Effects of Intra-Articular D-Amino Acids Combined With Systemic Vancomycin on an Experimental Staphylococcus Aureus-Induced Periprosthetic Joint Infection

Yicheng Li Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Shalitanati Wuermanbieke Karamay Central Hospital of Xinjiang Jiangdong Ren Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Wenbo Mu Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Hairong Ma State Key Laboratory of Pathogenesis, Prevention and Treatment of High Incidence Diseases in Central Asian Xinjiang Key Laboratory of Echinococcosis, Clinical Medical Research Institute Fei Qi Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Xiaoyue Sun Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Yang Liu Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Abdusami Amat Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Xiaogang Zhang Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi Li Cao (  xjbone@sina.com ) Department of Orthopaedics, First A liated Hospital of Xinjiang Medical University, Urumqi


Introduction
Periprosthetic joint infection (PJI) represents one of the most devastating complications that can occur following joint replacement surgery. Although the rates of PJI after primary arthroplasty have remained between 1% and 3%, recent analyses predict that the incidence of PJI will reach up to 4 million cases per year in the United States by 2030 1 . The increasing prevalence of PJI will result in a projected nancial burden for PJI in excess of $1.6 billion by 2020 1 . Despite recent advances in antiseptic protocols 2 , surgical techniques 3 and operating-room sterility 4 , efforts to prevent and treat PJI remain challenging.
Compared to other complications that can arise after joint replacement, PJI is di cult to treat due to the formation of a microbial bio lm that occurs as an adaptive response to hostile environments to protect the invading bacteria against the immune system and complement system of the host 5,6 . Speci cally, the bio lm augments bacterial resistance against the routine antibiotics by approximately 1000-fold 7 by impairing the penetration of antibiotics through the extracellular constituents 8 and by decreasing the metabolic activity of bio lm-embedded microorganisms 9 . Additionally bio lms promote the presence of slow-growing or quiescent "persister cells" 10,11 . Consequently, the higher values of minimum bio lm inhibitory concentration (MBIC) for the bacteria result in inadequate antibiotic treatment 12 , which can promote the appearance of mutated antibiotic-resistant strains 13 . Additionally, bacterial enterotoxins and the activation of polymorphonuclear neutrophils can elicit active bone resorption and inhibit bone formation, ultimately resulting in loosening of the prosthesis 14 . Therefore, there is an urgent need to develop strategies to disrupt the dense bio lm microarchitecture so as to promote the antibiotis ux towards deeper cell layers to kill bacteria within the bio lm.
The D-isoforms of amino acids (D-AAs) have been proved to break down bio lms in a diverse range of bacterial species, including Bacillus subtilis and Staphylococcus aureus 15 . In contrast to chemical agents used to disperse bio lms, D-AAs exhibit minimal cellular toxicity and can disturb the initial attachment of the bacteria to the surface while inhibiting subsequent growth of the microcolony into larger bacterial communities 16 . Moreover, the anti-bio lm activities of D-AAs are associated with multiple mechanisms, including the reduction of gene expression that is involved in extracellular matrix production 17 and the diminishment of the surface expression of bers that are required for bio lm formation that results from the incorporation of D-AAs into the bacterial cell wall 18 . However, it remains unclear if D-AAs are capable of disassembling bio lms and promoting antibacterial activity in the context of PJI.
Based on these previous ndings, we hypothesised that combining dispersal agents with antibiotics may provide an effective therapeutic strategy for PJI by functionally enhancing antimicrobial activity via the bio lm dispersion. To investigate this hypothesis, we assessed the e cacy of a combination therapy of D-AAs with vancomycin against a rat model of a Staphylococcus aureus-induced PJI.

Results
Effects of D-AAs-vancomycin combination therapy on systemic and local responses in vivo.
There were no signs of systemic illness in the rats that survived the postoperative period. For systemic response, a reduction in body weight was induced in the sham group at postoperative 8 weeks relative to that of the control group and vancomycin plus D-AAs group; however, there was no signi cant difference between the control group and vancomycin plus D-AAs group ( Fig. 2A). Moreover, the sham group exhibited a prominent increase in the levels of α2M, IL-1β, IL-6, IL-10, TNF-α and PGE2 compared to those of the control group, while PJI rats treated with vancomycin plus D-AAs but not vancomycin alone exhibited no statistical difference in these factors compared to these factors in the controls (Fig. 2B-G).
