Quantifying the natural history of biofilm formation in vivo during the establishment of chronic implant-associated Staphylococcus aureus osteomyelitis in mice to identify critical pathogen and host factors

Abstract

While it is well known that Staphylococcus aureus establishes chronic implant-associated osteomyelitis by generating and persisting in biofilm, research to elucidate pathogen, and host specific factors controlling this process has been limited due to the absence of a quantitative in vivo model. To address this, we developed a murine tibia implant model with ex vivo region of interest (ROI) imaging analysis by scanning electron microscopy (SEM). Implants were coated with Staphylococcus aureus strains (SH1000, UAMS-1, USA300LAC) with distinct in vitro biofilm phenotypes, were used to infect C57BL/6 or Balb/c mice. In contrast to their in vitro biofilm phenotype, results from all bacteria strains in vivo were similar, and demonstrated that biofilm on the implant is established within the first day, followed by a robust proliferation phase peaking on Day 3 in Balb/c mice, and persisting until Day 7 in C57BL/6 mice, as detected by SEM and bioluminescent imaging. Biofilm formation peaked at Day 14, covering ∼40% of the ROI coincident with massive agr-dependent bacterial emigration, as evidenced by large numbers of empty lacunae with few residual bacteria, which were largely culture negative (80%) and PCR positive (87.5%), supporting the clinical relevance of this implant model.

Keywords: Staphylococcus aureus; biofilm; implant; osteomyelitis; scanning electron microscopy.

Figure 1. Defining the 1 mm² region of interest (ROI) on the implant to quantify biofilm surface area

(A) The 2.0 × 0.5 mm surface area immediately adjacent to the bend is defined as the ROI from the primary SEM image of flat wire, and digitally cropped. Pseudo-color conversion is performed via thresholding pixels that are brighter (green) and darker (blue) than the stainless steel surface using ImageJ software. Finally, the green and blue pixels were converted to white, and all the other pixels were converted to black, to allow for binary quantitation of the % Biofilm Area (white bars = 0.2 mm). Existence of bacteria is not necessary to yield positive value. (B) To assess the spatial development of biofilm in our model, the ROI was divided into the Medial cortex region (0.5 mm), Medullary canal region (1.0 mm) and Lateral cortex region (0.5 mm) as illustrated.

Figure 2. S. aureus strains exhibit different biofilm phenotypes in a flow biofilm model

Stainless steel pins were placed in a flow chamber containing TSBGN (TSB supplemented with 1% glucose and 3% NaCl) or TSB+HP (TSB supplemented with 10% human plasma) media without bacteria, or inoculated with SH1000, UAMS-1 or USA300LAC S. aureus, and incubated for 24 hours with a flow rate of 0.2 mL/min at 37 °C. The pins were harvested and processed for SEM. Representative SEM images of two independent experiments are shown in which the ROIs (red box) of the biofilm in the left panels obtained at 150×, are presented at 5,000× in the right panels. SH1000 produced similar biofilms in TSBGN that was phenotypically similar to that produced in TSB+HP where the fibrin fibers are not abundant. In contrast UAMS-1 failed to produce biofilm in TSBGN, as evidenced by sparse cell clusters (black arrows), but generated robust biofilm containing abundant fibers (white arrows) in TSB+HP. USA300LAC produced robust biofilms in both media, but fibers (white arrows) were only present when grown in TSB+HP.

Figure 3. SEM assessment of the natural history of S. aureus adhesion, proliferation, biofilm formation and emigration during chronic implant-associated osteomyelitis in vivo

C57BL/6 mice (n = 5) were infected with S. aureus (SH1000, UAMS-1, or USA300LAC) contaminated stainless steel implants, and SEM images were obtained from the harvested implants. Representative low (5,000×: bar = 10 μm) and high (boxed region in low image; 15,000×: bar = 1 μm) magnification SEM images of S. aureus and biofilm on the implants at the indicated times are shown. Note the lack of biofilm associated with the adherent bacteria grown in vitro (top row), and the rapid accumulation of biofilm in the first 3 days following implantation. Also of note is the bacterial proliferation evidenced by the large clusters observed on Day 3 and Day 7, and S. aureus emigration suggested by the abundant empty lacunae observed after on Days 14 and 28 (also see enlarged images in Supplemental Figure 2). No remarkable differences between strains were observed.

