Osseointegration and antibacterial effect of an antimicrobial peptide releasing mesoporous titania implant.

Medical devices such as orthopedic and dental implants may get infected by bacteria, which results in treatment using antibiotics. Since antibiotic resistance is increasing in society there is a need of finding alternative strategies for infection control. One potential strategy is the use of antimicrobial peptides, AMPs. In this study, we investigated the antibiofilm effect of the AMP, RRP9W4N, using a local drug-delivery system based on mesoporous titania covered titanium implants. Biofilm formation was studied in vitro using a safranine biofilm assay and LIVE/DEAD staining. Moreover, we investigated what effect the AMP had on osseointegration of commercially available titanium implants in vivo, using a rabbit tibia model. The results showed a sustained release of AMP with equal or even better antibiofilm properties than the traditionally used antibiotic Cloxacillin. In addition, no negative effects on osseointegration in vivo was observed. These combined results demonstrate the potential of using mesoporous titania as an AMP delivery system and the potential use of the AMP RRP9W4N for infection control of osseointegrating implants.

tigated the antibiofilm effect of the AMP, RRP9W4N, using a local drug-delivery system based on mesoporous titania covered titanium implants. Biofilm formation was studied in vitro using a safranine biofilm assay and LIVE/DEAD staining. Moreover, we investigated what effect the AMP had on osseointegration of commercially available titanium implants in vivo, using a rabbit tibia model. The results showed a sustained release of AMP with equal or even better antibiofilm properties than the traditionally used antibiotic Cloxacillin. In addition, no negative effects on osseointegration in vivo was observed. These combined results demonstrate the potential of using mesoporous titania as an AMP delivery system and the potential use of the AMP RRP9W4N for infection control of osseointegrating implants. Despite that S. epidermidis has a low level of virulence and usually does not cause severe infections, they give persistent low-level infections that are difficult to treat. The virulence of S. epidermidis is connected to its ability to adhere to surfaces and form biofilms, 2 which can be up to 1,000 times more antibiotic resistant than their corresponding planktonic counterparts. 3 Due to their high antibiotic tolerance, alternatives to treat these infections must be found. A potential group of substances to eradicate biofilms are antimicrobial peptides, AMPs. They strike widely against bacteria, fungi, parasites and some viruses by destroying the cell membrane. The initial interaction between AMP and microbe is thought to occur via electrostatic interactions between the positively charged AMP and negatively charged groups on the bacterial membranes, for example, lipopolysaccharides and lipoteichoic acids. 4 This interaction is followed by rupturing of the bacterial membrane resulting in cell death. Although natural existing AMPs suffer from problems such as proteolytic degradation and cytotoxicity, they can be improved by peptide engineering. Then, proteolytically stable antimicrobial peptides with low toxicity for human cells and high bactericidal effect, even against multi-resistant bacterial strains of S. aureus, Group A streptococci, Escherichia coli and P. aeruginosa, can be obtained. 5 Engineered AMPs have also shown antibacterial properties against other bacteria like Enterococcus faecium, S. aureus, Klebsiella pneumoniae, Acinetobacter baumanii, Pseudomonas aeruginosa, Enterobacter cloacae and Escherichia coli. 6,7 In addition to engineering AMPs, their stability and function may also be improved by the mode of delivery. 8 As an alternative to administering antimicrobials systemically, local drug-delivery systems may be used when fighting biofilms, with benefits including high local concentrations without the risk of systemic toxic concentrations and an unaffected normal bacterial flora at uninfected areas. Some local strategies used are coatings preventing bacterial adhesion or coatings containing or releasing antimicrobials. 9 Local delivery systems of AMPs that have been evalauted include polymer functionalized carriers, 10 calcium phosphate carriers, 11 silica carriers 12 and cubosomes. 13 Titania surfaces can also be modified for local AMP delivery, using for example, covalent grafting of peptides 14 or nanotubes. 15,16 Mesoporous titania is a potential candicate for local delivery of AMPs, since it possess a good drug loading capacity with controlled relase kinetics, which can be altered by changing material properties such as pore size, surface area and surface chemistry. 17 For antimicrobial purposes, mesoporous titania coating has shown to be efficient in releasing antibiotics for elimination of S. aureus and P. aeruginosa. 17 In this study, we investigated the antibacterial effect of the PRELP-derived antimicrobial peptide RRPRPRPRPWWWW-NH2 (RRP9W4N) when incorporated into mesoporous titania. RRP9W4N has shown to have good bacterial killing properties combined with low toxicity to bone forming cells, human osteosarcoma MG63 cells, and human mesenchymal stem cells in vitro. 18 The loading and release of peptide from mesoporous titania was investigated as was the antibacterial effect, using a S. epidermidis biofilm safranine assay together with LIVE/DEAD BacLight staining and confocal laser scanning microscopy. The antimicrobial activity of the AMP was compared to the clinically used antibiotic Cloxacillin. Moreover, the effect of the AMP on the osseointegration process was evaluated in vivo using a rabbit tibia model. This was performed to evaluate if the procedure resulted in a selective implant, that is, high antimicrobial effect combined with no negative consequences on the osseointegration. In the in vivo study, commercially available dental titanium implants were installed and followed at different healing times.

