Error bars indicate SD.
D1 is a known Psl-positive strain PA715, and D21 is a PA715_psl-pr isogenic mutant of PA715 with a deletion in the psl promoter. mAb indicates monoclonal antibody; NA, not applicable.
The median biofilm formation is 1.03. Each point represents a study visit and the mean visual acuity for patients. Rate of vision gain over time is not statistically different (P = .67).
eFigure 1. Planktonic Growth During 12 Hours in LB Medium Expressed in Colony-forming Units of 15 SCUT P aeruginosa Isolates With the 5 Highest, 5 Lowest, and 5 of the Average Level of Biofilm Formation (3 Technical Replicates per Time Point)
eFigure 2. LogMAR Visual Acuity vs Time by Psl EPS Expression Dichotomized (Median Psl Formation = 0.86)
Zegans ME, DiGiandomenico A, Ray K, Naimie A, Keller AE, Stover CK, Lalitha P, Srinivasan M, Acharya NR, Lietman TM. Association of Biofilm Formation, Psl Exopolysaccharide Expression, and Clinical Outcomes in Pseudomonas aeruginosa KeratitisAnalysis of Isolates in the Steroids for Corneal Ulcers Trial . JAMA Ophthalmol. 2016;134(4):383-389. doi:10.1001/jamaophthalmol.2015.5956
Bacterial virulence factors are increasingly recognized as important in the understanding of clinical infections.
To determine whether 2 potential virulence factors, in vitro biofilm formation and Psl exopolysaccharide (EPS) expression, influence clinical presentation or outcomes in Pseudomonas aeruginosa keratitis.
Design, Setting, and Participants
Laboratory investigation using P aeruginosa clinical isolates from the double-blind Steroids for Corneal Ulcers Trial (SCUT), which included patients at Aravind Eye Hospital, Proctor Foundation, University of California, San Francisco, and Dartmouth-Hitchcock Medical Center. SCUT was conducted from September 1, 2006, through February 22, 2010. All data used in this study were obtained during this period. Pseudomonas aeruginosa clinical isolates from SCUT were evaluated for in vitro biofilm formation, and Psl EPS expression was assessed using an anti-Psl monoclonal antibody (mAb) enzyme-linked immunosorbent assay. Planktonic growth kinetics and the susceptibility to anti-Psl mAb-mediated opsonophagocytic killing (OPK) were also evaluated in a subset of isolates. Linear regression assessed associations between SCUT patients’ visual acuity and their corresponding biofilm formation and Psl EPS expression. Generalized estimating equation regression models were used to assess whether the change in visual acuity among SCUT patients was associated with Psl EPS expression or biofilm formation.
Main Outcomes and Measures
Biofilm formation, Psl production, OPK, and visual acuity.
The P aeruginosa SCUT strains produced a mean (SD) in vitro biofilm score of 1.06 (0.32) (range 0.17-2.12). A 1-unit increase in biofilm was associated with a worse visual acuity of 2 lines measured in SCUT patients at baseline (0.20 logMAR; 95% CI, −0.03 to 0.44; P = .09) and 3 months (0.21 logMAR; 95% CI, 0.003 to 0.44; P = .047). Of 101 confirmed P aeruginosa SCUT isolates, 100 expressed Psl EPSs. In addition, all Psl-positive strains evaluated in the OPK assay were susceptible to anti-Psl mAb-mediated OPK.
Conclusions and Relevance
The ability of P aeruginosa keratitis isolates to form biofilms in vitro was correlated with worse vision at presentation and after 3 months in SCUT. Ninety-nine percent of P aeruginosa keratitis isolates from SCUT produced Psl EPSs, and 100% of these evaluated Psl-positive isolates were susceptible to anti-Psl mAb-mediated OPK. These data indicate that biofilm formation and Psl EPSs may be candidate targets for novel therapeutics against P aeruginosa keratitis.
