Hydroxyapatite: An antibiotic recruiting moiety for local treatment and prevention of bone infections

Treatment of chronic osteomyelitis by radical debridement and filling of the dead space with antibiotic containing calcium sulfate/hydroxyapatite (CaS/HA) bone substitute has shown excellent long‐term outcomes. However, in extensive infections, sessile bacteria may remain in bone cells or soft tissues protected by biofilm leading to recurrences. The primary aim of this study was to evaluate if systemically administrated tetracycline (TET) could bind to pre‐implanted HA particles and impart an antibacterial effect locally. In vitro studies indicated that the binding of TET to nano‐ and micro‐sized HA particles was rapid and plateaued already at 1 h. Since protein passivation of HA after in‐vivo implantation could affect HA–TET interaction, we investigated the effect of serum exposure on HA–TET binding in an antibacterial assay. Although, serum exposure reduced the zone of inhibition (ZOI) of Staphylococcus aureus, a significant ZOI could still be observed after pre‐incubation of HA with serum. We could in addition show that zoledronic acid (ZA) competes for the same binding sites as TET and that exposure to high doses of ZA led to reduced TET–HA binding. In an in‐vivo setting, we then confirmed that systemically administered TET seeks HA particles that were pre‐implanted in muscle and subcutaneous pouches in rats and mice respectively, preventing HA particles from being colonized by S. aureus. Clinical Significance: This study describes a new drug delivery method that could prevent bacterial colonization of a HA biomaterial and reduce recurrences in bone infection.

2][13] However, after a few weeks, when the calcium sulfate is dissolved and the local antibiotic concentrations drops, bacteria could repopulate on apatite and lead to a relapse, especially in severe infections. 14Clinically, there is hitherto no proven method available to protect the remaining particulate apatite from bacterial recolonization.A promising scenario could be if systemically used antibiotics could seek and biologically activate local apatite to exert an antibacterial effect.
In a recent study, Raina et al. introduced the term "biomodulation" as a new concept of targeted delivery where locally implanted HA particles acted as a recruiting moiety for systemically administered biologically active molecules such as zoledronic acid (ZA), tetracycline (TET) and 18 F-fluoride. 15This drug delivery method was highlighted in an editorial in the same journal and described as a novel mechanism for advancing bone drug targeting. 16Systemically administered ZA not only binds to HA particles but also activates the HA particles leading to significantly higher bone formation in a boneimplant integration model. 15Using TET, the authors demonstrated that antibiotics could also be recruited by HA particles.However, the antibacterial effect of TET accreted to nano (n)-and micro (m)-HA particles or the in-vivo preventive effect, and potential interference of biological macromolecules like serum proteins in TET-HA interactions were not presented in the earlier study. 15T was discovered in 1948 by Benjamin Minge Duggar. 17It is a broad-spectrum antibiotic with an antibacterial effect against a wide range of gram-positive and gram-negative bacteria. 18The Swedish radiologist Torsten André was the first to describe TET binding to bone in humans in 1956. 19TET was reported to bind to calcium, hydroxyl, and phosphate groups also in dead bone by Perrin in   1965. 20Due to its fluorescence characteristics, systemic administration of TET has been used to measure bone growth in humans. 21In 1993, a study was published on 1894 femoral neck fractures that were included in a prospective analysis of bone turnover using systemic TET as a fluorescent bone marker. 22As a separate observation, the authors reported that patients administered TET pre-or per-operative for measuring bone turnover had significantly fewer deep infections than those without TET.Overall, these studies underline the strong binding of TET to bone and recent reports further confirm the affinity of TET to synthetic HA, which is the major mineral component of bone. 15, therefore, hypothesized that using synthetic HA particles as a recruiting moiety, specific antibiotics administered via systemic circulation could be recruited to the targeted tissue, and exert a local antibacterial effect.We first explored in-vitro if the HA particle size, exposure time to TET, and competing protein or bisphosphonate binding to HA would influence the HA-TET interaction.We then used two in-vivo models to demonstrate that TET biomodulated HA can both be used as a treatment and as a method to prevent bacteria from colonizing an HA-based material.In the first in-vivo model in rats, synthetic n-/m-HA particles were implanted in rats followed by systemically administering TET at clinical doses.The biomaterial pellets were then harvested and their antibacterial effect was tested in-vitro on both reference and clinical strains of S. aureus.Further, in a mouse model of subcutaneous infection, the ability of the biomodulated HA particles to prevent bacterial infection was studied.