| Formation of mesoporous titania films
Mesoporous titania films were formed on different substrates using the evaporation induced self-assembly method, EISA, as earlier described. 19 The block copolymer Pluronic 123 (Sigma Aldrich) was used as template and titanium(IV)tetraethoxide (Sigma Aldrich) was used as the inorganic precursor. The 0.5 g Pluronic 123 was dissolved in 8.5 g ethanol (99.5%, Solveco) and stirred for 2 hrs using a magnetic stirrer. In a separate vial, 2.1 g of titanium(IV)ethoxide and 1.6 g HCl (37%, Sigma Aldrich) were mixed and stirred for 2 hrs using a magnetic stirrer, before added to the Pluronic solution and stirred overnight.
The concentrations used were chosen to form a cubic mesoporous structure. 20 Mesoporous thin films were formed using spin-coating (Spin 150, SPS-Europe, 7,000 rpm, 60 s) on different substrates: cover glass slides (VWR), QCM-D sensors (Q-sense), titanium discs (Alfa Aesar) and commercially available threaded titanium implants with a diameter of 3.5 mm and a length of 7 mm (Neodent, Curitiba, Brazil).
After spin coating, the surfaces were left overnight in room temperature allowing for the self-assembly to complete and the ethanol to evaporate. The thin films were then calcined to remove the Pluronic template and increase titania cross-linking density through condensation. The samples were heated at a rate of 1 C/min to 350 C, and then left to dwell for 4 h before slowly cooled to room temperature.
Nonporous thin titania films were prepared using the same procedure, but without addition of Pluronic 123. The implants were immersed in 200 μM AMP for 24 h and air dried before sterilization by gamma-radiation. This procedure gave a film thickness of 200 nm. 19

| AMP loading and release from mesoporous titania
Titanium sensors (QSX 310, Q-Sense) were coated with mesoporous titania as described above. Non-coated titanium disks were used as reference. The sensors were washed as recommended by the manufacturer. First, sensors were treated in UV/Ozone for 20 min followed by cleaning in 2% sodium dodecyl sulphate (SDS) for 30 min before washing extensively with MilliQ water. Then samples were dried in nitrogen gas before being subjected to UV/Ozone for another 10 min.