Pseudomonas aeruginosa is an opportunistic gram-negative pathogen and a frequent cause of bacterial keratitis. It is the most common gram-negative pathogen isolated in many keratitis series and often the first or second most common pathogen overall.1- 4 The pathogenesis of P aeruginosa infection is mediated through the presence of numerous cell-associated and secreted virulence factors and the ability to form antibiotic, recalcitrant biofilms.5- 9 Biofilm formation refers to bacterial communities that grow attached to biotic or abiotic surfaces in contrast to planktonic or free-living growth.10 Bacterial biofilm formation on medical devices, such as contact lenses, scleral buckles, indwelling catheters, and artificial joints, is a common cause of serious persistent infections.11- 13 The role of biofilm formation in bacterial keratitis is complex and incompletely understood. Biofilm formation on intraocular lenses,14 contact lenses,15,16 and storage cases15 is well documented; however, whether biofilm formation takes place on the biotic surfaces of the cornea and whether in vitro measures of biofilm formation are associated with the severity of clinical disease remains unclear.
Bacteria produce exopolysaccharides (EPSs), which are secreted sugars that play a complex role in bacterial physiology and pathogenesis.17Pseudomonas aeruginosa is capable of producing at least 3 EPSs: Psl, Pel, and alginate.17- 19 Some P aeruginosa strains produce only 1 form of EPS, whereas others are capable of producing multiple EPSs.17 Regulation of EPS production appears to participate in the pathogenesis of specific infections. For instance, high alginate production is observed from phenotypic variant mucoid strains chronically infecting the lungs of patients with cystic fibrosis, suggesting that EPSs can help bacteria adapt to persistent growth in particular environments.19- 21
To our knowledge, few studies have focused on the EPS types associated with P aeruginosa keratitis because of the limited reagents capable of characterizing EPS types available for high-throughput screens. However, given the development of monoclonal antibodies (mAbs) targeting Psl, reliable analysis of Psl expression on larger numbers of clinical isolates is now possible.22,23 The unique primary sugar backbone structure of Psl EPS is composed of mannose, glucose, and rhamnose, and its biosynthesis requires a previously reported 15-gene operon.24- 26 Psl EPS has been widely studied as a colonization and persistence factor in multiple types of P aeruginosa infections.27,28
In the present study, we surveyed P aeruginosa strains from the National Eye Institute–sponsored Steroids for Cornea Ulcers Trial (SCUT)29 because we have standardized clinical measures associated with these isolates, thus allowing an analysis of the association of these disease measures with in vitro measures of biofilm production and Psl EPS expression. We assessed the ability of P aeruginosa strains to form abiotic biofilms and produce Psl EPSs and compared these findings to clinical outcomes for evaluation of potential correlations.
Question: Is there a correlation between biofilm formation and production of the exopolysaccharide Psl in Pseudomonas aeruginosa keratitis isolates from the Steroids for Corneal Ulcer Trial (SCUT) with clinical outcomes in the patients infected with these isolates?
Findings: Robust in vitro biofilm formation is associated with worse vision at presentation and after treatment. A total of 99% of SCUT P aeruginosa keratitis isolates produced the exopolysaccharide Psl and were susceptible to killing by anti-Psl monoclonal antibodies.
Meaning: Biofilm formation and Psl may play a role in virulence in patients with P aeruginosa keratitis and thus may be relevant therapeutic targets for future treatments.
Institutional review board approval for SCUT was granted by the Aravind Eye Care System’s Institutional Review Board, the Dartmouth-Hitchcock Medical Center Committee for the Protection of Human Subjects, and the University of California, San Francisco, Committee on Human Research. Written informed consent was obtained from all study participants. The trial adhered to the Health Insurance Portability and Accountability Act and the tenets of the Declaration of Helsinki and was registered at clinicaltrials.gov (NCT00324168).30 SCUT was conducted from September 1, 2006, through February 22, 2010. All data used in this study were obtained during this period.
The clinical isolates used, unless otherwise noted, came from SCUT or the laboratory strain collection in the laboratory of Professor Michael E. Zegans at the Geisel School of Medicine at Dartmouth. Bacterial planktonic cultures were grown in lysogeny broth (LB)31 at 37°C unless otherwise noted, and their growth was monitored at an optical density of 600 nm. Colonies of bacteria were grown on plates of LB agar or LB agar with no sodium unless otherwise noted.
