Ureolytic Biomineralization Reduces Proteus mirabilis Biofilm Susceptibility to Ciprofloxacin

Abstract

Ureolytic biomineralization induced by urease-producing bacteria, particularly Proteus mirabilis, is responsible for the formation of urinary tract calculi and the encrustation of indwelling urinary catheters. Such microbial biofilms are challenging to eradicate and contribute to the persistence of catheter-associated urinary tract infections, but the mechanisms responsible for this recalcitrance remain obscure. In this study, we characterized the susceptibility of wild-type (ure+) and urease-negative (ure−) P. mirabilis biofilms to killing by ciprofloxacin. Ure+ biofilms produced fine biomineral precipitates that were homogeneously distributed within the biofilm biomass in artificial urine, while ure− biofilms did not produce biomineral deposits under identical growth conditions. Following exposure to ciprofloxacin, ure+ biofilms showed greater survival (less killing) than ure− biofilms, indicating that biomineralization protected biofilm-resident cells against the antimicrobial. To evaluate the mechanism responsible for this recalcitrance, we observed and quantified the transport of Cy5-conjugated ciprofloxacin into the biofilm by video confocal microscopy. These observations revealed that the reduced susceptibility of ure+ biofilms resulted from hindered delivery of ciprofloxacin into biomineralized regions of the biofilm. Further, biomineralization enhanced retention of viable cells on the surface following antimicrobial exposure. These findings together show that ureolytic biomineralization induced by P. mirabilis metabolism strongly regulates antimicrobial susceptibility by reducing internal solute transport and increasing biofilm stability.

INTRODUCTION

Proteus mirabilis is one of the most common human pathogens found in catheter-associated urinary tract infections (CAUTIs) (1, 2). Although long-term CAUTIs are usually polymicrobial, P. mirabilis is considered to play a unique and important role in both CAUTI establishment and catheter blockage, due to its strong ability to induce mineral precipitation (3, 4). P. mirabilis produces extensive urease, which hydrolyzes urea to ammonia, increasing pH and inducing mineral precipitation. This process is commonly referred to as ureolytic biomineralization (5). In CAUTIs, P. mirabilis forms distinctive crystalline biofilms through ureolytic biomineralization (6). Typical minerals produced by P. mirabilis in CAUTIs are struvite (magnesium ammonium phosphate), apatite (calcium phosphate), and calcite (calcium carbonate) (7–9).

The treatment of CAUTIs is challenging. Although antimicrobials such as ciprofloxacin and ampicillin are still the most commonly recommended therapeutic agents for urinary tract infections (UTIs) (10), crystalline biofilms are generally recalcitrant to antimicrobials (11). As a result, many patients suffer recurrent biofilm-based infections and catheter encrustation and blockage (12). Alternative therapies, including irrigation with antibiotic-containing solutions and the use of antibiotic-impregnated catheters, also show limited efficacy in preventing biofilm growth and resulting mineral precipitation (13, 14).

Several mechanisms are known to protect biofilm-resident microorganisms from antimicrobials (15, 16). Excreted extracellular polymeric substances (EPSs) protect microorganisms by creating transport barriers that hinder the penetration of antimicrobials into the biofilm and/or sequester antimicrobials within the EPS matrix (17, 18). Physiological heterogeneity within biofilms also contributes to reduced antimicrobial susceptibility of biofilms, by yielding subpopulations with low metabolic activity and other resistant phenotypes (such as persisters) (18, 19). Therefore, the observed challenges in eradicating biofilm-associated CAUTIs are not surprising. However, most of the current understanding of interactions between biofilms and antimicrobials is based on studies of mucoid biofilms, in which the matrix is composed primarily of polymers, and the effects of biomineralization on the antimicrobial susceptibility of biofilms are not known.