| Biofilm formation
Two bacterial strains of Staphylococcus epidermidis were used, the fresh isolate Mia 21 and the type strain CCUG 39508 (Culture Collection University of Göteborg). The strains were cultured on Brain heart infusion agar plates at 37 C. Colonies were transferred to 5 ml liquid Todd Hewitt medium and incubated overnight (37 C) before being transferred to 100 ml fresh Todd Hewitt medium and incubated overnight. Then, bacteria were harvested, washed by centrifugation (2,500 rpm, 10 min) and resuspended in fresh Todd Hewitt medium.
Heat-sterilized mesoporous titania samples were submerged into a solution of 0.5 g/L Cloxacillin, pH 7.5 or 200 μM of the antimicrobial peptide RRPRPRPRPWWWW-NH2 (Biopeptide, 95%, RRP9W4N) and left to soak for 24 hrs to incorporate the antimicrobial substances into the mesoporous film.
Bacteria were cultured for 24 hrs (37 C) to allow biofilm adherence and then media was discarded, fresh media was added, and biofilm formation was allowed for another 24 hrs (37 C).

| Biofilm safranine assay
To assess the amount of biofilm formed on the mesoporous titania samples, surfaces of the samples after bacterial culture were stained with safranine. Each surface was rinsed in 2*0.5 ml MilliQ water and the biofilms were fixed with 0.5 ml methanol for 10 min. Surfaces were allowed to air dry before stained with 1 ml 0.1% safranine O (Alfa Aeser) for 10 min and then rinsed with 4*1 ml MilliQ water. The samples were air-dried and safranine was solubilized for 10 min in 95% ethanol. The solutions were collected, and absorbance was recorded with UV-Vis spectroscopy (HP8453, Hewlett Packard) at 532 nm.

| Biofilm viability
To assess the bacterial viability, biofilms were stained with LIVE/ DEAD BacLight (Molecular probes, kit 7007). Samples were washed twice in MilliQ water, and then stained by 4 μl of BacLight (2.5 μl of sample A, 0.5 μl of sample B in 1 ml 0.85% NaCl) and left in the dark for at least 15 min before studied with confocal microscopy using a LSM 700 inverted confocal laser scanning microscope (Zeiss). Light emitted below 555 nm was collected for propidium iodide and light emitted above 560 nm was collected from Syto 9. In each experiment, images covering a total of 1 mm 2 were obtained using the tile scan function and the surface area coverage of live and dead bacteria were measured using the software Volocity.

| In vivo study
To assess the effect of the antimicrobial peptide (RRP9W4N) on osseointegration and bone healing an in vivo study in rabbit tibia was performed. Screw shaped titanium implants coated with a thin mesoporous titania film were used as controls and the same implants loaded with AMP were used as tests.

| Statistics
Statistical analysis of the in vitro results was performed using one-way ANOVA or Kruskal-Wallis tests and p < 0.05 was considered statistically significant. Data are presented as mean ± SD.
In the in vivo study, the histomorphometrical results of test and control implants at the different healing times were compared using the non-parametric Wilcoxon Signed Rank test for depended samples and the test and control implants on the same animal were considered as a pair. The analysis was performed with SPSS software (IBM, Version 23.0). Data are presented as mean ± SD.

| Material characterization
The mesoporous titania was characterized using electron microscopy. In TEM it was clearly observed that the titania consisted of a continuous mesoporous network with pore-diameter of 6 nm, see micrograph in Figure 1(a). The surface of the mesoporous network was examined using SEM and pores present on its surface could be seen, Figure 1(b). This indicate the possibility of loading antimicrobial substances into the mesoporous titania.
The water wettability was examined using contact angle measurements and is was shown that the mesoporous titania surfaces were hydrophilic, as can be seen in Table 1. After loading the mesoporous network with either AMP or Cloxacillin, the contact angles changed, indicating the presence of antimicrobials on the surface.
To confirm the presence of antimicrobials, XPS was performed.
According to XPS, pure mesoporous titania has a surface composition of about 18% C, 59% O and 23% Ti, whereas the AMP loaded sample had a considerable amount of nitrogen (14% of the sample). Also, the cloxacillin loaded sample contained nitrogen, as well as chlorine, as expected from the compositions of the antimicrobial ( Table 2). In addition, some salts were also found on the AMP loaded sample.