| Study design
The experiments performed in this study were divided into two parts: in vitro and in vivo experiments.The in vitro part was initiated by evaluating how exposure time of HA to TET affects the binding kinetics of TET-HA.This was done by exposing clinical concentrations of TET to n-/m-HA and mixing them for different time intervals.
In the second in vitro experiment, by exposing HA particles to fetal bovine serum, we studied the effect of serum proteins on the interaction of TET-HA.In the final in vitro experiment, we tried to saturate the TET binding sites of n-HA particles by treating them with ZA, a drug known for its high HA affinity.We then performed two in vivo studies to confirm the TET-HA binding.In the first study, using a rat muscle-pouch model, TET biomodulation was tested by implanting n-and m-HA pellets in an abdominal muscle pouch model followed by systemic TET administration.The antibacterial effects of the TET biomodulated HA pellets were tested by harvesting the pellets followed by agar diffusion test in-vitro.In the final experiment, the infection prevention ability of TET biomodulated n-HA pellets was studied using a subcutaneous mouse infection model.A detailed overview of the study design is provided in the schematic below (Figure 1).

| In vitro binding experiment
To understand the binding rate of HA particles and antibiotics under different exposure time conditions, we conducted an in-vitro binding experiment.n-/m-HA powder (100 mg; Fluidnova) was measured in 2 mL microcentrifuge tubes in triplicates and 1 mL of TET solution (4 μg/mL, serum concentration of TET achieved clinically) was added.
After vortex mixing, the tubes were kept on a rotator at 180 RPM for 1, 6, 12, and 24 h.At desired time points, the tubes were removed and centrifuged at 14,000 RPM for 2 min.The supernatant was discarded and sediments were then washed three times by adding SEBASTIAN ET AL.
| 213 1 mL of saline (NaCl 0.9%) followed by vortex mixing and centrifuging at 14,000 RPM for 2 min.After three washing-cycles, supernatants were discarded and the HA sediments were kept at −20°C before testing their antibacterial effects.Using Kirby-Bauer disk diffusion assay, antibacterial effect of the obtained paste was tested by placing them on the Muller Hinton agar (MHA) plates inoculated with bacterial suspension of S. aureus ATCC 25923 (OD 600 = 0.1).After overnight incubation at 37°C, the diameter of the zone of inhibition (ZOI) around the paste were measured.

| Protein passivation test
To understand the interaction between serum proteins and the HA binding sites for TET, a protein passivation test was performed.
Briefly, n-/m-HA powder (100 mg; Fluidnova) was measured in 2 mL microcentrifuge tubes in triplicates.To the test tubes, 1 mL of undiluted FBS was added whereas control tubes received 1 mL saline.
After vortex mixing, tubes were kept on a rotator at 180 RPM.After 24 h, the tubes were removed and centrifuged at 14,000 RPM for 2 min.The supernatant was discarded and sediments were then washed three times by adding 1 mL of saline followed by vortex mixing and centrifuging at 14,000 RPM for 2 min.After three washing-cycles, supernatants from the test samples were discarded and 1 mL of TET (4 μg/mL) prepared in undiluted FBS was added to the sediments and mixed.To one set of saline treated control nano-/ micro-HA samples, 1 mL of TET (4 μg/mL) prepared in undiluted FBS was added whereas the other set of samples received 1 mL of TET (4 μg/mL) prepared in saline.After 1 and 24 h of mixing, tubes were removed and centrifuged at 14,000 RPM for 2 min.The supernatants were discarded and sediments were then washed and their antibacterial effects were tested as mentioned in Section 2.3.

