MRSA strain, in vitro culture, and colony forming units (CFU)

The most prevalent community-acquired MRSA strain is USA300. Thus, we used USA300 LAC::Lux, which has a chromosomal insertion of the bacterial luciferase lux gene that allows for in vivo bioluminescent imaging (BLI).35 This bioluminescent construct constitutively emits a blue-green light with a maximal emission wavelength of 490 nm only in live and metabolically active bacteria.36 The USA300 LAC::Lux bacteria were cultured in tryptic soy broth media at 37 °C overnight as previously described.9,37 CFU assays of overnight cultures and sonicated implants were performed as previously described.14,38

Animal surgeries and antibiotics treatments

When we first developed our murine models of implant-associated osteomyelitis we evaluated male and female mice and failed to identify any sexual dimorphisms.14,32,39 Thus, since female mice are more cost-effective for research (e.g., 5 mice per cage), easier to handle, and exhibit less aggressive behavior towards post-surgery implant destruction, female mice are the perfect option for our current study. In terms of mouse stains, BALB/c mice are frequently used for a variety of infection studies, mainly because they display TH2-biased immune responses.40 All animal experiments were conducted under IACUC-approved protocols. Female BALB/c mice that were 10 weeks old were purchased from Jackson Research Labs (Bar Harbor, ME) and acclimated to a standard microisolator environment for one week prior to surgeries, as previously described.14,39 Briefly, the mice were anesthetized with xylazine (20 mg·kg−1) and ketamine (100 mg·kg−1) administered intraperitoneally, and the right femur or tibia was exposed in accordance with animal surgery guidelines to follow either the septic femoral plate 1-stage revision model or septic transtibial pin model, respectively, with treatment of bisphosphonate-conjugated sitafloxacin (BCS) and hydroxybisphosphonate-conjugate sitafloxacin (HBCS) as previous descripted.9,10

Experiment 1: The septic femoral plate 1-stage revision model

A total of 28 mice underwent septic femoral plate surgery as described in Fig. 1. The right femur was implanted with a 6-hole femoral titanium plate and screws (RISystem, Davos, Switzerland) that was secured by a sterile screw in hole #3 and a screw containing ~105 CFU of USA300LAC::Lux in hole #4. The presence of osteomyelitis was confirmed via X-ray, BLI, and CFU assays on day 7 as previously described.14 Also on day 7, all implants were removed with debridement and irrigation, and a sterile plate was implanted and secured by sterile screws in holes #1 and #6 with the initiation of antibiotic therapy (Fig. 1a, b). To verify the efficacy of SOC with bisphosphonate-conjugated sitafloxacin (BCS) and hydroxybisphosphonate-conjugated sitafloxacin (HBCS), the mice were randomized into seven treatment groups: 1) Baseline (day 7 post-op to serve as a cross-section control for peri-implant osteolysis) (n = 4); 2) Hydroxyphenylethanebisphosphonate (HPBP, BV6001) + Vancomycin (n = 4); 3) Hydroxyphenylethanehydroxybisphosphonate (HPHBP, BV6301) + Vancomycin (n = 4); 4) Vancomycin (n = 4); 5) Sitafloxacin(n = 4); 6) BCS (BV600072) + Vancomycin (n = 4); 7) HBCS (BV63072) + Vancomycin (n = 4). The parent sitafloxacin was used to assess the effects of non-bone targeted antibiotics, while HPHBP and HPBP were used as control for non-nitrogen-containing bisphosphonate effects on bone, and the mice were euthanized on day 21(Fig. 1c). The chemical structures and commercial product numbers of these compounds are shown in Fig. 1d.

Experiment 2: Septic transtibial pin model

To further assess the antimicrobial efficacy of HBCS and its mechanism of action, a transtibial pin model was employed as previously described.39 Briefly, the right tibia was shaved, and the skin was cleansed with 70% ethanol. Infection was initiated by placing a stainless steel L-shape pin contaminated with ~105 CFU of USA300LAC::Lux transcortical through the tibia via medical to lateral implantation which remained untreated for 1 week to establish chronic osteomyelitis (Fig. 7a). On day 7, the mice were randomized into three monotherapy treatment groups (n = 10 per group): 1) Vancomycin (110 mg·kg−1 per 24 h); 2) Sitafloxacin (2.5 mg·kg−1 per 24 h); and 3) HBCS (BV63072) (3 mg·kg−1 per 24 h). The infected mice were treated for another 14 days, euthanized via heart perfusion with 4% paraformaldehyde/2.5% glutaraldehyde on day 21, and the infected tibiae were processed for histology and TEM (Fig. 7b).