For local response, our ndings revealed that weight-bearing activity was markedly decreased following PJI and PJI rats treated with vancomycin alone, while the combination of D-AAs and vancomycin improved weight-bearing activity to a level similar to that of the controls (Fig. 3A,B). Moreover, rats provided with sham treatment, the sizes and widths of distal femurs were obviously reduced in rats treated with vancomycin or a D-AAs-vancomycin combination, and no statistically signi cant difference was observed between the D-AAs plus vancomycin and control groups ( Fig. 4A-C). Additionally, D-AAsvancomycin combination therapy clearly reduced the Rissing scale score of PJI rats to levels that were similar to those of the controls (Fig. 3C). Taken together, D-AAs-vancomycin combination therapy was more effective than was vancomycin alone in inhibiting infection-induced systemic and local responses.
Effects of D-AAs on bio lm formationin vivo and in vitro.
Scanning electron microscopy (SEM) was performed to con rm the e cacy of bio lm dispersion in response to D-AAs in rat PJI models. In the formal experiment, substantial bio lm formation and large clusters of cocci were visualised around the surface of the implant in the sham and vancomycin groups, while the D-AAs plus vancomycin group did not possess detectable levels of dense bio lm or even colonised bacteria, and this nding was similar to that from the control group (Fig. 5A). Moreover, D-AAs exhibited the capacity to dissociate bio lms in a dose-dependent manner; however, D-AAs failed to clear the colonised bacteria (Fig. 5B).
Subsequently, the activity of D-AAs in regard to bio lm disassembly and prevention was further determined in vitro using crystal violet staining. D-Trp, D-Pro, and D-Phe dispersed bio lms in a dosedependent manner and were most effective at concentrations of 10 mM (Fig. 6A,B). Additionally, D-Trp, D-Pro, and D-Phe prominently inhibited bio lm formation when Staphylococcus aureus was cultured in the presence of individual D-AAs (Fig. 6C). Importantly, the bio lm-dispersive activity of the mixture of D-AAs was elevated, and a reduction in bio lm biomass was observed at a concentration of 1 mM (Fig. 6B,C). In parallel, individual D-AAs at a concentration of 10 mM exhibited the ability to dissociate existing bio lms and to prevent bio lm formation by the clinical strains (Fig. 6D,E). Collectively, D-AAs possess the potential for anti-bio lm activity in vivo and in vitro.

Effects of D-AAs on the antibacterial activity of vancomycinin vivo and ex vivo.
To determine if bio lm disassembly by D-AAs promotes the ability of vancomycin to clear bacteria, CFUs were isolated from peri-implant bone, soft tissue, and implants. The CFUs for the tissue specimens were statistically higher in the sham group compared to those in the groups treated with vancomycin alone or a D-AAs-vancomycin combination (Fig. 7A,B). The implants in the sham group also exhibited signi cantly higher CFUs compared to those of the combined treatment group; however there were no differences in comparison to the group treated with vancomycin alone (Fig. 7C). Finally, µCT analyses and staining were applied to assess the changes in the microstructure and bone remodelling of the distal femur. The ndings revealed that vancomycin alone and the D-AAs-vancomycin combination treatment resulted in higher MBD, BV/TV, and Tb.Th values than those observed in shamtreated rats ( Fig. 8A-E). Moreover, no signi cant difference was noted between the D-AAs-vancomycintreated rats and the uninfected controls ( Fig. 8A-E). Additionally, the number of Trap + osteoclasts in the sham group was markedly reduced in the D-AAs plus vancomycin group (Fig. 9A,C). Minimal levels of osterix + cells were observed in infected rats, while the D-AAs-vancomycin combination restored osterix + cells to levels similar to those of uninfected rats (Fig. 9B,D). Notably, the number of osterix + cells failed to recover in response to treatment with vancomycin alone. Collectively, abnormal bone remodelling around the prosthesis was prominently attenuated by bio lm disruption and bacterial elimination.

Discussion
This study was the rst to assess the e cacy of a D-AAs-vancomycin combination therapy using a rat PJI model. As expected, although systemic vancomycin improved systemic and local responses and decreased bacterial burden, this therapy failed to eliminate the infection in all cases. However, the combination treatment of D-AAs and vancomycin achieved a marked therapeutic bene t relative to that of vancomycin alone. Mechanistically, D-AAs prevented bio lm formation and disassembled established bio lms, thus facilitating the diffusion of vancomycin into the deeper cell layers to eradicate bacteria. Additionally, we also found that the D-AAs-vancomycin combination therapy was highly effective at redressing abnormal bone remodelling by re-establishing the dynamic balance between bone resorption and formation.