Figure 4. Agr is required for S. aureus dispersal from the implant-associated biofilm

Mice (n=5) were challenged with a UAMS-1Δagr contaminated implant, which was harvested for SEM analysis on Day 14 as described in Fig. 3. Representative SEM images obtained at 5,000× (A) and 15,000× (B) illustrate the biofilm filled with bacteria and the absence of empty lacunae, demonstrating that agr is required for S. aureus emigration from biofilm (white bar = 1 μm). Wild-type biofilms remain intact with empty lacunae and a few residual cells. SEM images of the implants harvested at 6 months were obtained at 5,000× (C, D) and 15,000× (E). Note that empty lacunae were still the dominant biofilm feature at this late time point (C), although residual cocci could be found in all samples (D, E). (white bar = 1 μm).

Figure 5. Mature S. aureus biofilm is detected within 1 week and its growth peaks at 2-weeks post-implantation

Mice (n = 5) were challenged with a sterile implant (No bacteria control), or an implant contaminated with SH1000, UAMS-1 or USA300LAC, and the implant was harvested at the indicated time for SEM analysis as described in Fig. 2. (A) Representative SEM images of the biofilm within the ROI on the implant obtained at 25× (top) with the rendered biofilm area illustrate the growth of the biofilm. (white bars = 0.2 mm) (B) The % Biofilm Area on the implant for each strain at the indicated time is presented as the mean ± SD (*p < 0.05 vs. No bacteria; ^ap < 0.05 vs. Day 1 of the same strain; ^bp < 0.05 vs. Day 1 and Day 3 of the same strain; ^cp < 0.05 vs. Days 1, 3, and 7 of the same strain).

Figure 6. Biofilm growth on the implants commences on the medial side and progresses laterally

Mice were infected with the indicated strain of S. aureus as described in Fig. 3, and the implants were harvested for quantitative SEM on the indicated day. (A) Representative SEM images of SH1000 biofilm obtained at 25× illustrate the temporal and spatial biofilm growth over time. For each time point, the primary image of the ROI (top) is presented with the rendered binary image that was generated to quantify % Biofilm Area on the Medial cortex region (M), Medullary canal region (C) and Lateral cortex region (L) of the implant. Note that biofilm is first seen in the Medial Cortex region, and progressed laterally to cover surface at Day 14. The % Biofilm Area at the indicated time point was calculated (B) SH1000, (C) UAMS-1 and (D) USA300LAC, and the data are presented as the mean ± SD (*p < 0.05 vs. Medial Cortex region; ^ap < 0.05 vs. Day 1 in the same region; ^bp < 0.05 vs. Day 1 and Day 3 in the same region; ^cp < 0.05 vs. Days 1, 3 and 7 in the same region).

Figure 7. Assessment host strain variability, dynamics of bacterial metabolic activity, biofilm formation and maturation during establishment of chronic implant-associated osteomyelitis

Mice were infected with a bioluminescent strain of UAMS-1 (Xen40), and the BLI signal for each Balb/c (A) and C57BL/6 (B) mouse is presented along with the mean BLI at each time point (black line). Note that the peak BLI signal in Balb/c mice is brief (Day 3 only), while the peak in C57BL/6 mice extends from Day 3 through Day 7. The BLI signal intensity in Balb/c is 10-fold greater vs. C57BL/6, which is likely due to skin pigment absorption. (C) Representative 5,000× and 15,000× (boxed region) SEM images of the biofilm on the implant harvested from each mouse strain at 56 days illustrate the similarity (white bar = 10 μm and 1 μm, respectively). Yellow arrows indicates S. aureus cells. No remarkable differences in bacteria or biofilm were observed at any time point, which was confirmed by quantitative assessment of the % Biofilm Area (D).

Figure 8. Assessment of the bacterial load on the implant and the adjacent tissues during the establishment of chronic implant-associated osteomyelitis

(A) C57BL/6 mice (n > 5) were challenged with UAMS-1 infected implants (described in Fig. 3) that was harvested at the indicated time points for CFU assays. The day post-infection for each mouse is plotted against the mean CFU count measured at each time point. (B) By using the part of the (A), bacterial burden of the adjacent tissues of the infection site was determined at day14 and 28. **:p<0.01 in Mann–Whitney U test.(C) RT-PCR was performed with primers specific for S. aureus 16S rRNA with total RNA isolated from 8 implants harvested at Day 28. Note that while only 2 implants out of 10 had CFU, most of implants (7/8) were RT-PCR positive for S. aureus.