Bioluminescence imaging

For mice in Experiment 1, in vivo BLI was performed on days 0, 1, 3, 5, 7, 7 (postoperative), 8, 10, 12 and 14 as indicated in Fig. 1. Prior to performing the BLI, the mice were anesthetized with ketamine/xylazine (100 mg/20 mg/kg) and the femur with implant was included in the region of interest (ROI) for imaging and analysis. The BLI data are presented on a color-scale overlapped with a grayscale photograph of the femur and quantified as total flux (photons/s) within a standardized circular ROI using Living Image software (Caliper).41 Of note is that absence of BLI signal on day 1 is an exclusion criteria for this outcome measure, as the data at later timepoints are normalized to this value.14 Thus, as seven mice had no BLI signal on day 1 post-op, BLI analyses for all treatment groups were reduced to n = 3. Following randomization into treatment groups, one mouse in Group 2 and one mouse in Group 3 died due to accidental overdose of anesthesia. In order to meet the requirements for statistical analysis, we did another set of BLI experiments to make the sample size of each group reach n = 5 (Fig. 3c).

Radiographic and Micro-CT

Infected femurs were harvested on day 21 in experiment 1, peri-implant osteolysis and osseointegration was assessed radiographically using a Faxitron Cabinet X-ray system (Faxitron, Wheeling, IL, USA) as previously described.42 Micro-CT was performed to quantify screw hole volume as previously described.41,43

Histology

Infected long bones were dissected to remove the implants, decalcified, and embedded in paraffin. At the mid-point of the femur or tibia on the sagittal cuts, five 5 µm tissue sections were obtained from three levels and mounted onto slides. One slide per orientation of cut was stained with H&E, Brown-Brenn, or for tartrate-resistant acid phosphatase (TRAP) as previously described.10 Semi-automated histomorphometry to quantify biofilm area and TRAP+ area was performed with Visiopharm software as previously described.10

Transmission electron microscopy (TEM)

For mice in Experiment 2, regions of interest (ROI) within serial sectioned paraffin blocks of infected tibia samples were identified from the Brown-Brenn stained sections, and adjacent unstained slides were reprocessed for TEM using the “pop-off” technique, as previously described.44 Slides were deparaffinized in 3 changes xylene and then rehydrated through a graded series of ethanol back to ddH2O. Rehydrated sections on slides were post fixed overnight at 4 °C in 0.1 mol·L−1 phosphate buffered 2.5% glutaraldehyde, then 1% OsO4 for 20 min at room temperature. The slides were washed in ddH2O, dehydrated in a graded series ethanol to 100% (x3), infiltrated with a 1:1 mixture of 100% ethanol and Spurr’s resin, then 100% Spurr’s resin overnight at room temperature. The ROI (etched with a diamond pen on the back side of the slides) was re-identified after the slides were drained of excess resin. A size 3 BEEM capsule filled with fresh resin was placed over the paraffin section’s ROI and polymerized 24 h at 65˚C. BEEM capsules, with entrapped ROIs, were popped off slides by dipping them 5–10 s rapidly in liquid nitrogen. The popped off block was trimmed to the ROI, thin sectioned at ~70 nm, and placed onto formvar carbon coated nickel slot grids for imaging using a Gatan Erlangshen 11-megapixel digital camera (Pleasanton, CA) and a Hitachi 7650 TEM (Pleasanton, CA).

Statistics

CFU data were quantified for each implant to determine the mean ± SD for the group, and statistical significance was assessed by 1-way ANOVA. Percent of body weight from pre-op day 0 was quantified for each mouse to determine the mean ± SD for the group, and statistical significance was assessed by 2-way ANOVA. Longitudinal BLI data are presented as the % BLI normalized to day 1 for each group and for each mouse with mean ± SD for the group, and statistical significance was assessed by Wilcoxon matched-pairs signed rank test. Boolean assessment of catastrophic fractures was determined from X-ray prior to removal of the implants by Fisher’s exact test. Micro-CT assessment of screw hole volume was determined for each mouse with mean ± SD for the group, and statistical significance was assessed by 1-way ANOVA. Histomorphometric quantification of biofilm and TRAP+ area was determined with ImageJ for each mouse with mean ± SD for the group, and statistical significance was assessed by 1-way ANOVA. Boolean assessments of ongoing infection (presence of necrotic marrow, viable SACs, Gram positive bone fragments, and endosteal ring of osteoclasts) and infection control (presence of bridging new bone and osseous integration of the implant) were quantified as the percentage of animals per group and significant differences from vancomycin control were assessed by Fisher’s exact test.