All microbiologic assays in this study (biofilm formation, Psl EPS enzyme-linked immunosorbent assay [ELISA], and opsonophagocytic killing [OPK] assays) were performed before the statistical analysis, and laboratory investigators were masked to the clinical outcomes associated with the strains.
Biofilm assays in 96-well polyvinyl chloride microtiter plates with crystal violet staining of biofilms were performed as previously described.32,33 All biofilm cultures for the assays were incubated for 24 hours at 37°C in LB medium unless otherwise specified.34 The amount of biofilm present in each well was measured by solubilization of crystal violet with 30% acetic acid and quantified at an optical density of 550 nm. All biofilm data presented were collected from 1 representative experiment, with at least 4 replicates per strain, of a minimum of 3 independent assays. Biofilm formation is expressed as a percentage of biofilm formed by laboratory strain PAO1 analyzed on the same plate.
Bacterial strains were grown on LB agar overnight. Three colonies were picked and grown in LB liquid. These cultures were normalized to an optical density of 600 nm. ELISA was performed as previously described.22 Primary antibody WapR-001 was used at 0.1 μg/mL22 followed by donkey antihuman horseradish peroxidase–conjugated secondary antibody (1:5000) (Jackson Immuno Research Laboratories Inc). TMB SureBlue microwell peroxidase substrate (Kirkegaard & Perry Laboratories Inc) was used for development, and 0.2N acetic acid was used to stop the reaction. Laboratory strains PA14 and PAO1 were used as positive and negative controls, respectively, for Psl EPS expression. A negative control omitting the primary antibody was performed for each isolate. The 3 biological replicates were tested the same day in triplicate. ELISA results were expressed as a percentage of the PAO1 results analyzed on the same plate. Psl EPS–positive strains were defined as those with at least 5% of PAO1 ELISA results. Strains that produced ELISA results of less than 5% of PAO1 were considered Psl negative.
We performed OPK assays as previously described with the investigator masked to the Psl expression status of each isolate.22 When possible, luminescent P aeruginosa isolates were constructed as described to facilitate rapid anti-Psl mAb–mediated OPK analysis.22 Experiments were performed in white 96-well plates (Nunc; Thermo Scientific) using 0.025 mL of each OPK component: P aeruginosa strains from log-phase cultures diluted to approximately 2 × 106 cells/mL, diluted baby rabbit serum (1:10), differentiated HL-60 cells (2 × 107 cells/mL), and mAb. After mixture of all components, plates were incubated at 37°C for 2 hours while shaking at 250 rpm. For nonluminescent P aeruginosa, colony-forming units per milliliter from each treatment condition were enumerated after overnight incubation at 37°C on LB agar plates. For luminescent OPK assays, determination of relative luciferase units (RLUs) was performed using an Envision Multilabel Plate Reader (PerkinElmer). The percentage of killing was determined by comparing the number of colonies or RLUs derived from assays that lacked mAb with the number of colonies or RLUs obtained from assays with anti-Psl mAbs or the control R347 mAb.
Baseline characteristics were compared between upper and lower medians of biofilm and Psl using the Fisher exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. Simple linear regression was used to assess the association between Psl or biofilm formation and visual acuity measured at baseline and 3 months in SCUT patients. In addition, generalized estimating equation (GEE) regression models with an exchangeable correlation structure for multiple visual acuity measurements (taken at baseline, 3 weeks, 3 months, and 12 months) were used to assess the association between biomarkers and the rate of visual acuity gain, clustered by patient.