| Saturation test using zoledronic acid
To further confirm the binding between TET and synthetic HA particles, n-HA was pretreated with ZA (Novartis, Switzerland, concentration 4 mg/5 mL), which is known to have very strong binding with HA. 1 mL of varying concentrations of ZA (10, 100, and 250 μg/mL) was added to the microcentrifuge tubes containing n-HA powder (100 mg).After vortex mixing, tubes were kept on a rotator at 180 RPM for 24 h.After 24 h, the tubes were removed and centrifuged at 14,000 RPM for 2 min.The supernatant was discarded F I G U R E 1 Overview of the experimental design used in the study.and sediments were washed three times by adding 1 mL of saline followed by vortex mixing and centrifuging at 14,000 RPM for 2 min.
After three washing-cycles, supernatants were discarded and 1 mL of TET (4 μg/mL) was added to the sediments and mixed for 24 h.Nano-HA without any ZA pretreatment mixed with TET, and ZA-pretreated n-HA without any TET mixing were used as controls.After 24 h of mixing, the tubes were centrifuged at 14,000 RPM for 2 min and the supernatants were then discarded.As mentioned in Section 2.3, sediments were then washed and their antibacterial effect was tested.

| In vivo HA biomodulation and its antibacterial effect
All animal experiments were conducted with prior consent from the Swedish Board of Agriculture (Jordbruksverket) following protocol number 15288/2019.All personnel working with animals had received prior training and certification as per the FELASA guidelines.
Animals were housed in pairs in ventilated and temperaturecontrolled rooms with 12 h day and night cycles and free access to food and water.Wound healing and animal health (fur condition, body posture, and mobility) was monitored twice daily for 72-h postsurgery to comply with humane endpoints.No animals were removed from the analysis of data.
Pellets of n-/m-HA were prepared by mixing 100 mg of n-/m-HA (Fluidnova) with 60 and 80 µL of hyaluronic acid (HAD, 1 mg/mL), respectively.Using the mixture, pellets were cast by transferring them to an elastic mold (Ø = 4.7 mm) and placed at −20°C until further use.Following an established muscle pouch model (17), six male Sprague-Dawley (SD) rats (average weight 353 g) were operated in the abdominal muscle (under isoflurane anesthesia and buprenorphine analgesia), and each rat received m-HA (n = 6) and n-HA (n = 6) pellets on the right and left side of the abdominal midline, respectively. 23Due to only one treatment group, animals were not randomized at the time of surgery.At 7 days post-surgery, all animals received one intraperitoneal injection of TET (25 mg/rat) every day for three continuous days.24 h after the third injection, the animals were sacrificed.Implanted pellets devoid of muscle tissue were retrieved from the animals and placed in a 2 mL microcentrifuge tube.
Similarly, bone (trabecular and cortical), kidney, and liver tissues were collected from n = 3 animals, and muscle tissue from all animals (n = 6).All samples were placed in liquid nitrogen and immediately crushed to form a slurry.The slurry was then put through the disk diffusion assay as described above, using S. aureus ATCC 25923.In case of muscle tissue and n-/m-HA, by using same slurry and moving them to new plates, the above process was repeated for three consecutive days but on the third day the antibacterial effect was tested on a clinical strain, S. aureus P3 isolated from a case of prosthetic joint infection at our hospital.Both S. aureus ATCC 25923 and S. aureus P3 were TET sensitive strains.