Our previous investigations showed that in situ biomineralization profoundly alters biofilm architecture, displaces biofilm biomass, and detaches cells from the biofilm (20). These changes in biofilm morphology and cell density are expected to alter biofilm permeability (21) and may either increase or decrease the transport of antimicrobials within biofilms, depending on the spatial pattern of biomineralization (22). Internal growth of mineral deposits is expected to increase susceptibility to antimicrobials by perforating the biofilm and displacing cells (20), while precipitation at the biofilm surface is expected to decrease susceptibility by forming a transport barrier. P. mirabilis biofilms in CAUTIs usually show a high degree of crystallization (12), but spatial patterns of biomineralization within P. mirabilis biofilms are still unclear. Therefore, we hypothesized that biomineralization would reduce the susceptibility of P. mirabilis to antimicrobials by hindering antimicrobial transport. To test this hypothesis, we resolved spatial and temporal patterns of ureolytic biomineralization in P. mirabilis biofilms grown in artificial urine, and then we tested the susceptibility of both these biomineralized biofilms and a nonbiomineralizing, urease-negative mutant strain to ciprofloxacin. Finally, we compared the transport of Cy5-conjugated ciprofloxacin (Cy5-cipro) in biomineralized and nonbiomineralized biofilms, to determine directly the effects of biomineralization on ciprofloxacin transport within the biofilms.

MATERIALS AND METHODS

Strains and biofilm growth conditions.

A human urinary tract isolate, Proteus mirabilis strain HI4320 (referred to hereafter as ure+) (23), was used to evaluate the effects of biomineralization on the susceptibility of biofilms to ciprofloxacin, a commonly used antimicrobial for P. mirabilis infections. A urease-negative derivative of P. mirabilis HI4320 (referred to hereafter as ure−) (24) was used as a control for nonbiomineralizing biofilm formation. P. mirabilis ure− was constructed through homologous recombination to disrupt ureR, which is a transcriptional regulator that positively activates the expression of ure genes in the presence of urea (25). Previous studies demonstrated that ure− is incapable of producing urease (24, 25). P. mirabilis ure+ was grown on regular LB plates, and ure− was grown on LB plates with the addition of 100 μg/ml ampicillin (25).

Biofilm growth and biomineralization were observed in flow cell systems, with artificial urine as the growth medium. Artificial urine was prepared according to the recipe described in reference 26 and was filter sterilized before use. The artificial urine composition can be found in the supplemental material. The flow cell system was described previously (27); it contains a peristaltic pump to deliver growth medium to the flow cell. Flow cells were sterilized with 1% bleach for 24 h and then thoroughly rinsed with autoclaved deionized water before inoculation. Flow cell inoculation followed procedures described previously (28). Briefly, log-phase ure+ or ure− cells were diluted in 200 ml artificial urine to an initial cell concentration of 10³ cells/ml. Inoculated artificial urine was then circulated in the flow cell system via a peristaltic pump, at a flow rate of 10 ml/h (28). Biofilms and biomineral deposits were observed 6, 12, 18, and 24 h after inoculation. The pH of flow cell effluent filtered through a 0.2-μm-pore-size filter was also analyzed at those times by using a pH meter (Fisher Scientific accumet AR25).

Confocal imaging.

Biofilms and biomineral deposits were visualized using confocal laser scanning microscopy (CLSM) (Leica TCS SP5) with a 63× objective. Biofilm biomass was visualized through staining with SYTO 62 (649/680 nm; Life Technology). Biofilms were stained by injecting 1 ml of 25 μM SYTO 62 into the flow cell, retaining the stain in the flow cell for 15 min in the dark, and then rinsing unbound stain with 1 ml of water before imaging. To visualize mineral deposits with confocal imaging, filter-sterilized calcein (Sigma-Aldrich) was added to the artificial urine at a concentration of 3.2 μM, to label precipitated minerals. Calcein specifically binds to divalent metals, such as calcium and magnesium, and is incorporated into calcium and magnesium minerals during precipitation (29). Residual calcein was rinsed from the bulk fluid with 1 ml of water before SYTO 62 staining and imaging procedures. Confocal imaging was performed at three different locations in the flow cell for each experiment.

Ciprofloxacin treatment.

