Efficacy of Bisphosphonate-Conjugated Sitafloxacin in a Murine Model of S. aureus Osteomyelitis: Evidence of "Target & Release" Kinetics and Killing of Bacteria Within Canaliculi

Abstract

S. aureus infection of bone is difficult to eradicate due to its ability to colonize the osteocyte-lacuno-canalicular network (OLCN), rendering it resistant to standard-of-care (SOC) antibiotics. To overcome this, we proposed two bone-targeted bisphosphonate-conjugated antibiotics (BCA): bisphosphonate-conjugated sitafloxacin (BCS) and hydroxybisphosphonate-conjugate sitafloxacin (HBCS). Initial studies demonstrated that the BCA kills S. aureus in vitro. Here we demonstrate the in vivo efficacy of BCS and HBCS versus bisphosphonate, sitafloxacin, and vancomycin in mice with implant-associated osteomyelitis. Longitudinal bioluminescent imaging (BLI) confirmed the hypothesized "target and release"-type kinetics of BCS and HBCS. Micro-CT of the infected tibiae demonstrated that HBCS significantly inhibited peri-implant osteolysis versus placebo and free sitafloxacin (p < 0.05), which was not seen with the corresponding non-antibiotic-conjugated bisphosphonate control. TRAP-stained histology confirmed that HBCS significantly reduced peri-implant osteoclast numbers versus placebo and free sitafloxacin controls (p < 0.05). To confirm S. aureus killing, we compared the morphology of S. aureus autolysis within in vitro biofilm and infected tibiae via transmission electron microscopy (TEM). Live bacteria in vitro and in vivo presented as dense cocci ~1 μm in diameter. In vitro evidence of autolysis presented remnant cell walls of dead bacteria or "ghosts" and degenerating (non-dense) bacteria. These features of autolyzed bacteria were also present among the colonizing S. aureus within OLCN of infected tibiae from placebo-, vancomycin-, and sitafloxacin-treated mice, similar to placebo. However, most of the bacteria within OLCN of infected tibiae from BCA-treated mice were less dense and contained small vacuoles and holes >100 nm. Histomorphometry of the bacteria within the OLCN demonstrated that BCA significantly increased their diameter versus placebo and free antibiotic controls (p < 0.05). As these abnormal features are consistent with antibiotic-induced vacuolization, bacterial swelling, and necrotic phenotype, we interpret these findings to be the initial evidence of BCA-induced killing of S. aureus within the OLCN of infected bone. Collectively, these results support the bone targeting strategy of BCA to overcome the biodistribution limits of SOC antibiotics and warrant future studies to confirm the novel TEM phenotypes of bacteria within OLCN of S. aureus-infected bone of animals treated with BCS and HBCS.

Keywords: Staphylococcus aureus; antibiotic; bisphosphonate; osteomyelitis; transmission electron microscopy.

Figure 1

In vivo evidence of extended antimicrobial efficacy of bisphosphonate-conjugated sitafloxacin. Mice (n = 4) were challenged with a USA300LAC::Lux-contaminated transtibial pin and were given the indicated drug on days 0, 3, 6, 9, and 12 postinfection. In vivo BLI was performed on the indicated days 3, 5, and 7 (A), and the data for each group are shown continuously as mean +/- SD for the group (**p < 0.001 two-way ANOVA) (B). Of note is that the BLI in HPHBP-treated mice (negative control for HBCS) remained elevated throughout infection, while the free sitafloxacin treatment led to an immediate reduction in BLI and maximal inhibition of BLI in both BCS- and HBCS-treated mice was delayed until ~5–7 days postinfection and remained low thereafter.

Figure 2

In vivo evidence of reduced peri-implant osteolysis in HBCS- and BCS-treated mice vs. free sitafloxacin and controls. Mice (n = 4) were challenged with a USA300LAC::Lux contaminated transtibial pin and were given the indicated drug on days 0, 2, 4, 6, 8, 10, 12, and 14 postinfection. The infected tibiae were harvested on day 14 and processed for micro-CT to quantify peri-implant osteolysis. Representative 3D lateral view of renderings of infected tibiae from mice treated with placebo (A), HPHBP (B), sitafloxacin (C), BCS (D), and HBCS (E) are shown with the osteolysis area (green). The osteolysis area for each tibia is presented with the mean +/- SD for each group (F) **p < 0.01 one-way ANOVA).