| TET biomodulated n-HA and its efficacy in infection prevention
Pellets of n-HA were prepared in a sterile environment by mixing n-HA particles with 1% (w/v) HAD to form a paste.The paste was then transferred into a nylon mold and casted into hemispherical pellets of 3.2 mm diameter.Only n-HA was chosen in this experiment because it was established in the earlier in-vitro and in vivo experiments that n-HA was better at TET accretion than m-HA.
In Group 1, BALB/c mice (5 mice, total n = 5) were anesthetized using 2%-4% (4% for induction and 2% for maintenance) isoflurane followed by shaving the hair on their back.With the tip of a sterile scissor, a 0.5 cm incision was made on the right dorsum and one pellet of n-HA was implanted in the subcutaneous pocket in each mouse (n = 5 n-HA pellets in total).In Group 2 (3 mice, total n = 3), two pellets of n-HA were implanted on either side of the dorsum in a similar fashion as described above (n = 6 n-HA pellets in total).The wound was closed using a Vicryl suture.Three days postimplantation, the animals in Group 1 (i.e., animals with one n-HA pellet) were intraperitoneally injected with TET suspended in normal saline at a dose of 20 mg/kg body weight.TET administration was performed once every day on postoperative days 3, 4, and 5. Three days after the last tetracycline injection, that is, postoperative day 8, animals were injected with a bioluminescent strain of S. aureus SAP229 (kindly provided by Dr. Roger D. Plaut, Division of Bacterial, Parasitic, and Allergenic Products, FDA) at a dose of 10 6 CFU (in 50 μL Tris buffer).Bacteria were subcutaneously introduced at two locations; (1) on the right side where the n-HA pellet was implanted and (2) on the left side parallel to Site 1, which acted as a non-n-HA control.On Day 8, all mice in Group 2 (i.e., animals with 2 n-HA pellets/mouse without TET administration) also received the same number of bacteria in both n-HA implantation sites.At 30 min and 6 h after bacterial inoculation, animals were imaged using an in vivo imaging system (IVIS spectrum, PerkinElmer).The bioluminescence signal at the injection sites was quantified as a surrogate marker for bacterial proliferation at the site of infection.Bioluminescent signals from the mice were quantified using Living Image 4.0 Software (PerkinElmer).
After the last IVIS imaging, mice were killed and the implanted materials and/or surrounding subcutaneous tissue from the infected site were harvested and collected in 0.5 mL ice-cold PBS.For CFU analysis, collected materials and tissues were homogenized and serial dilutions were plated on TH broth agar and incubated overnight at 37°C.

| Statistical analysis
All data were presented as mean ± SD.Details of statistical analysis are indicated in each figure legend.All data processing was carried out on GraphPad Prism version 9.1.2for MacOS (GraphPad Software).

| In vitro binding experiment
At the tested time points, TET exhibited a uniform binding with both nand m-HA particles which was evident with almost similar ZOI obtained by disk diffusion assay done on the sediments from different time intervals (Figure 2A,B).With uniform and similar ZOI's exhibited by both nano-and micro-HA particles from 1 to 24 h time points, it was confirmed that TET binding to HA particles is rather quick.Compared with ZOI exhibited by m-HA particles, n-HA particles showed larger ZOI indicating a better binding between n-HA and TET (Figure 2B).

| Protein passivation test
The protein passivation of the n-/m-HA particles with FBS significantly affected the ability of TET to chemically interact with HA (Figure 2C-F).Although protein passivation affected the total binding capacity of TET to both type of HA particles, the ZOI of protein passivated HA-TET particles indicates that protein-HA interaction did not completely hinder TET-HA binding and they acquired adequate antibacterial property as well.

| Saturation test using zoledronic acid
As reflected in the ZOI's, saturation of TET binding sites of n-HA was directly associated with the concentration of ZA (Figure 3A,B).At lower concentrations of ZA, no significant effect on the TET-HA interaction could be observed.At the highest ZA concentration, almost no TET-HA binding occurred and the corresponding ZOI was 0 mm.

| In vivo biomodulation and its antibacterial effect
The ZOI of kidney samples (14-17 mm) were higher than that of liver (10-12 mm), mirroring the route of excretion of TET and validating the experiment model (Figure 4B,D).The antibacterial effect of both trabecular and cortical bone samples confirmed the affinity of TET for HA (Figure 4B,D).Of the six muscle samples, two of them showed a narrow ZOI on Day-1 that disappeared by Day-2 (Figure 4E).Both nand m-HA exhibited ZOI, confirming their biomodulation by systemic TET administration (Figure 4C).Compared with m-HA and other tissue samples, n-HA showed a strong antibacterial effect as evident by the ZOI on Day-1 (Figure 4D,E).Even on Day-3, both n-/m-HA exhibited antibacterial effects against the clinical strain S. aureus P3 (Figure 4E).