Twenty-four-hour-old ure+ and ure− biofilms were subjected to ciprofloxacin treatment at ciprofloxacin concentrations of 20 and 100 μg/ml, under both static and flowthrough conditions. These concentrations are comparable to ciprofloxacin levels in urine after regular oral-uptake doses (30). For static ciprofloxacin treatment, 1 ml of ciprofloxacin solution was slowly injected into the flow cell and maintained without flow for 2 h. For flowthrough ciprofloxacin treatment, 100 μg/ml ciprofloxacin was continuously supplied to the flow cell at a flow rate of 10 ml/h, through the peristaltic pump. Both static and flowthrough ciprofloxacin treatments were tested in triplicate in independent experiments. Ciprofloxacin stock solution was prepared at a concentration of 10 mg/ml and stored in a −20°C freezer before being diluted for injection into the flow cells.

Live/dead staining was used to evaluate biofilm killing resulting from ciprofloxacin exposure. A 1-ml mixture containing 6.7 μM SYTO 9 and 100 μM propidium iodide (PI) (FilmTracer LIVE/DEAD biofilm viability kit; Life Technology) was carefully injected into the flow cell, and biofilms were stained for 15 min. The unbound stains were then rinsed out with 1 ml of water before imaging. Due to the overlap between SYTO 9 and calcein fluorescence, biomineral deposits were not labeled in experiments with ciprofloxacin treatment.

Detachment of both live and dead cells during flowthrough ciprofloxacin treatment was analyzed using flow cytometry and direct microscopic counting. Flow cell effluent samples were collected for live/dead analyses 0.5 and 2 h after the initiation of flowthrough ciprofloxacin treatment. These antimicrobial killing experiments were replicated independently in triplicate. SYTO 9 and PI were added to 1 ml of flow cell effluent to final concentrations of 3.4 μM and 33 μM, respectively. Stained effluent samples were run through a flow cytometer (BD LSR II) with a 488-nm laser, in log mode, until at least 10,000 events were recorded. For direct microscopy, stained effluent was centrifuged at 9,193 × g for 1 min. The supernatant was then carefully removed, and the cell pellet was resuspended in 100 μl phosphate-buffered saline (PBS). For microscopic counting, 10 μl of the cell resuspension was transferred to a glass slide with a coverslip and was observed under a Leica epifluorescence microscope with a 20× objective. Images were obtained using an EXi Aqua Bio-Imaging camera and MetaMorph software. At least three different locations were imaged for each sample.

Transport of Cy5-conjugated ciprofloxacin.

The transport of Cy5-conjugated ciprofloxacin (Cy5-cipro) (Bio-Synthesis Inc.) was observed in 24-h ure+ and ure− biofilms, following methods reported previously (17). Cy5-cipro stock solution (1 mg/ml) was prepared and stored in a −20°C freezer. Ure+ and ure− biofilms were stained with SYTO 9 and imaged with confocal imaging before the introduction of Cy5-cipro; 1 ml of 20 μg/ml Cy5-cipro was then slowly injected into the flow cell and retained under static conditions for 30 min. The penetration of Cy5-cipro into the biofilm was imaged using methods described previously (17), with the flow cell fixed on the confocal microscope stage during the injection to enable direct registration of SYTO9 and Cy5-cipro data within a single field of view. In separate experiments, we observed the penetration of Cy5-cipro in ure+ biofilms with biomineral deposits labeled by calcein but with biomass not labeled due to the overlap of SYTO 9 and calcein fluorescence, as described previously.

The transport of Cy5-cipro in ure+ biofilms was also visualized under flowthrough conditions, with Cy5-cipro being continuously delivered into the flow cell at a concentration of 20 μg/ml and a flow rate of 10 ml/h. In this experiment, time-lapse confocal z-stack images of the same field of view were taken every 5 min for 30 min.

Image processing.

Three-dimensional (3D) images were generated from confocal image stacks using the Volocity software package (PerkinElmer, Inc.). Biofilm biomass and live/dead fractions were quantified using the BioSPA (Biofilm Spatial Pattern Analysis) software package running in Matlab. Live and dead cells in flow cell effluent were imaged by epifluorescence microscopy and counted using the particle analysis function in ImageJ (NIH).

