to OD₆₀₀ values of 0.9 and 0.7, respectively. To 5.5 ml of bacterial suspension, 500 µl of a 20% suspension of CaOx monohydrate or HA prepared in the corresponding buffer was added. Samples were incubated for 40 min with gentle inversion followed by 30 min sedimentation. Immediately after mixing, 500 µl aliquots were collected for further analyses. Control samples were prepared by replacing crystals with PBS. After incubation, the optical density of the supernatant was measured at 600 nm. The percentage of bacterial adhesion was calculated as described for the MATH assay, where A₀ represents the absorbance of the control without crystals and A₁ the absorbance of the sample incubated with CaOx or HA. To evaluate bacterial adhesion under conditions resembling urine, the same assay was performed using AU instead of PBS.

For validation experiments, the clinical strains IH11128 and DR14 were examined only for interactions with CaOx. Overnight cultures were resuspended in AU to OD₆₀₀ = 0.8, and the adhesion assay was performed as described above.

Three independent biological replicates, each including three technical replicates, were performed. Results were statistically evaluated using the non-parametric Kruskal–Wallis test.

Interaction of the model AAEC191A Dr⁺ and Dr⁻ strains with HA and CaOx crystals during growth in artificial urine

A sterile polystyrene 24-well plate was prepared by adding 1.5 ml of AU to each well. Overnight bacterial cultures were diluted in AU to OD₆₀₀ = 0.025 and used to inoculate the wells. HA and CaOx were added to selected wells to a final concentration of 0.05% (w/v), while two control wells contained AU inoculated with Dr⁺ or Dr⁻ bacteria, without minerals. The plate was incubated at 37 °C under static conditions. After 2 h of incubation, all experimental variants were analyzed by phase-contrast microscopy using the same imaging setup as in the experiment assessing bacterial detachment by the moving meniscus and recorded as 10 s video sequences acquired at 10 frames s⁻¹ to assess early bacterial interactions with crystal surfaces. After 24 h of incubation, 5 µl of DAPI (2 mg/ml stock solution) was added to each well, followed by incubation for 1 h at room temperature. After staining, the supernatant was gently removed, each well was carefully rinsed once with 1.5 ml of AU, and 1.5 ml of fresh AU was gently added to the crystals remaining at the bottom of the well. The suspensions were subsequently analyzed microscopically using phase-contrast microscopy to visualize HA and CaOx crystals and fluorescence microscopy under DAPI excitation to detect bacteria associated with crystal surfaces. Four independent biological replicates, each including three technical replicates, were performed.

Sedimentation and autoaggregation assay

Overnight cultures of E. coli AAEC191A, BL21(DE3), and JM101 carrying either the Dr fimbriae plasmid pCC90 (Dr⁺) or control pACYC177 (Dr⁻) were grown in LB medium and adjusted to OD₆₀₀ = 2.0 using the strain-specific spent culture medium obtained after centrifugation. Forty milliliters of each adjusted culture were transferred into sterile flat-bottom screw-cap tubes (25 × 145 mm) and left undisturbed at room temperature for sedimentation. Tubes were photographed at 0, 0.5, 1, 2, 2.5, 3, 3.5, 4, 10, 22, 24, 28, and 48 h using a Canon EF 50 mm f/1.4 lens under constant white-light illumination. Changes in culture density were analyzed at three positions along each tube based on grayscale luminance measurements. Luminance values were compared with control tubes containing spent culture medium from centrifuged bacterial cultures incubated identically. Differences in luminance were interpreted as changes in optical density, and the percentage of remaining OD over time was calculated. Autoaggregation and settling behavior of overnight cultures were additionally monitored by phase-contrast microscopy. Three independent biological replicates, each including three technical replicates, were performed.

Bacteriophage efficiency of plating assay

The efficiency of plating of T7, λ vir and P1 vir bacteriophages was determined using Dr⁺ and Dr⁻ variants of E. coli AAEC191A and JM101. M13 bacteriophage was analyzed analogously, but only on JM101 variants due to the presence of sex pili required for phage adsorption. A volume of 100 µl of bacterial culture was mixed with 3 ml of molten LB top agar (0.5%, w/v) containing the appropriate antibiotic and poured onto LB agar plates. For assays involving P1 vir, the top agar was additionally supplemented with calcium chloride to a final concentration of 5 mM. Ten-fold serial dilutions of bacteriophage lysates were prepared in TM buffer containing 10 mM magnesium sulfate and 50 mM Tris hydrochloride (pH 7.5), and 10 µl of each dilution was spotted onto the solidified top agar surface. Plates were incubated upright overnight at 37 °C. Plaques were counted after incubation, and bacteriophage efficiency of plating was expressed as the ratio of plaque numbers obtained on Dr⁺ and Dr⁻ variants.

To examine susceptibility of the clinical strains, undiluted lysates of T7, λ vir, and P1 vir bacteriophages were spotted onto bacterial lawns of IH11128 and DR14 prepared as described above. Because no plaques were observed with any of the undiluted lysates, serial dilution analysis was not performed for these strains.