| Drug loading and release
To investigate how much antimicrobials that were loaded into the mesoporous titania, QCM-D experiments were performed see Figure   2. When RRP9W4N was added to the mesoporous titania an immediate, substantial loading of the mesopores could be observed

| Biofilm formation
When using the safranine assay to investigate the biofilm content

| In vivo results
One rabbit allocated to 2 weeks of healing died during surgery, proba-  Biofilm area (μm 2 ) Syto PI * * * * F I G U R E 4 The mean (± SD) biofilm surface coverage was assessed using BacLight LIVE/DEAD staining and confocal microscopy. Using a Kruskal Wallis test statistical significance could be shown between MpTiO 2 compared to Cloxacillin and AMP (p < 0.01) but the difference did not reach a statistical level of significance (p = 0.05). The BA% values were comparable for the test and control implants (24.25% for the test group and 23.21 for the control group).
At week 4 of healing, the new bone had matured and formed a compact and lamellar bone in intimate contact with the implant surface and filling the threads in the cortical region of the tibia. Again, no qualitative differences were observed in the bone maturation between the test and the control groups. The histomorphometrical values were significantly increased from the 2-week observations and were comparable for the test and control samples (BIC%: 16.27% and 21.23% for the test and control group respectively; BA%: 64.60 and 60.50% for the test and control group respectively). Further bone maturation was observed after 12 weeks of healing and the implants were all surrounded by compact and lamellar bone, which extended also in the marrow region, encapsulating large areas of the implants. The average BIC% was 21.46% for the tests and 22.46% for the controls, while the mean BA% was 54.66% for the tests and 52.98% for the controls. The histomorphometrical parameters are reported in Table 3 and Figure 6.