RESULTS

The development of P. mirabilis wild-type (ure+) and urease-negative mutant (ure−) biofilms in artificial urine was observed in a flow cell (see Materials and Methods). After 24 h of development, ure+ biofilms produced fine mineral precipitates that were homogeneously distributed throughout the biofilm (Fig. 1; also see Fig. S1 in the supplemental material). This mineral precipitation caused the biofilm biomass to be discontinuous, with small colonies scattered on the surface (see Fig. S1). The main component of the mineral precipitates was identified as magnesium ammonium phosphate (struvite) by means of confocal Raman spectroscopy (see the text and Fig. S3 in the supplemental material). Time series observations showed that biomass accumulated faster than minerals in the early stage of ure+ biofilm formation (before 12 h) (Fig. 1). However, biomineralization accelerated greatly starting at 18 h, causing the biomineral deposits to occupy 78% of the total volume of the biofilm by 24 h (Fig. 1b). Unlike the wild-type biofilms, ure− biofilms did not induce biomineralization and formed dense and continuous lawns of bacteria on the surface after 24 h (Fig. 2; also see Fig. S2 in the supplemental material). The mechanism of biomineralization was confirmed to be the conventional process associated with increasing pH induced by ureolytic metabolism during biofilm development (31). Ure+ biofilm growth increased the bulk pH from 6.8 to 9.0 in 24 h, while the pH did not change during ure− biofilm development, as expected because of the lack of ureolytic metabolism in this mutant strain (see Fig. S4 in the supplemental material). These results show that biomineralization induced by ureolytic metabolism strongly regulated P. mirabilis biofilm development and morphology.

FIG 1.

Development of P. mirabilis (ure+) biofilms in artificial urine. (a) Confocal micrographs obtained at 6, 12, 18, and 24 h. Red, biofilm biomass stained with SYTO 62; blue, biomineral deposits stained with calcein. The component orthogonal planar images of these 3D renderings are presented in Fig. S1 in the supplemental material. (b) Growth of biomass and biomineral deposits over time. The volume of biofilm biomass was greater than the volume occupied by biomineral deposits for the first 18 h of biofilm development, but the biomineralization accelerated greatly between 18 and 24 h and mineral deposits were the predominant component of the biofilm after 24 h.

FIG 2.

Development of P. mirabilis urease-negative (ure−) biofilms in artificial urine. (a) Biomass stained by SYTO 62 (red). The component orthogonal planar images of these 3D renderings are presented in Fig. S2 in the supplemental material. (b) Growth of biomass over time. Ure− produced high-cell-density biofilms without biomineralization.

We exposed 24-h ure+ and ure− biofilms to ciprofloxacin to test the effects of biomineralization on biofilm susceptibility to a clinically important antimicrobial. With static treatment (see Materials and Methods), biomineralization decreased biofilm killing by ciprofloxacin (Fig. 3 and 4). Live/dead staining showed greater killing of ure− biofilms than ure+ biofilms with ciprofloxacin concentrations of both 20 and 100 μg/ml (P < 0.05). Dead cells were sparse and heterogeneous in ure+ biofilms but were homogeneously distributed throughout ure− biofilms (Fig. 3). Overall, 20 μg/ml ciprofloxacin resulted in 18.7 ± 5.0% killing of ure+ biofilms and 31.5 ± 1.2% killing of ure− biofilms, while 100 μg/ml ciprofloxacin resulted in 22.8 ± 0.5% killing of ure+ biofilms and 42.1 ± 7.3% killing of ure− biofilms (Fig. 4), despite similar total biofilm biomass values. We conducted control experiments to confirm that the reduced susceptibility of the biomineralized biofilms resulted specifically from biomineralization induced by biofilm metabolism. Planktonic ure+ and ure− cells exhibited similar susceptibilities to ciprofloxacin at a wide range of pH values (see the text and Fig. S5 and S6 in the supplemental material). Further, ure+ and ure− biofilms grown in 3% LB medium, which supports P. mirabilis growth but have little urea and lower salt concentrations than the artificial urine and do not yield biomineralization, exhibited similar lawn morphologies and ciprofloxacin killing levels (see Fig. S7 in the supplemental material). Together, these results indicate that urease-induced biomineralization was responsible for the reduced susceptibility of ure+ biofilms to ciprofloxacin.