Three independent biological replicates, each including three technical replicates, were performed. Results were statistically evaluated with the test for proportion.

Bacteriophage adsorption assay

Overnight bacterial cultures were refreshed in fresh LB medium until reaching OD₆₀₀ of 0.3. Samples of 2.4 x 10⁸ bacteria were collected and volumes were adjusted with LB medium to 1 ml, where required, and supplemented with 5 mM MgCl₂ and 5 mM CaCl₂. P1 vir and λ vir lysates were added to obtain the multiplicity of infection (MOI) of 0.05. A control sample contained LB medium supplemented with salts indicated above. Samples were incubated for 30 min at 37 °C and centrifuged at 14,000 x g for 15 min at 4 °C. Phages present in supernatants were titrated on appropriate bacterial strains harboring no plasmids. The percentage of adsorption was calculated from numbers of phages present in samples containing no bacteria vs. numbers of unadsorbed phages in samples with bacteria. In the case of T7 bacteriophage, phage lysate was added to bacterial samples to obtain MOI = 0.001. Samples were incubated for 5 min at 37 °C and two 200 µl aliquots were collected from each sample. One sample was centrifuged for 1 min at 14,000 x g. 100 µl of supernatants and uncentrifuged samples were added to 3 ml of molten top LB agar (0.5%) and poured onto the LB agar plates. Plates were incubated upright overnight at 37 °C. Three independent biological replicates, each including three technical replicates, were performed. The percentage of adsorption was calculated with numbers of plaques obtained for centrifuged samples (unadsorbed phages) vs. uncentrifuged ones (total phages). Results were statistically evaluated with unpaired Student’s t-test.

Determination of LamB abundance

To determine whether altered susceptibility to bacteriophage λ was due to steric shielding rather than changes in receptor abundance, LamB protein levels were quantified in AAEC191A Dr⁺ and Dr⁻ strains. Two unique tryptic peptides, NLIEWLPGSTIWAGK and ITLAQQWQAGDSIWSRPAIR, were analyzed by label-free LC–MS in three biological replicates per condition. No significant differences in peak areas were observed between strains for either peptide (unpaired two-tailed Student’s t-test, p > 0.05), as shown in S6 Table in S1 File and S15 Fig.

Minimum inhibitory concentration (MIC) determination

The minimum inhibitory concentration (MIC) was determined by the broth microdilution method using 96-well microtiter plates with E. coli AAEC191A Dr⁺ and Dr⁻ strains. Bacterial suspensions were standardized to OD₆₀₀ and diluted in LB broth to a final inoculum of approximately OD₆₀₀ = 0.001. Susceptibility was tested against chloramphenicol, erythromycin, kanamycin, spectinomycin, and tetracycline, with antibiotics twofold serially diluted in LB to final ranges of 0.008–128 µg/ml (chloramphenicol), 0.05–200 µg/ml (erythromycin), 0.008–64 µg/ml (kanamycin), 0.025–100 µg/ml (spectinomycin), and 0.002–64 µg/ml (tetracycline). Equal volumes of bacterial inoculum were added to each well. Growth controls, medium sterility controls, and antibiotic-only wells were included. Plates were incubated for 20 h at 37 °C, shaken for 12 s before measurement, and OD₆₀₀ was recorded using a 2 × 2 matrix per well (bandwidth 10 nm, 25 flashes). MIC values were defined as the lowest antibiotic concentration completely inhibiting growth. Three independent biological replicates, each including three technical replicates, were performed.

Antibiotic susceptibility by disk diffusion assay

The susceptibility of E. coli AAEC191A Dr⁺ and Dr⁻ was assessed using a disk diffusion assay on a two-layer agar system. The bottom layer contained LB agar medium, while the upper layer was prepared by mixing LB broth and LB agar in a 2:1 (v/v) ratio with 100 µl of overnight bacterial culture. Antibiotics tested were chloramphenicol (34 mg/ml), erythromycin (100 mg/ml), kanamycin (20 mg/ml), spectinomycin (100 mg/ml), and tetracycline (12.5 mg/ml), with serial tenfold dilutions from 10⁻¹ to 10⁻⁸. Sterile cellulose disks were loaded with 10 µl of antibiotic in 70% ethanol or 15 µl of aqueous solution, depending on solubility. Disks were arranged around the plate perimeter with undiluted stock in the center. Plates were incubated overnight at 37 °C. Three independent biological replicates, each including three technical replicates, were performed.

Statistical analysis

Unless otherwise stated, all experiments were performed with three biological replicates, each including three technical replicates. Bacterial cell counts are presented as mean values ± standard deviation (SD). Unless stated otherwise, statistical significance between the analyzed groups was evaluated using the non-parametric Kruskal-Wallis test. This method was specifically chosen because cell counting yields discrete data that often do not meet the assumptions of normal distribution required for standard parametric tests. The levels of statistical significance are defined as follows: * p < 0.05, ** p < 0.01, and *** p < 0.001.