Pseudomonas aeruginosa isolates were evaluated for their ability to form abiotic biofilms in a microtiter plate assay. In vitro biofilm formation among SCUT P aeruginosa isolates, as a percentage of the PAO1 control, varied from 17% to 242% (Figure 1). SCUT strains produced a mean (SD) number of biofilms in vitro of 1.06 (0.32) (range, 0.17-3.12). To assess whether this variability reflected a generalized deficit in bacterial growth in liquid culture, 15 strains were analyzed for planktonic growth in liquid culture; 5 strains were low biofilm formers, 5 were intermediate biofilm formers, and 5 were high biofilm formers. In contrast to growth in a biofilm, planktonic growth occurs when bacterial cells are not attached to a surface. Despite wide variations among strains in sessile biofilm growth, no difference in the growth kinetics of the isolates under planktonic conditions was observed, indicating that the variability in biofilm formation was not related to deficient planktonic growth (eFigure 1 in the Supplement).
A total of 101 SCUT P aeruginosa isolates were evaluated for their ability to express Psl EPSs by ELISA using an anti-Psl mAb, WapR-001.22 Results were validated with controls, including strains that have been previously classified as Psl positive (laboratory strain PAO1) or Psl negative (laboratory strain PA14).35 We tested a previously described ocular clinical isolate, PA715, and an isogenic mutant with a deletion in the Psl promoter PA715 Δpsl-pr.35 As expected, control strains known to produce Psl were ELISA positive, and strains previously established to be Psl negative did not react with the anti-Psl mAb. ELISA results were normalized to strain PAO1 in which Psl-positive strains were defined as those with Psl reactivity of 5% or greater of PAO1. By these criteria, 100 of the 101 SCUT isolates were found to express Psl, indicating that this EPS is highly prevalent among keratitis isolates (Figure 1). Overall, anti-Psl mAb reactivity of SCUT isolates ranged from 1.6% to 176.1% to that of PAO1.
We next sought to evaluate the ability of an anti-Psl mAb, Psl0096, to mediate OPK of selected P aeruginosa isolates.23 We chose 19 SCUT strains with Psl ELISA reactivity that ranged from 1.6% to 176.1% to that of PAO1. PA715 and PA715 Δpsl-pr served as positive and negative controls, respectively.35 As expected, PA715 but not PA715 Δpsl-pr was susceptible to OPK. Susceptibility to killing in the OPK assay was observed in 18 of the 19 SCUT isolates (Figure 2). Of interest, Psl-positive strains with reactivity as low as 11.7% to that of PAO1 were susceptible to anti-Psl mAb-targeted killing. The only isolate that was not killed in the OPK assay had the lowest anti-Psl mAb reactivity, which was 1.6% of PAO1 reactivity. These data support our criteria to consider strains of 5% or greater of PAO1 as Psl positive. Together with the overall ELISA results, these data suggest that most SCUT isolates produce Psl EPSs and are susceptible to anti-Psl mAb-mediated OPK.
We investigated the association between in vitro biofilm formation and the clinical outcomes of patients in SCUT. Table 1 summarizes the baseline clinical characteristics for SCUT patients among the upper and lower median values of their in vitro biofilm formation as a percentage of the PAO1 control. A 1-unit increase in biofilm formation was associated with 0.20 worse logMAR acuity measured at the baseline visit (0.20 logMAR; 2 Snellen lines; P = .09) and the 3-month visit (0.21 logMAR; 2.1 Snellen lines; P = .047) in SCUT patients (Table 2). The GEE regression model indicated that there was not a statistically significant difference in the rate of visual acuity improvement during the 12-month study period between the isolates with higher and lower levels of biofilm formation (P = .67). Figure 3 shows that the visual gain over time is similar between the upper and lower biofilm formation groups, indicating that worse visual acuity at later visits in the higher biofilm group is associated with the underlying poor vision at baseline. Although the finding was not statistically significant, the duration of symptoms and infiltrate size were numerically greater in the high biofilm formation group (≥1.03) than in the low biofilm formation group (<1.03).
Although the finding was not statistically significant, a 1-unit increase in Psl EPS expression was associated with 0.28 better logMAR acuity at baseline (−0.28 logMAR; 2.8 Snellen lines; P = .16) and 0.14 better logMAR acuity at 3 months (−0.14 logMAR; 1.4 Snellen lines; P = .48) in SCUT patients. eFigure 2 in the Supplement reveals that the association between visual gain over time is different between the upper and lower halves of the Psl EPS expression group. Although the finding was not statistically significant, the patients with Psl EPS expression greater than 0.86 had better visual acuity throughout the study and improved at a faster rate than their counterparts who had Psl EPS expression less than 0.86 (P = .06).