Figure 3

HBCS treatment reduces osteoclast numbers in mice with implant-associated osteomyelitis. The infected tibiae (n = 4) described in Figure 2 were processed for TRAP stained histology, and semiautomated histomorphometry was performed to quantify the TRAP+ area. Representative high-resolution (×40) images of placebo (A), sitafloxacin (B), BCS (C), and HBCS (D) are shown with quantification of the TRAP-stained area (E), in which the data for each tibia are presented with the mean +/-SD for the group (**p < 0.05 one-way ANOVA). All black arrows in (A–D) indicate the osteoclast and # mean the presence of sequestrum in the medullary canal. The TRAP+ area for HPBP-treated mice was similar to placebo (data not shown).

Figure 4

Morphology of S. aureus within untreated and sitafloxacin-treated in vitro SACs analyzed by TEM. In vitro Staphylococcal abscess communities (SAC) grown in collagen hydrogels were processed for TEM imaging to illustrate the morphologic features of 100% viable untreated bacteria with dense black interiors (A–C) ×30,000, compared to sitafloxacin-treated SACs (D–F) ×30,000, displaying vacuolated degenerating (gold arrows) and dead bacteria (ghosts, red arrows). Note in (D), a single viable S. aureus (blue arrow) non-vacuolated coccus with an intact septal wall compared to the majority of vacuolated dying bacteria.

Figure 5

TEM evidence of HBCS and BCS killing of MRSA within canaliculi. Histology from the infected tibiae described in Figure 3 were processed for Brown and Brenn staining (A–J, U, W) and subsequent TEM (K–T, V, X). Representative low-power (×0.5) images containing necrotic bone fragments with Gram-positive bacteria in the marrow space (A, C, E, G, I), with high-power (×5) images of the dashed boxed region of interest (B, D, F, H, J) are shown to illustrate the extent of biofilm formation (black arrows) in the indicated treatment groups on day 14 postinfection. Representative TEM images of the bacteria within canaliculi of the infected bone are shown at low power (×12,000, K, M, O, Q, S), with high-power images (×25,000) of the dashed boxed region of interest (L, N, P, R, T) shown to illustrate the morphology of the bacteria within canaliculi on day 14 postinfection. Note that most bacteria in the placebo and vancomycin- and sitafloxacin-treated tibiae are dense (dark), and there are a few ghosts (red arrows) and degenerating bacteria (yellow arrows). In contrast, bacteria in BCS- and HBCS-treated tibiae are less dense and contain small vacuoles and holes >100 nm. To see if the morphology of these vacuole-containing dying bacteria with hole >100 nm changes to a more necrotic phenotype, we repeated the HBCS treatment of mice with an MRSA-infected transtibial pin for 14 days, left these mice untreated for another 14 days (U, V), or throughout all 28 days of treatment (W, X), then harvested the infected tibiae 28 days post-op for histology and TEM. A representative Brown and Brenn-stained section containing a necrotic bone fragment with Gram-positive bacteria is shown at ×10 (U, W) with a high-power (×15,000) TEM of the bacteria in the canaliculus (V, X). Note that the large number of ghosts and necrotic bacteria had lost their semirigid structure, several ruptures and displacement of membranes, lysis, and extrusion of the intracellular content with large holes (red arrows) in HBCS treatment groups.

Figure 6

HBCS and BCS treatment of MRSA-infected mice causes swelling of bacteria within canaliculi. The TEM images described in Figure 5 were used to quantify the diameter of the bacteria in the canaliculi of mice given the indicated treatment for 14 days following infection. The diameter of each bacterium (n = 4) is presented with the mean +/-SD for the group (two-way ANOVA: ns: no statistical significant, ****p < 0.0001, **p < 0.005, *p < 0.05).