FIG 3.

Killing of 24-h ure+ and ure− biofilms following 2 h of static ciprofloxacin treatment. (a) Confocal micrographs of ciprofloxacin-treated biofilms with live/dead staining. Green, live cells; red, dead cells. (b) Killing heat maps. Warm colors indicate high levels of local killing. Ciprofloxacin produced greater overall killing and more homogeneous killing in ure− biofilms (mucoid) than in ure+ biofilms (biomineralized).

FIG 4.

Quantification of biofilm killing after 2 h of static ciprofloxacin treatment. Bars, live and dead biomass values observed following exposure to ciprofloxacin; numbers above each pair of bars, overall killing percentages (dead biomass/total biomass). Killing levels were significantly higher in ure− biofilms than in ure+ biofilms with both ciprofloxacin concentrations tested (P < 0.05 for both concentrations). Results are means ± 1 standard deviation from triplicate experiments.

Biomineralized deposits can alter the delivery of antimicrobials to cells buried within the biofilm by deforming the biofilm structure (20) and forming transport barriers (32, 33). To investigate whether such transport limitations were responsible for the reduced susceptibility of ure+ biofilms to ciprofloxacin that we observed, we visualized the transport of Cy5-conjugated ciprofloxacin (Cy5-cipro) into 24-h ure+ and ure− biofilms. The transport of Cy5-cipro was greatly hindered in ure+ biofilms (Fig. 5). After 30 min of static treatment, Cy5-cipro did not penetrate significantly into ure+ biofilms, and the limited penetration that was observed was highly localized (Fig. 5). This pattern of limited and localized penetration is similar to the killing patterns observed in ure+ biofilms (Fig. 3). In contrast, Cy5-cipro readily penetrated throughout ure− biofilms (Fig. 5). Based on these results, we conclude that hindered transport of ciprofloxacin through biomineralized deposits contributed to the reduced susceptibility of ure+ biofilms.

FIG 5.

Penetration of Cy5-cipro in ure+ and ure− biofilms. Magenta, Cy5-cipro; green, biofilm biomass; blue, biomineral deposits (note that biomineralization occurred only in ure+ biofilms); white, overlay of biomass (green) and Cy5-cipro (magenta). Left, 3D opacity views; right, orthogonal planar views. Ciprofloxacin transport was greatly hindered in ure+ biofilms with biomineral deposits, whereas ciprofloxacin fully penetrated ure− biofilms lacking biomineral deposits.

To evaluate further the effects of transport limitations on biofilm susceptibility, we repeated ciprofloxacin exposure under flowthrough conditions (which increase solute transport into and within biofilms [34]); 100 μg/ml ciprofloxacin was continuously supplied to 24-h biofilms for 2 h at a flow rate of 10 ml/h. Significantly greater overall killing of ure+ biofilms was observed under flowthrough exposure conditions (47.1 ± 7.1%) than under static exposure conditions (22.8 ± 0.5%). Further, Cy5-cipro fully penetrated ure+ biofilms in 30 min under flowthrough conditions (see Fig. S8 in the supplemental material), whereas very little penetration of the biofilms was observed under static conditions (Fig. 5). These results provide further support for the conclusion that biomineralization limits ciprofloxacin penetration and resulting biofilm killing, and they also indicate that enhanced transport (here controlled by influent flow conditions) increases the effectiveness of ciprofloxacin in killing biofilms.