| TET biomodulated n-HA and infection prevention
As seen from Figure 5, irrespective of the time interval, the animals in  side compared with the left (non-n-HA) side signifying that n-HA was protected by the accreted TET.The bioluminescence signal from the animals in Group 2 (irrespective of the side) was similar to the non-n-HA side of the animals from group 1 indicating that pristine n-HA alone does not impart any antibacterial activity (Figure 5A,B).Upon quantifying the total bioluminescence from animals in Group 1 and Group 2, the n-HA side of group 1 showed significantly lower bioluminescence when compared to the left side of Group 1 or either side in Group 2.
As expected, with the passage of time (i.e., 30 min vs. 6 h), the bioluminescence increased in the non-n-HA side in Group 1 and in Group 2, indicating continued bacterial growth (Figure 5B).This change stagnated in the n-HA side in Group 1 indicating that the bound TET to HA conferred a protective effect on the n-HA particles and stopped further bacterial growth.The bioluminescence data were also confirmed by performing a CFU assay and it was observed that the n-HA side in Group 1 had significantly lower CFU units/g tissue when compared with non-n-HA side in Group 1 (p < 0.01) and Group 2 (p < 0.05) (Figure 5C).

| DISCUSSION
In the present study, we have shown that by systemically administering TET at serum concentrations reported clinically, TET can be recruited by synthetic HA particles placed in a targeted site and exert an adequate antibacterial effect.5][26] It is compelling that some antibiotics for bone infection such as rifampicin and tetracycline have a chemical structure indicating that they could bind to apatite. 15,27Recently, nanoparticles of HA with a proven tumor drug accretion (doxorubicin) are reported to be internalized into cells retarding osteosarcoma growth. 28This opens up possibilities for the development of a new drug delivery modality for intracellular drug delivery with apatite-bound antibiotics to eradicate bacteria known to reside intracellularly.
Irrespective of the size of the HA particles used, the chemical interaction of TET and HA was found to be quick and remained constant at tested time points.In an in-vitro study, Cazalbou et al.,   showed that TET adsorption kinetics on apatite particles reached quasi-equilibrium in 30 min, which corroborates with our current findings. 29This is promising, since during in-vivo conditions, the drug-HA interaction will be for a short time with a half-life of TET being 6-11 h. 30Moreover, it is important to highlight that all our in-vitro results were obtained with serum concentration of TET that is normally achieved clinically.Although in the present study, the exact mechanism behind TET-HA interaction was not delineated, but various factors including the chemical structure of the drug, and physio-chemical properties of HA such as its porosity, degree of crystallinity, number, and size of particles could play an important role in drug-HA binding. 31Song et al., evaluated the interaction mechanism of TET-HA, and observed a strong affinity that formed via hydrogen, covalent bonds, and weak van der Waals interactions. 32 has been reported that various biological macromolecules such as serum proteins bind to HA surface. 33By implanting HA particles in an ectopic muscle pouch model in rats, Raina et al. reported that ZA, an antiresorptive bisphosphonate administered systemically twoweeks postimplantation not only sought HA moiety but it was possible to reload them when ZA was given at 4 weeks. 15In a recent study, Liu et al. showed that protein passivation of HA surfaces both in vitro and in in vivo conditions did not significantly affect the ability of doxorubicin, an anticancer drug used clinically, to chemically interact with HA. 28 In contrast to these reports, following protein passivation, we did observe a significant change in the interaction of TET-HA.However, the HA particles possessed sufficient antibacterial effect.Taken together, our results as well as previous reports indicate that even after continuous exposure to serum proteins, HAseeking biological molecules including antibiotics such as TET could still interact specifically with HA particles.
The ZA saturation experiment reflects a competitive binding scenario.ZA which has a high affinity to HA, outperformed TET in HA interaction at high concentrations. 34Saturation of HA using lower concentrations of ZA did not have any notable effect on TET-HA interaction.Also, it has been reported that compared to n-HA, ZA has more affinity to m-HA. 15If we had used only m-HA, even lower concentrations of ZA may have had a significant effect on TET-HA interaction.However, we selected n-HA based on the larger ZOI obtained from the in-vivo biomodulation experiment.Our results suggest that in a clinical scenario involving particularly the elderly, where administration of a potent bisphosphonate such as ZA is foreseen, TET should be administered before the administration of ZA to achieve better antibacterial effects of TET.
Using a clinically mimicking dose, our in vivo experiment results further validated our in vitro findings on TET-HA interaction.Serum levels of TET achieved in vivo not only biomodulated the synthetic HA particles but were effective in imparting antibacterial properties to HA even 3 days after harvest.Nevertheless, we tested the antibacterial effect on both standard and clinical strains from a PJI.
Since it has been reported that reloading of HA with the same drug is possible, we speculate that repeated TET injections may have increased the total accretion of TET on HA particles. 15Similar to the differences in ZOI's observed between the type of HA particles used in the time kinetics experiment, nano-HA had a stronger affinity for TET compared to micro-HA.In congruence with our findings, Liu et al. reported significantly stronger binding of doxorubicin to n-HA than m-HA. 28However, few reports have contradictory findings with more binding of biological molecules to microparticles than nanosized HA particles. 15,35Although binding efficiency of HA and TET could be evaluated through various methods such as spectrofluorimetry, microscopy, or histological analysis we focused on measuring the ZOI obtained through agar diffusion assay used as a surrogate marker for the analysis of HA-TET binding both in vitro and in vivo. 15,28It was mainly because our main goal was to protect the

