Combining bacteriophage and vancomycin is efficacious against MRSA biofilm-like aggregates formed in synovial fluid

Abstract

Background: Biofilm formation is a major clinical challenge contributing to treatment failure of periprosthetic joint infection (PJI). Lytic bacteriophages (phages) can target biofilm associated bacteria at localized sites of infection. The aim of this study is to investigate whether combination therapy of phage and vancomycin is capable of clearing Staphylococcus aureus biofilm-like aggregates formed in human synovial fluid.

Methods: In this study, S. aureus BP043, a PJI clinical isolate was utilized. This strain is a methicillin-resistant S. aureus (MRSA) biofilm-former. Phage Remus, known to infect S. aureus, was selected for the treatment protocol. BP043 was grown as aggregates in human synovial fluid. The characterization of S. aureus aggregates was assessed for structure and size using scanning electron microscopy (SEM) and flow cytometry, respectively. Moreover, the formed aggregates were subsequently treated in vitro with: (a) phage Remus [∼10⁸ plaque-forming units (PFU)/ml], (b) vancomycin (500 μg/ml), or (c) phage Remus (∼10⁸ PFU/ml) followed by vancomycin (500 μg/ml), for 48 h. Bacterial survival was quantified by enumeration [colony-forming units (CFU)/ml]. The efficacy of phage and vancomycin against BP043 aggregates was assessed in vivo as individual treatments and in combination. The in vivo model utilized Galleria mellonella larvae which were infected with BP043 aggregates pre-formed in synovial fluid.

Results: Scanning electron microscopy (SEM) images and flow cytometry data demonstrated the ability of human synovial fluid to promote formation of S. aureus aggregates. Treatment with Remus resulted in significant reduction in viable S. aureus residing within the synovial fluid aggregates compared to the aggregates that did not receive Remus (p < 0.0001). Remus was more efficient in eliminating viable bacteria within the aggregates compared to vancomycin (p < 0.0001). Combination treatment of Remus followed by vancomycin was more efficacious in reducing bacterial load compared to using either Remus or vancomycin alone (p = 0.0023, p < 0.0001, respectively). When tested in vivo, this combination treatment also resulted in the highest survival rate (37%) 96 h post-treatment, compared to untreated larvae (3%; p < 0.0001).

Conclusion: We demonstrate that combining phage Remus and vancomycin led to synergistic interaction against MRSA biofilm-like aggregates in vitro and in vivo.

Keywords: Galleria mellonella; Staphylococcus aureus; aggregates; biofilm; peri-prosthetic joint infection; phage; synovial fluid; vancomycin.

FIGURE 1

Staphylococcus aureus BP043 aggregate formation in human synovial fluid (HSF) after incubation for 24 h at 37°C. (A) Macroscopic images of the S. aureus aggregates forming in human synovial fluid in 96-well plate, with and without Remus treatment. (B) Images taken with bright field and fluorescence microscopy (SYTO9 nucleic acid stain) at 20× for S. aureus planktonic cultures grown in TSB or S. aureus grown as aggregates in human synovial fluid (HSF), scale bar = 100 μm. (C) Scanning electron microscopy image of S. aureus human synovial fluid aggregates, scale bar = 10 μm.

FIGURE 2

Human synovial fluid (HSF) enhances aggregate formation compared to TSB. Flow cytometry data shows (A) bigger (median fluorescence intensity of forward scatter, MFI-FSC) and, (B) more aggregates formation in HSF compared to TSB. (C) P2 gate was set to show SYTO9 green positive bacterial cells. N = 3, mean ± SE. Unpaired, two-tailed t-test was utilized to determine statistical difference, *p < 0.05.