Despite meticulous clinical management that incorporates irrigation, surgical debridement, and the use of antibiotics 19 , a small percentage of "persister cells" in bio lms can still survive and rapidly grow after the cessation of antimicrobial therapy, thus leading to recurrent PJIs 20 . Therefore, dissociation of these bio lms may provide a vital target for PJI therapy. Recent studies have shown considerable interest in bio lm dispersal agents such as D-AAs 21 , quorum-sensing inhibitors 22 , and vapour nanobubbles 23 . A previous study demonstrated the e cacy of the local delivery of D-AAs in eliminating bacterial contamination by targeting bacteria within bio lms 24 . Similarly, a rat model in our study strictly mimicked clinical PJI, where mature bio lm formation occurred 2 weeks after bacterial colonisation on the prosthesis surface 25 . Notably, a mixture of D-AAs that included D-Trp, D-Pro, and D-Phe was active in not only dispersing mature bio lms but also in inhibiting bio lm formation to expose colonised bacteria to vancomycin to achieve a stronger sterilisation effect. Moreover, our in vitro ndings also con rmed that the concentrations of ≥ 1 mM of D-Trp, D-Pro and D-Phe prevent bio lm formation and disrupt the existing bio lm integrity of Staphylococcus aureus. Speci cally, the identi ed concentrations exhibited the effects of bio lm-dispersive and anti-bio lm activity against various clinical strains. However, strain heterogeneity and different bacterial species result in a discrepancy in regard to anti-bio lm activity.  27 that demonstrated that an equimolar mixture of D-AAs shifted the dose-response curve toward lower doses compared to that of the individual D-AAs. Collectively, the anti-bio lm activity of D-AAs highlights their potential usefulness in the treatment of PJI.
The cell wall thickness of S. aureus located in bio lms gradually increases 25 , thereby resulting in reduced susceptibility to vancomycin 28 . In this case, the effect of vancomycin on inhibiting bacterial cell wall synthesis was signi cantly reduced 29 . In contrast, D-AAs can modulate cell wall remodelling in bacteria by causing the release of amyloid bres that link the cells in the bio lm together 15 . Although D-AAs are unable to kill bacteria 30 , they can likely provide an effective adjuvant therapy that can be used in combination with antibiotics. Indeed, our results revealed that the use of D-AAs signi cantly elevated the bactericidal activity of vancomycin, where 1-to 2-log CFUs decreases were observed compared to decreases caused by the agent alone. Furthermore, combining D-AAs with vancomycin therapy is more effective than vancomycin therapy is more effective than is the use of vancomycin alone in regard to the reduction of systemic and local reactions and the attenuation of X-ray results. This result was somewhat unanticipated, as the vancomycin-D-AAs combination was applied in the absence of debridement or irrigation, both of which were routinely performed in our previous clinical studies 31 . This may be attributed to the possibility that interventions using D-AAs augment the ability of drugs to diffuse into deeper cell layers by establishing more cracks between bacteria to ultimately eradicate these bacteria. Bacteria or their components that can induce abnormal bone remodelling are the main causes of prosthesis loosening 14 . On the one hand, osteoclastic activity can be stimulated by S. aureus, including protein A 33 , or its secreted substances 34 . It has been observed that soluble factors produced by S. aureus can reduce osteoblast viability and increase cell apoptosis and these factors can inhibit bone mineralisation 35 . Taken together, these factors drive the dynamic balance of bone remodelling towards osteolysis. Our micro-CT results also showed that the MBD, BV/TV, and Tb.Th values of the distal femur were decreased in the sham group. In parallel with the results of microarchitecture analyses, our study found increased Trap + osteoclasts and reduced osterix + osteo-progenitors around the prosthesis. However, D-AAs-vancomycin combination therapy alleviated this aberrant microstructure by reducing the number of osteoclasts and osteoblasts to normal levels, thereby stabilising the joint prosthesis.