We analyzed 101 P aeruginosa isolates from SCUT and found that strains with greater in vitro biofilm formation were associated with worse presenting vision (P = .09) and worse vision at 3 months (P = .05) (Table 2 and Figure 3). The GEE regression model, which included all visual acuity measurements, did not reveal a difference in the rate of visual acuity improvement between the 2 biofilm groups, indicating that the vision at presentation was the main driver of the differences in visual outcomes. The relative ability to form robust biofilms was not attributable to generalized slower or faster planktonic growth kinetics in liquid broth among low, medium, and high biofilm formers. These data indicate that the degree of biofilm formation is an independent predictor of the clinical severity of the infection.
We found that 100 (99.0%) of the 101 isolates expressed Psl, an abundant surface-associated EPS implicated in host cell attachment and immune evasion and a component in biofilm formation and maintenance in nonmucoid and mucoid P aeruginosa.17,27,36 There was no effect of corticosteroid treatment on visual acuity within the higher and lower subgroups of Psl (P = .81) and biofilm (P = .48). Of importance, we also observed potent anti-Psl mAb-mediated OPK against 18 of 19 strains that were evaluated, even against a low Psl-expressing strain that exhibited only 11.7% of PAO1 Psl expression (Figure 2). The only isolate that was not killed in the OPK assay had the lowest anti-Psl mAb reactivity (1.6% of PAO1 reactivity), which we would classify as a Psl-negative strain. Opsonophagocytic killing has been widely used to test the functional capacities of therapeutic antibodies.37,38 Opsonophagocytic killing demonstrates the ability of such antibodies to attach to a pathogen, fix complement, and drive uptake and killing by host effector cells. The same antibodies have demonstrated efficacy in preventing P aeruginosa infection in a variety of animal disease models, including P aeruginosa keratitis.22,23 Taken together, our data indicate that most P aeruginosa keratitis isolates express Psl EPSs and are susceptible to mAb-targeted killing, suggesting the potential utility of anti-Psl mAbs in the treatment of P aeruginosa keratitis.
Biofilm formation has been implicated as a persistence factor that is associated with antibiotic resistance in a variety of clinical infections.10,11,21,39,40 Most often, it is associated with growth on abiotic implants, such as contact lenses or indwelling catheters. These abiotic surfaces are more vulnerable to bacterial colonization than biotic surfaces because they lack antimicrobial defenses (eg, the epithelial barrier, lysozyme, lactoferrin, and cationic peptides) of the innate immune system. However, only 8 of 500 patients in SCUT presented with keratitis associated with contact lens use, and of our 101 P aeruginosa–infected patients, 3 with high biofilm formation (≥1.03) and 1 with low biofilm formation (<1.03) wore contacts (P = .62). Thus, we cannot attribute the worse clinical outcomes associated with strains capable of forming robust biofilms to attachment to contact lenses or cases because they were not used by most patients.
In the absence of a large number of SCUT patients using contact lenses, 2 explanations of the association of biofilm formation with worse clinical outcomes seem reasonable. First, biofilm formation may occur on the cornea itself in P aeruginosa keratitis. In other words, strains that form robust biofilms in the microtiter assay also form biofilms on the cornea, leading to a more severe clinical course. Biofilm formation on biotic surfaces has been demonstrated in some infections, such as P aeruginosa pneumonia in patients with cystic fibrosis21 and in infectious crystalline keratopathy.41 Studies42,43 comparing biofilm formation in vitro on abiotic surfaces to biofilm formation on biotic surfaces, such as cultured epithelium, suggest that biofilm formation on abiotic surfaces share some but not all of the characteristics and molecular requirements of biofilm formation on biotic surfaces.