FIGURE 3

Biochemical disruption assay of Staphylococcus aureus BP043 synovial fluid aggregates. (A) Destruction of synovial fluid aggregates after adding proteinase K (150 μg/ml) for 1 h at 37°C. (B) No clear destruction of synovial fluid aggregates after adding DNase I (0.5 mg/ml dissolved in 2 mM MgCl₂) for 24 h at 37^°C. (C) Proteinase K treatment resulted in the release of more bacterial cells and increased the CFU compared to no treatment (1x PBS), mean ± SE, N = 4 pools of synovial fluids. (D) DNase I treatment did not affect the bacterial load indicating the extracellular DNA is probably not a major component of the synovial fluid aggregates, ± SE, N = 3 pools of synovial fluids. Unpaired, two-tailed t-test was utilized to determine statistical difference, ^**p < 0.01, ns, non-significant.

FIGURE 4

Fluorescence staining and microscopy examination of synovial fluid-derived Staphylococcus aureus aggregates to detect the matrix composition. Aggregates were stained with SYTO9 (green) for nucleic acid (cellular biomass), SYPRO (red) for proteins and WGA (blue) for GlcNAc (carbohydrates). Scale bar = 100 μm.

FIGURE 5

Lytic activities of phage Remus against Staphylococcus aureus (BP043) planktonic cells (∼10⁸ CFU/ml) in TSB with Remus at 10-fold changes of MOIs ranging from 10 to 10^{– 6}. Phage Remus was applied right at the start of the experiment. Cell density was assessed by measuring optical densities at 600 nm every 120 min for 48 h. N = 2, mean ± SE. One-way ANOVA (Tukey’s multiple comparisons test) was utilized to determine statistical difference at 48 h. Asterisks show comparison between the different MOIs and the media blank (no bacteria). *p < 0.01, ***p < 0.001, ****p < 0.001, ns, non-significant.

FIGURE 6

Phage Remus bacteriolytic activities against Staphylococcus aureus BP043 preformed aggregates in human synovial fluid. Viable bacterial counts (log₁₀ CFU/ml) after applying various titers of Remus (∼3 × 10⁹–10⁶ PFU/ml) for 48 h against pre-formed aggregates of ∼4 × 10⁸ CFU/ml in synovial fluid (MOIs: 10–10^{– 2}). Control is non-phage treated aggregates. N = 2–6 synovial fluids, mean ± SE. Asterisks on top of the columns show comparison between the control (no treatment) and the different PFUs treatment groups. One-way ANOVA (Tukey’s multiple comparisons test) was utilized to determine statistical difference, *p < 0.05, ****p < 0.0001, ns, non-significant.

FIGURE 7

Phage Remus and vancomycin efficacy against pre-formed Staphylococcus aureus BP043 aggregates in human synovial fluid. Bacterial load (log₁₀ CFU/ml) in synovial fluid after applying phage Remus alone at ∼3 × 10⁸ PFU/ml (48 h), vancomycin at 500 μg/ml (48 h) and sequential treatment of phage (24 h) followed by vancomycin (24 h). N = 5 synovial fluids, mean ± SE. Asterisks on top of the columns show comparison between the control (no treatment) and the different treatment groups. Statistical significance was assessed by performing one-way ANOVA (Tukey’s multiple comparisons test), *p < 0.05, **p < 0.01, ****p < 0.0001.

FIGURE 8

Kaplan–Meier survival analysis of G. mellonella infected with planktonic cells or synovial fluid aggregates of Staphylococcus aureus BP043. Survival was evaluated after administration of 10⁸ PFU of phage Remus and/or 5 μg of vancomycin to larvae infected with S. aureus BP043 as, (A) planktonic cells grown in Ringer’s solution TSB (2.2 × 10⁷ CFU) or (B) aggregates (1.7 × 10⁷ CFU). Mortality was scored daily for 96 h according to Supplementary Table 1. Uninfected larvae receiving a Remus-vancomycin combination treatment were included as a control group up to 48 h. The data represented are the mean of three biological replicates with ten larvae per group. Data of synovial fluid controls from the same experiments are presented in both figures for clarity and comparison. Statistically significant differences by the log-rank test were defined as p < 0.0001 between treatment groups.