Diflunisal‐loaded poly(propylene sulfide) nanoparticles decrease S. aureus‐mediated bone destruction during osteomyelitis

Abstract Osteomyelitis is a debilitating infection of bone that results in substantial morbidity. Staphylococcus aureus is the most commonly isolated pathogen causing bone infections and features an arsenal of virulence factors that contribute to bone destruction and counteract immune responses. We previously demonstrated that diflunisal, a nonsteroidal anti‐inflammatory drug, decreases S. aureus‐induced bone destruction during osteomyelitis when delivered locally from a resorbable drug delivery depot. However, local diflunisal therapy was complicated by bacterial colonization of the depot's surface, highlighting a common pitfall of devices for local drug delivery to infected tissue. It is, therefore, critical to develop an alternative drug delivery method for diflunisal to successfully repurpose this drug as an antivirulence therapy for osteomyelitis. We hypothesized that a nanoparticle‐based parenteral delivery strategy would provide a method for delivering diflunisal to infected tissue while circumventing the complications associated with local delivery. In this study, we demonstrate that poly(propylene sulfide) (PPS) nanoparticles accumulate at the infectious focus in a murine model of staphylococcal osteomyelitis and are capable of efficaciously delivering diflunisal to infected bone. Moreover, diflunisal‐loaded PPS nanoparticles effectively decrease S. aureus‐mediated bone destruction, establishing the feasibility of systemic delivery of an antivirulence compound to mitigate bone pathology during osteomyelitis.


| INTRODUCTION
Osteomyelitis, or inflammation of bone, is commonly caused by bacterial infection. This disease afflicts an estimated 1 in 4000 people annually and is projected to impact up to 30% of orthopedic procedures. 1,2 Due in part to the widespread emergence of antimicrobial resistance, treatment of osteomyelitis can be extremely difficult. 3,4 Efforts to cure osteomyelitis often involve invasive debridement procedures and long-term antibiotic therapy that together result in substantial strain on the patient and healthcare system. 2,[5][6][7] Staphylococcus aureus, a Gram-positive bacterium, is the most common etiologic agent of osteomyelitis. 1 S. aureus possesses an arsenal of virulence factors that lyse host cells, including skeletal cells, thereby contributing to osteomyelitis-induced bone loss. 8 Thus, effective therapies are necessary to ameliorate concomitant morbidities such as bone loss that may increase the risk of fracture or treatment failure.
Antivirulence therapies inhibit bacterial virulence pathways without directly impacting bacterial viability and are actively being investigated as adjunctive treatment strategies. 3 We have recently demonstrated the antivirulence potential of diflunisal, a nonsteroidal anti-inflammatory drug, to decrease S. aureus-induced bone destruction in a murine osteomyelitis model. 9 Diflunisal inhibits the quorum-sensing agr pathway of S. aureus, limiting production of numerous virulence factors including cytolytic toxins. 10 In previous studies, local delivery of diflunisal from resorbable poly(ester urethane) foams significantly reduced bone resorption. 9,11 While local delivery presents the advantage of achieving high drug concentrations near target sites, the avascular delivery depot can function as a nidus for bacterial colonization. [11][12][13] Thus, effective delivery of diflunisal and other antivirulence compounds requires an alternative method to avoid exacerbation of infection.
While parenteral therapy potentially circumvents the challenges of local delivery devices, diflunisal is hydrophobic and therefore has low aqueous solubility. Encapsulation of compounds within nanoparticles has enabled effective systemic delivery of hydrophobic drugs and demonstrated distribution to target sites. [14][15][16][17][18] Our group has previously shown that poly(propylene sulfide) (PPS) nanoparticles provide a reactive oxygen species (ROS)-responsive carrier for delivery of the Gli2 inhibitor, GANT58, to sites of bone cancer metastases. 18 The PPS nanoparticles distributed preferentially to tumor-bearing limbs compared to contralateral limbs, presumably due to increased vascular permeability at tumor sites that allows for nanoparticle extravasation and decreased lymphatic drainage. These phenomena allow for nanoparticle retention and are known as the enhanced permeability and retention (EPR) effect. 19 Furthermore, PPS-based biomaterials break down in the presence of high levels of ROS, [20][21][22] providing a potential mechanism for targeted drug release at inflamed sites. However, few studies have investigated systemically (e.g., intravenously) delivered nanoparticles in the context of osteomyelitis. [23][24][25][26] The objectives of this study were to understand the biodistribution of PPS nanoparticles during osteomyelitis and evaluate the efficacy of diflunisal-loaded nanoparticles in limiting S. aureusinduced bone loss. We hypothesized that PPS nanoparticles would accumulate at infectious foci during osteomyelitis and that diflunisalloaded PPS nanoparticles would limit S. aureus-mediated cortical bone destruction. To test these hypotheses, we evaluated PPS nanoparticle delivery in a murine model of osteomyelitis and investigated the efficacy of diflunisal-loaded PPS nanoparticles both in vitro and in vivo.