We also observed a minor increase in killing of ure− biofilms with flowthrough ciprofloxacin exposure (46.7 ± 10.8%), relative to static exposure (42.1 ± 7.3%). However, flowthrough ciprofloxacin exposure also triggered significant detachment of biofilm biomass, and this detachment was significantly greater in ure− biofilms than in ure+ biofilms (P < 0.01) (Fig. 6). Over 90% of ure− biofilm biomass detached within 2 h of flowthrough exposure to ciprofloxacin, while less than 20% of ure+ biofilm biomass detached under the same conditions. Further, after 24 h of flowthrough ciprofloxacin exposure, ure− biofilms maintained only 3.6% of their initial biomass, while ure+ biofilms maintained 16.9% of their initial biomass (Fig. 6b). Killing of detached cells was analyzed both by direct microscopy and by flow cytometry. Direct counting and flow cytometry yielded comparable ratios of dead and live cells, with an average of 20.9 ± 7.8% and 20.2 ± 4.6% killing in effluents from ure+ and ure− biofilms, respectively (see Table S1 and Fig. S9 and S10 in the supplemental material). This observed killing of detached cells was significantly less than the killing of cells remaining in the biofilm (47.1% and 46.7% for ure+ and ure− biofilms, respectively). Beyond protecting biofilm-resident cells from antimicrobial killing, these results indicate that biomineralization stabilized biofilms to enhance the retention of biofilm biomass during antimicrobial treatment.

FIG 6.

Killing and biomass reductions in ure+ and ure− biofilms after flowthrough ciprofloxacin treatment. (a) Patterns of biofilm killing. Green, live cells; red, dead cells. Overall ciprofloxacin killing levels were similar in ure+ and ure− biofilms (47.1 ± 7.1% and 46.7 ± 10.8%, respectively). (b) Biomass reduction. Total biomass (live plus dead) was quantified before and after flowthrough ciprofloxacin (cipro) treatment. Ure− biofilms lost more biomass during the antimicrobial treatment than did ure+ biofilms.

DISCUSSION

P. mirabilis biofilms and associated biomineral deposits have commonly been imaged using electron microscopy (12, 35). This method reveals surface structures and cell-mineral associations but does not resolve spatial patterns of biofilm biomass and biomineral deposits in three dimensions. We combined fluorescent labeling and confocal imaging to resolve successfully the morphology of biomineralized biofilms in three dimensions and to quantify the amounts and distributions of cells and mineral deposits throughout the biofilms (Fig. 1). We observed that the urease-positive (ure+) P. mirabilis wild-type strain induced struvite biomineralization in artificial urine, producing biofilms that were morphologically distinct from those produced by the ure− mutant (Fig. 1 and 2). These results match those of many prior studies that demonstrated that urease production and urea hydrolysis play crucial roles in biomineralization and crystalline biofilm development both in vitro and in CAUTIs (11, 23). With detailed temporal observations, we found that biomass growth dominated the early stage of biofilm development, and mineral formation accelerated dramatically only after the bulk pH rose (Fig. 1; also see Fig. S4 in the supplemental material). After 24 h of growth in artificial urine, P. mirabilis ure+ biofilms were primarily composed of biomineral deposits and had substantially different overall morphology, compared to ure− biofilms, which did not induce biomineralization (Fig. 1). Conversely, ure+ and ure− biofilms developed similar morphologies when grown in 3% LB medium, which does not contain urea (see Fig. S6 in the supplemental material), indicating that the specific mutation in the ure− strain that renders it unable to produce urease does not impair its ability to form biofilms.