F
I G U R E 2 Tetracycline binding to hydroxyapatite: Time kinetics and protein passivation.(A, B) Interaction of n-and m-HA particles with TET studied as a function of time.(C-F) Effect of protein passivation on the interaction of TET to n-/m-HA.A student's t-test was used to compare the in vitro binding of n-and m-HA particles with or without protein passivation.**p < 0.01, and ***p < 0.001.FBS, fetal bovine serum; m-HA, micro-hydroxyapatite; n-HA, nano-hydroxyapatite; TET, tetracycline.
group 1 had a reduced bioluminescence signal on the right (n-HA) F I G U R E 3 Effect of zoledronic acid on tetracycline binding to hydroxyapatite.(A) Muller Hinton agar plates inoculated with Staphylococcus aureus ATCC 25923 showing ZOI of nano-hydroxyapatite (n-HA) particles with and without pretreatment of zoledronic acid (ZA) exposed to clinically achieved serum concentration levels of tetracycline (TET).(B) A graph data showing the antibacterial effect of the ZA pretreated n-HA particles following exposure to TET.F I G U R E 4 Systemically administered tetracycline seeks hydroxyapatite particles and provides antibacterial effects to them.(A) Schematic of the experimental procedure.(B, C) Muller Hinton agar plates inoculated with Staphylococcus aureus ATCC 25923 showing zone of inhibition (ZOI) of various tissue samples and hydroxyapatite (HA) particles collected from rats who received three doses of systemic tetracycline (TET) injections.(D) Graph showing the ZOI of tissue samples harvested following TET biomodulation.The red-dotted arrow corresponds to the average ZOI of muscle tissue (2.5 mm).(E) Graph showing the antibacterial effect of HA particles and muscle tissue obtained from continuous disk diffusion assay.

F I G U R E 5
Antibacterial effect of tetracycline biomodulated hydroxyapatite particles in a mouse model of sub-cutaneous Staphylococcus aureus infection.(A) shows IVIS images of three mice (left and middle shows images of mice from Group 1 and the image in the right shows a mouse from Group 2) taken 30 min after local bacterial infection and the graph on the right provides a quantification of the bioluminescence measured at the site of infection after 30 min of infection.(B) IVIS images depicting three mice (left and middle shows images of mice from Group 1 and the image in the right shows a mouse from Group 2) taken 6 h after local bacterial infection and the graph on the right provides a quantification of the bioluminescence measured at the site of infection at 6 h postinfection.(C) Presents the data from the colony forming unit assay after harvesting the infected tissues from the killed animals.Statistical.