Due to the existence of bio lms, PJI patients must be administered antibiotics for long periods of time and at high doses to control infection, and this can lead to a number of risks to patient health and impose an economic burden on the healthcare system. Fortunately, our study indicated that the combination treatment of D-AAs and vancomycin represents a potentially e cacious therapy for PJIs, as the disassembly of bio lms results in an increase in antibiotic activity, ultimately improving abnormal bone remodelling, and preventing prosthesis loosening.

Materials And Methods
Animals.
Ethical statement. All procedures complied with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care, and the protocol was approved by the Institutional Animal Care and Use Committee of First A liated Hospital of Xinjiang Medical University (protocol number IACUC20191011-01).
Amino acids and bacterial strain.
D-isomers of amino acids, including phenylalanine (Phe), proline (Pro) and tryptophan (Trp), were purchased from Sigma. For bacterial cultures, D-AA stocks were prepared in 0.5 M HCl at concentrations between 150 and 200 mM. There were diluted into Mueller Hinton (MHB-II) broth that was neutralised to pH 7.4, and the stocks were then stored at -80 °C. Staphylococcus aureus ATCC 25923 was used for this study. Four clinical Staphylococcus aureus strains that were characterised according to bio lm formation were isolated from PJI patients as described in our previous study 31 (Supplementary Table 1). For all studies, the strains were cultured at 37 °C overnight in tryptic soy broth (TSB) media with agitation.
Induction and treatment of PJI in a rat model.
Eight-week-old male Sprague-Dawley (SD) rats were purchased from the Animal Centre of Xinjiang Medical University. Animal handling conditions included a humidity of 55 ± 5%, a temperature of 25 ± 2°C , and a 12 h light/dark cycle. All rats were maintained under speci c pathogen-free conditions and provided with autoclaved food and water ad libitum. The rats were anaesthetized using 10% hydrated chloral solution (Aladdin, 4 mL kg −1 body weight). The joint capsule of the right knee was opened through a medial parapatellar arthrotomy (Fig. 1A). After exposing the intercondylar notch, the femoral canal was reamed with sequentially larger needles until an orthopaedic-grade Kirschner wire (1.0 mm in diameter and 20.0 mm in length; Synthes) was exactly inserted in a retrograde fashion with 1 mm of wire protruding into the joint space (Fig. 1B-D). The arthrotomy site was sutured using with interrupted 4 # Monocryl (Ethicon) and then injected with Staphylococcus aureus (1×10 4 colony-forming units [CFUs] in 10 µL saline) (Fig. 1E,F). Finally, the skin was closed. Beginning the second week afer Kirschner wire implantation, a therapeutic dose of vancomycin (110 mg/kg twice daily) (Novaplus; Hospira, Inc., Lake Forest, IL) was administered by subcutaneous injection in the vancomycin gruop. On this basis, D-AAs were injected into the articular cavity once weekly in the vancomycin plus D-AAs group.
A preliminary experiment was performed rst to identify the optimal dose of D-AAs. Rats were randomly assigned to the control group, sham group, vancomycin group, and D-AAs groups at various concentrations (0.5, 1, and 10 mM; n = 8 per group). After treatment for 6 weeks, lower concentrations of D-AAs (0.5 or 1 mM) exerted minimal effects on the local response ( Supplementary Fig. 1). Based on this, in the formal experiment, the rats were randomised to the control group, sham group, vancomycin group, and a 10 mM D-AAs combined with vancomycin group ( n = 20 per group, where 8 rats were used for tissue homogenate, 8 rats were used for immunostaining and 4 rats were used for SEM analysis). Additionally, 16 rats were randomly assigned to the sham group and various concentrations of D-AAs groups (0.5, 1, and 10 mM) to evaluate the bio lm lysis potential of D-AAs alone after 6 weeks by SEM.
Systemic and local response analysis.
The weights of all rats were measured and recorded once every two weeks. Serum samples were collected from the left ventricle immediately after sacri ce. The serum concentrations of α2M, IL-1β, IL-6, IL-10, TNF-α, and PGE2 were determined by ELISA according to the manufacturer's instructions (CUSABIO, China). ELISA results were quanti ed according to absorbance at 450 nm as assed using a microplate reader (Bio-Rad, Hercules, CA, USA), and these values were normalised according to the number of cells per well.
The weight-bearing activity of rats was assessed using ink blot analysis and was graded for each rat as full (3 points), partial (2 points), toe-touch (1 point), or non-weight-bearing (0 points) 36 . The front paws of the rats were covered with dark blue ink, and the hind paws were covered with red ink.