| DISCUSSION
In this study, we wanted to investigate the antimicrobial behavior of an engineered AMP, RRP9W4N, in addition to investigating its effect on osseointegration, using commercially available implants coated with AMP loaded mesoporous titania. This was accomplished using both in vitro and in vivo studies, showing a greatly reduced in vitro biofilm formation on mesoporous titania loaded with the AMP and no negative effects on osseointegration in vivo.
The rabbit in vivo study showed that release of the AMP from the mesoporous titania coated implants did not cause any sign of cellular or tissue toxicity in the proximity of the implants and did not interfere with bone healing and osseointegration of the implants. This is in agreement with our earlier in vitro results, showing this AMP, when covalently surface-immobilized, have no toxic effects on mammalian cells. 18 In native form however, AMPs may be toxic to eukaryotic cells, although cytotoxicity often decreases in plasma due to binding of plasma proteins. 23 Due to their endogenous origin, AMPs also F I G U R E 5 Representative toluidine blue stained histological images of the control (no AMP) and test (with AMP) implants at 2, 4 and 12 weeks of healing. Red scale bar: 1000 μm. Yellow scale bar: 100 μm  24 and by varying the end-tag length proteolytic stability can be achieved. 5 The AMP used in this study, RRP9W4N, has been engineered according to these measures and the in vivo results showed the AMP-releasing implants performed in a manner comparable to the control implants with identical surface, but without addition of AMP. In addition, the BIC% was more than twice as high for the test implants than for the control implants after 2 weeks of healing.
Despite the difference was not statistically significant, this AMP might have a mild enhancing effect in initial osseointegration. Mild, non-significant, enhancing osseointegrative effects have also been shown for other AMPs. 11 However, more studies need to be performed to make the statement that these AMPs improves the early stages of osseointegration.
In addition to chemically engineered AMPs, their stability and function may also be improved by mode of delivery. 8 For example, elastases from Pseudomonas aeruginosa and human neutrophils degrades pure AMP LL-37, but when loaded in cubosomes it can be protected against proteolytic degradation. 13 In this study, mesoporous titania coatings were used and electron microscopy showed a porous network extending to the surface, into which antimicrobials were loaded. The cubic mesoporous network had pore diameters of 6 nm, well above the size of both antimicrobials used, Cloxacillin and RRP9W4N. As has been previously shown using different sized dendrimers, the absorption into mesoporous films is dependent on both pore size and morphology (cubic or hexagonal) where a cubic arrangement results in an increased absorption rate. 25 The cubic mesoporous titania used in the present study clearly facilitate AMP loading into the pores; however, the pore size is small enough to prevent entry of most proteolytic enzymes and other proteins, and thus the mesoporous titania may not only act as a drug carrier but also provide some means of physical protection against proteolytic degradation and other protein interactions that may interfere with the antimicrobial activity. For example, mesoporous silica have shown to prevent proteolytic degradation of a green fluorescent protein. 26 In the in vitro studies, rich growth medium was used, meaning a plethora of proteins (although denatured in the autoclave) were present during both culturing and bacterial elimination. Despite this, the AMP was as good a biofilm eliminator as Cloxacillin, indicating the AMP ability to maintain its function despite potential protein binding. This is probably due to its engineered properties to increase its proteolytic stability and the potential of physical protection of the AMPs inside the mesopores. should also be emphasized that the drug-release in vitro is by no means the same as in the in vivo situation, and it is challenging to make direct comparisons. We do not claim that the release kinetics observed here can be directly translated to a clinical situation, but still it gives a good indication that the release is sustained and not delivered in a burst fashion, at least during the first day. In addition to functioning as a drug-delivery system, pure mesoporous titania coatings on implants have also shown a tendency to improve osseointegration in vivo compared to non-porous titania. 19 The major function of AMP is to eliminate microorganisms, and both in vitro experiments in this study showed a significantly smaller biofilm on the AMP charged mesoporous surface compared to the non-loaded mesoporous control. The antimicrobial effect of the AMP was comparable or better than what was found for the antibiotic Cloxacillin. Other AMPs, both natural and synthetic, have also shown considerable antibiofilm properties. 28,29 We did not perform any in vivo antimicrobial tests, but this AMP has shown an antibacterial effects in human infected blood and in a pig skin wound model ex vivo, 5 indicating its potential use in future patient implants. Studying the small proportion of bacteria that were able to adhere to the AMP charged surface, a majority (86-99%) were alive. One hypothesis is that bacteria may adhere to the surface, but a slow, sustained release of AMP efficiently eliminate bacteria in due time and prevent formation of a mature biofilm. We have earlier shown this AMP to initiate swelling and bursting of bacteria on surfaces and complete bacterial eradication, depending on time and concentration. 30 The results of this study show the combined good in vitro bacterial elimination, even in presence of a high initial bacterial load, far larger than what would be found in the clinical setting with no negative effects on in vivo osseointegration. We have earlier shown this AMP to be efficient in killing both normal and persister bacteria of S. epidermidis, at a concentration of 100-200 μM in vitro. 30 This is an important factor in the clinical setting where recurring infections caused by persister cells is a challenging problem. 31 Although the AMP concentration in the mesoporous titania of this study is less, it may be modulated to create an efficient bacterial persister elimination and sustained release of AMP.

| CONCLUSION
In this study, the antibacterial effect of the engineered AMP, RRP9W4N on bacterial biofilms, and its influence on osseointegration directly at the implant healing site were examined. The peptide was loaded in mesoporous titania, which was coated into titanium implants. The anti-bacterial property of the AMP was preserved, as demonstrated by the in vitro evaluation, while it did not negatively affect osseointegration in vivo. The potential benefits of introducing antimicrobial substances into mesoporous titania on implant surfaces is local delivery, facilitating high antimicrobial substance concentration and avoiding systemic side effects. Here, we show the potential of mesoporous titania to deliver antimicrobial peptides in a proteinaceous environment. More than 90% of the biofilms were eliminated in vitro, a result equal to that of the clinically used antibiotic Cloxacillin, at the same time as ossoeintegration proceeded without any negative observations. The combined result of this study suggests that AMP loaded mesoporous titania may be a good candidate to lower the risk of implant associated infections.