| Cell lines, bacterial strains, and reagents
The murine preosteoblast MC3T3-E1 subclone 4 cell line was obtained from the American Type Culture Collection. The cells were propagated in a humidified 37°C incubator with 5% CO 2 and maintained in ⍺-MEM (Gibco #A1049001; Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (Bio-Techne) and 1X penicillin-streptomycin (Ther-   After determining optimal parameters for drug loading, nanoparticles were fabricated in large batches by microfluidics processes as described previously for animal studies. 33 Briefly, PPS 135 -b-p (Cy7 1 -ran-DMA 149 ) (60.0 mg) was codissolved with pHPMA-Bz (6.0 mg) and/or diflunisal (6.0 mg) in methanol (0.6 ml) and mixed with sterile PBS using a benchtop NanoAssemblr (Precision Nanosystems, Inc.). All formulations were prepared with a 10:1::aqueous:organic flow rate ratio and 4 ml/min total flow rate.

| Fabrication and characterization of nanoparticles
Methanol was removed using a rotovap heated to 40°C for 30 min.
Resulting solutions were passed through 0.45-μm syringe filters. All nanoparticles contained Cy7-grafted polymer for imaging purposes.
Dif-NPs refers to nanoparticles loaded with diflunisal and pHPMA-

| Preparation of concentrated supernatants
One colony of S. aureus from a tryptic soy agar plate was used to was recorded each hour to monitor bacterial growth using a BioTek Synergy HT microplate reader (BioTek Instruments, Inc.). The initial OD 600 reading was subtracted from each well to serve as a baseline.

| MC3T3 cytotoxicity assay
MC3T3 cytotoxicity was analyzed as reported previously. 9  treated and euthanized at Day 7, and bacterial burdens were assessed as conducted previously. 34 In total, 86 animals were used to complete these studies. All statistical analyses were performed with GraphPad Prism.

| PPS diblock copolymer nanoparticle synthesis and cargo release
To generate the building blocks necessary for fluorescent nanoparticle synthesis, PPS-b-p(Cy7 1 -ran-DMA 149 ) polymer ( Figure 1A) was synthesized by RAFT polymerization. Polymer structure was confirmed by 1 H NMR ( Figure S1). An oil-in-water emulsion formed the micellar nanoparticles in which the hydrophilic DMA blocks compose the hydrophilic corona and the hydrophobic PPS blocks compose the ROS-responsive core which releases loaded drug upon destabilization ( Figure 1B).

| Formation of diflunisal-loaded PPS nanoparticles for drug delivery
To determine the optimal process for encapsulation of diflunisal within PPS nanoparticles (Dif-NPs), the quantity of loaded drug and encapsulation efficiency of two different drug-to-polymer ratios were characterized. The addition of pHPMA-Bz as an excipient was also tested to determine the benefits of facilitated π-π stacking on diflunisal encapsulation. Increasing the drug-to-polymer ratio from FORD ET AL.   Figure 2D).