Radiographs were assessed using Image J software. The maximal femoral width was calculated perpendicular to the anatomical axis of the distal portion of the femur. The area of the distal 25% of the femur (from the midpoint of a line extending from the intercondylar notch to its intersection with a perpendicular line that bisected the third trochanter) was also measured. The local tissue response was evaluated at the time of implant-bone harvest according to the Rissing score 37 .

SEM analysis.
After acquisition, samples were xed with 2.5% (w/v) glutaraldehyde and 0.15 M sodium cacodylate buffer for 3 h. Samples were then rinsed with 0.15 M sodium cacodylate buffer and xed for 1h in 1% (v/v) osmium tetroxide in sodium cacodylate buffer. Samples were dehydrated with ethanol and then incubated with hexamethyldisilizane, and this was followed by drying in a desiccator overnight. Samples were sputter-coated with gold palladium and observed using a JEOL-6610 scanning electron microscope.
Bio lm formation and dispersal assays.
Bio lm formation was assessed under static conditions for 24 h in polystyrene 24-well plates (Corning, Inc., Corning, NY, USA). Brie y, after overnight incubation, the culture medium was removed and fresh medium containing either an individual D-AA or a 1:1:1 mixture of D-Trp: D-Pro: D-Phe was added at the indicated concentrations. After incubation for 24 h, the wells were washed with phosphate buffered saline (PBS) and then stained with 0.1% (w/v) crystal violet (Sigma Aldrich, St. Louis, MI, USA) at room temperature for 15 min. Next, bio lm biomass was examined by measuring the optical density at 570 nm of the crystal violet that was solubilised in 80% (v/v) ethanol. All assays were performed in triplicate.
Ex vivo bacterial burden.
After euthanasia, the peri-implant bone and soft tissues were harvested along with the implanted K-wires. The bone and soft tissues were homogenised using a sterile tissue grinder, and this was followed by inoculation of 20 μL aliquots onto sheep-blood agar plates (Hardy Diagnostics) for 24 h at 37 °C. The Kwires were sonicated for 10 min and then vortexed for 2 min. Subsequently, 20 μL of sonicated uid was plated and incubated as described for tissue cultures. The number of CFUs was counted after overnight incubation of the plates. Additionally, to further con rm if the bone tissues, soft tissues, or K-wires retained any bacteria, the homogenates and sonicates were cultured again for an additional 48 h at 37°C. The presence or absence of bacterial CFUs was determined by assessing for the presence or absence of CFUs after 48 h culture of the plates.
A high-resolution micro-CT (µCT) was conducted using a SkyScan 1172 Scanner. The data were subsequently reconstructed (NRecon v1.6), analysed (CTAN, v1.9), and re-established for 3D model visualisation (CTVol, v2.0). The coronal view of the 1.5 cm distal femur was selected for 3D histomorphometric analysis. Around the K-wire, a 3 mm region was identi ed as the region of interest.
Histological analysis and immunostaining.
Following euthanasia, the right knee joints of rats were harvested and xed in 4% paraformaldehyde for 24 h. Then, the knee joints were decalci ed for 3 weeks and embedded in para n. Sagittal sections of the femur were processed for tartrate-resistant acid phosphatase (TRAP) staining and immunostaining.
For immunostaining, the sections were rehydrated and quenched with endogenous peroxidase before treatment with 0.1% trypsin for 30 min at 37 °C to retrieve the antigen. Then, 20% normal horse serum was used to block the sections to reduce non-speci c staining. Sections were then incubated with primary antibodies against osterix (Abcam, 1:400, ab22552). A horseradish peroxidase streptavidin detection system (ZSGB BIO) was used to detect the immunoactivity, and this was followed by counterstaining with haematoxylin (ZSGB BIO). The number of positively stained cells was determined in a blinded manner using cellSens software (Olympus, Int, USA).
Statistical analysis.
Graphpad Prism 5.0 was used to analyse data and to draw diagrams. Signi cance was determined using a One-Way ANOVA for the comparison of animal systemic and local responses, radiographic ndings, and µCT ndings using the nonparametric Wilcoxon rank-sum test for the comparison of ex vivo CFUs between different treatment conditions and using the Fisher exact test for the comparison of the percentages of cultures that exhibited any bacterial growth. A p value of < 0.05 was considered signi cant.

Declarations
Data availability statement.
The datasets generated for this study are available upon request from the corresponding author.