Alternatively, one may argue that biofilms, as measured in a microtiter dish assay, are a surrogate marker for other factors, such as pili and flagellum expression, which are known to play a role in both biofilm formation and virulence.44 Our current data do not allow us to distinguish between these possibilities, but these are noteworthy questions for future investigations. In either case, our data present a unique association between in vitro data on biofilm formation with worse clinical outcomes using a large body of prospective clinical data.
Although Psl is believed to be important for biofilm formation, we did not identify a positive correlation between increased levels of Psl expression and increased biofilm formation. Because other EPSs, such as Pel, can substitute for Psl to support biofilm formation, a Psl-negative or low-producing strain may have ample Pel or alginate production. In addition, Psl reactivity may not correlate with absolute levels of Psl production, potentially obscuring the association between Psl levels on biofilm formation. Finally, in addition to Psl EPS, other factors, such as pili and flagellum, and other adhesion factors important to biofilm formation influence the degree of biofilm formation in our assay.44 What is clear is that, in our study, Psl reactivity was not a surrogate measure of biofilm formation.
Limitations of the study include the fact that it was retrospective in nature, with the laboratory analysis occurring after the clinical data were collected. However, the laboratory investigations were masked to the clinical results while they were being performed. In addition, most of the samples came from patients in India who had bacterial keratitis not associated with contact lens wear. It is possible that the results would be different in a group of patients with contact lens–associated P aeruginosa keratitis isolates or in patients from a different geographic region. As discussed above, we did not establish whether biofilm formation occurred on the cornea during infection. Finally, the true efficacy of mAbs directed against Psl in corneal infection awaits further studies in animal models.
In summary, using P aeruginosa keratitis isolates and clinical data from SCUT, we found that the ability to form biofilms in vitro was correlated with worse vision at presentation and 3 months. A total of 100 (99.0%) of 101 P aeruginosa keratitis isolates from SCUT produce Psl EPSs, and of 18 Psl-positive strains tested 18 of 18 (100%) were susceptible to anti-Psl mAb-mediated OPK. These data indicate that biofilm formation and Psl EPS may be candidate targets for novel therapeutics against P aeruginosa keratitis.
Submitted for Publication: July 29, 2015; final revision received December 11, 2015; accepted December 12, 2015.
Corresponding Author: Michael E. Zegans, MD, Departments of Surgery (Ophthalmology) and Microbiology and Immunology, Geisel School of Medicine at Dartmouth, Hanover, NH 03756 (email@example.com).
Published Online: February 4, 2016. doi:10.1001/jamaophthalmol.2015.5956.
Author Contributions: Dr Zegans had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Zegans, DiGiandomenico, Naimie, Lietman, Srinivasan, Acharya.
Acquisition, analysis, or interpretation of data: Zegans, DiGiandomenico, Ray, Naimie, Keller, Stover, Lalitha, Archarya, Lietman.
Drafting of the manuscript: Zegans, DiGiandomenico, Naimie, Keller, Stover.
Critical revision of the manuscript for important intellectual content: Zegans, DiGiandomenico, Ray, Lalitha, Srinivasan, Acharya, Lietman.
Statistical analysis: Keller, Lietman.
Obtained funding: Zegans, DiGiandomenico, Stover.
Administrative, technical, or material support: Zegans, DiGiandomenico, Naimie, Keller, Lalitha, Acharya.
Study supervision: Zegans, DiGiandomenico.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr DiGiandomenico, Ms Keller, and Dr Stover reported being collaborating scientists on the project and employees of MedImmune/AstraZeneca and may own stock in that company. No other disclosures were reported.
Funding/Support: The National Eye Institute sponsored the Steroids for Corneal Ulcers Trial through grant U10-EY015114-01, which provided support for the conduct of SCUT and the collection, management, and analysis of the clinical data (Dr Lietman, principal investigator). The laboratory analysis (biofilm formation, Psl, and opsonophagocytic killing ) of the Pseudomonas aeruginosa isolates was supported by an investigator-initiated grant (Dr Zegans) from MedImmune.
Role of the Funder/Sponsor: The National Eye Institute did not control the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. MedImmune reviewed but did not have final approval of the manuscript.