| Systemically administered nanoparticles accumulate at infected femurs
Having identified an optimal nanoparticle formulation, we sought to  Figures 5G and S4), suggesting that Dif-NPs had no effect on bacterial burdens. Thus, diflunisal-loaded PPS nanoparticles decrease S. aureus-induced bone loss in infected femurs during osteomyelitis without significantly influencing bacterial burdens. Taken together, these data support findings that PPS nanoparticles efficaciously deliver diflunisal to infectious foci to decrease bone destruction during osteomyelitis.

| DISCUSSION
Delivery of hydrophobic drugs such as diflunisal is limited by low aqueous solubility, which can lead to unfavorable pharmacokinetic profiles and poor biodistribution when delivered parenterally. 36,37 Local delivery systems have been designed to overcome solubility limitations; however, foreign devices are known to be a nidus for bacterial colonization and biofilm formation. [11][12][13] While some compounds (including diflunisal) can achieve systemic delivery through oral delivery, oral administration is not feasible in all clinical settings (e.g., moribund or perioperative patients), and alternative parenteral options may be advantageous. For such compounds without parenteral compatibility, nanoparticle delivery systems offer a parenteral delivery vehicle for pharmaceuticals to target sites. Although nanoparticle accumulation has not been extensively studied in the context of osteomyelitis, effective treatment of bone infection with systemically administered nanoparticles has been reported. 23,26 Delivery of antimicrobial compounds using locally administered nanoparticles has also been investigated both in vitro 14,38 and in vivo, [39][40][41] but systemic delivery of nanoparticles capable of carrying hydrophobic drugs is under-investigated in osteomyelitis. Delivery of diflunisal using nanoparticles may provide effective therapy and limit potential complications associated with avascular local delivery devices.
In this study, we evaluated the efficacy of PPS nanoparticles to deliver diflunisal, which we previously demonstrated inhibits S. aureus-induced cortical bone destruction when delivered locally  Compared to free-drug administration via intravenous or oral delivery routes, synthetic nanoparticles offer the potential to accumulate and release loaded compounds at target sites. 14-18 As described by the EPR effect, both tumors and inflammation result in enhanced vascular permeability allowing for extravasation of nanoparticles. 42 Our results suggest that PPS nanoparticles accumulate at the infectious foci; however, the exact mechanisms that drive nanoparticle retention during posttraumatic osteomyelitis must be investigated further. One possible mechanism may include phagocytic cell uptake as described in the "ELVIS" effect (extravasation via leaky vasculature followed by inflammatory cell sequestration). 43,44 Nevertheless, modifications to the nanoparticle chemistry have shown enhanced retention at target sites and allow for further improvement of nanoparticle accumulation in bone in other disease models. 29 Considering that sites of inflammation and infection are known to produce ROS 45 and that release of compounds from PPS nanoparticles is responsive to ROS concentration, it is likely that ROS levels at infected sites contribute to drug cargo release within bone.
However, more extensive in vivo analyses must be performed to conclude that ROS-mediated degradation is the primary mechanism of drug release at the infectious site.  One mouse in the Blank-NPs group experienced more than 20% weight loss and was euthanized. Different symbols (circles, triangles, and squares) represent three independent trials that included the groups as indicated by the corresponding symbols. Effect size (Hedges' g) between Blank-NP and Dif-NP groups = −1.500 (95% confidence interval: −0.684, −2.317). The median femur from each group is shown in a three-dimensional reconstruction to the right of the graph. Error bars represent mean ± SEM. **p < .01 and ****p < .0001 as determined by one-way ANOVA. (F) Quantification of bacterial burden by colony-forming units enumeration 7 days postinfection following daily treatment with PBS, Blank-NPs, or Dif-NPs. N = 5 mice per group. One mouse in the PBS group was euthanized following an adverse response to anesthesia. Error bars represent mean ± SEM. ns denotes no significance as determined by one-way ANOVA. (G) Representative histology images of femurs harvested from mice treated with Blank-NPs or Dif-NPs and stained with a modified hematoxylin and eosin stain. Scale bars are as shown in the lower right corner of images FORD ET AL.