The distinctive morphologies and compositions of ure+ and ure− biofilms yielded dramatic differences in biofilm killing by ciprofloxacin. We found that ure+ biofilms were much less susceptible to ciprofloxacin than were ure− biofilms (Fig. 4). By visualizing the transport of Cy5-cipro, we showed that ureolytic biomineralization decreased biofilm susceptibility by impeding the penetration of the antimicrobial into the biofilm (Fig. 5). These findings are consistent with prior observations that transport limitations largely account for biofilm tolerance to antimicrobials (18, 36) and biocides (20, 37). We observed fast penetration of Cy5-cipro in ure− biofilms without biomineralization (Fig. 5), similar to the rapid ciprofloxacin penetration observed previously in Pseudomonas aeruginosa biofilms (17). Higher Cy5 intensity was observed in biofilms than in bulk flow, indicating that these biofilms accumulated ciprofloxacin (Fig. 5 and reference 17). Biomineralized ure+ biofilms, however, had hindered penetration and did not accumulate ciprofloxacin because of the transport barrier created by the accumulation of ureolytic struvite deposits within the biofilm (Fig. 5). We observed greater ciprofloxacin penetration and biofilm killing after flowthrough ciprofloxacin treatment (Fig. 6; also see Fig. S7 in the supplemental material), as expected because flow generally enhances solute transport in biofilms (34). These findings together show a strong link between antimicrobial transport and biofilm killing. Since internal transport limitation is recognized as a major cause of biofilm physiological heterogeneity (38, 39), biomineralization is expected to further regulate biofilm complexity by changing nutrient and oxygen distributions. Therefore, while we clearly demonstrated here that biomineralization hindered delivery of ciprofloxacin, associated processes such as low levels of cellular metabolism due to nutritional limitations also could contribute to the observed reduction in antimicrobial susceptibility.

Beyond reducing antimicrobial killing, we also observed that ureolytic biomineralization stabilized biofilms and enhanced retention of viable cells on the surface (Fig. 6). Retained viable cells are expected to regrow following cessation of antimicrobial treatment (40), contributing to the persistence of infection. We showed that flowthrough ciprofloxacin treatment nearly eliminated nonbiomineralized P. mirabilis biofilms from the surface, but biomineralized biofilms retained significant numbers of viable cells after treatment. We also found that ∼80% of cells that detached during flowthrough ciprofloxacin treatment remained viable (see Table S1 in the supplemental material). Flow- or stress-induced detachment plays an important role in biofilm infections by disseminating large numbers of bacteria (41, 42). Once biofilms disperse, the released viable cells are able to spread the infection and recolonize surfaces (43).

The information we obtained regarding the structure and antimicrobial susceptibility of biomineralized P. mirabilis biofilms can be used to improve treatment of persistent CAUTIs. Our results clearly show that P. mirabilis biofilms become much more recalcitrant following biomineralization. Therefore, our results suggest that controlling mineral formation is critical for successfully treating CAUTIs. This explains the difficulties in eliminating CAUTIs via oral antimicrobial therapy that are commonly observed in clinical settings. The normal clinical approach to the lack of efficacy of antimicrobials against CAUTIs involving biomineralization is to replace the catheter, which significantly improves infection control outcomes (4, 44). However, catheter replacement is not desirable as a primary means of controlling CAUTIs, because it causes patient discomfort, incurs health risks when blocked catheters are not removed promptly, and requires significant effort on the part of health care providers (45). Further, CAUTIs are often found to recur chronically (46), especially in long-term care, suggesting that recalcitrant biomineralized biofilms occur outside the catheter itself (12) and provide a nidus for catheter reinfection. This idea was confirmed by clinical evidence showing that most patients with recurrent catheter encrustation had bladder stones that harbored P. mirabilis (47). As an alternative, it has been observed in clinical settings that increased fluid intake can slow the rate of catheter encrustation by diluting salt concentrations in the urine (48). Our results further suggest that reducing biomineralization would render biofilms more susceptible to killing, indicating that control of catheter encrustation and biofilm formation elsewhere in the urinary tract is critical not only to the longevity of catheters but also to the effectiveness of antimicrobial therapies for CAUTIs.

Overall, our findings strongly suggest that ureolytic biomineralization complicates CAUTIs and their treatment by increasing biofilm mechanical stability and decreasing biofilm susceptibility to antimicrobials. Both factors suggest that simple antimicrobial therapy is likely to be insufficient for treating biomineralized biofilm infections, and additional strategies should be used to hinder biomineralization, disrupt existing biomineral deposits, or otherwise increase the penetration of antimicrobials into biomineralized biofilms.

Funding Statement

This work was supported by grant R01AI081983 from the National Institutes of Health, National Institute of Allergy and Infectious Diseases.