Antibacterial Surface Design of Titanium-Based Biomaterials ...Antibacterial Surface Design of...

Antibacterial Surface Design of Titanium-Based Biomaterials forEnhanced Bacteria-Killing and Cell-Assisting Functions AgainstPeriprosthetic Joint InfectionJiaxing Wang,†,∇ Jinhua Li,‡,⊥,∇ Shi Qian,‡,∇ Geyong Guo,† Qiaojie Wang,† Jin Tang,∥ Hao Shen,*,†

Xuanyong Liu,*,‡ Xianlong Zhang,*,† and Paul K. Chu§

†Department of Orthopaedics, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai Jiao Tong University,Shanghai 200233, China‡State Key Laboratory of High Performance Ceramics and Superfine Microstructure, Shanghai Institute of Ceramics, ChineseAcademy of Sciences, Shanghai 200050, China§Department of Physics and Materials Science, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong, China∥Department of Clinical Laboratory, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai Jiao Tong University,Shanghai 200233, China⊥University of Chinese Academy of Sciences, Beijing 100049, China

*S Supporting Information

ABSTRACT: Periprosthetic joint infection (PJI) is one of the formidable andrecalcitrant complications after orthopedic surgery, and inhibiting biofilmformation on the implant surface is considered crucial to prophylaxis of PJI.However, it has recently been demonstrated that free-floating biofilm-likeaggregates in the local body fluid and bacterial colonization on the implant andperi-implant tissues can coexist and are involved in the pathogenesis of PJI. Aneffective surface with both contact-killing and release-killing antimicrobialcapabilities can potentially abate these concerns and minimize PJI caused byadherent/planktonic bacteria. Herein, Ag nanoparticles (NPs) are embedded intitania (TiO2) nanotubes by anodic oxidation and plasma immersion ionimplantation (PIII) to form a contact-killing surface. Vancomycin is thenincorporated into the nanotubes by vacuum extraction and lyophilization toproduce the release-killing effect. A novel clinical PJI model system involvingboth in vitro and in vivo use of methicillin-resistant Staphylococcus aureus(MRSA) ST239 is established to systematically evaluate the antibacterialproperties of the hybrid surface against planktonic and sessile bacteria. Thevancomycin-loaded and Ag-implanted TiO2 nanotubular surface exhibitsexcellent antimicrobial and antibiofilm effects against planktonic/adherent bacteria without appreciable silver ion release. Thefibroblasts/bacteria cocultures reveal that the surface can help fibroblasts to combat bacteria. We first utilize the nanoarchitectureof implant surface as a bridge between the inorganic bactericide (Ag NPs) and organic antibacterial agent (vancomycin) toachieve total victory in the battle of PJI. The combination of contact-killing and release-killing together with cell-assistingfunction also provides a novel and effective strategy to mitigate bacterial infection and biofilm formation on biomaterials and haslarge potential in orthopedic applications.

KEYWORDS: antibiotics, silver nanoparticles, titania nanotubes, cells/bacteria coculturing, antimicrobial properties

1. INTRODUCTION

As the demand for total joint arthroplasty (TJA) is increasingdue to the aging population, periprosthetic joint infection (PJI)is becoming a formidable and recalcitrant complicationrequiring costly treatment, multiple operations, and extendedhospitalization.1−3 PJI can be categorized as early (less than 3months after surgery), delayed (3 to 24 months), or late (morethan 2 years).1,4 Early PJIs are generally caused by high virulentbacteria such as Staphylococcus aureus (S. aureus), especiallymethicillin-resistant S. aureus (MRSA).1,5,6 MRSA-PJI involves

the interactions among bacteria, biomaterials, and tissues, andthe formation of a biofilm composed of bacteria andextracellular matrix plays a vital role in the pathogenesis ofPJI.7,8 After a mature biofilm has developed, bacteria can leavethe biofilm penetrating the biomaterials surface into surround-ing fluids and tissues.7 Although radical debridement, empirical

Received: March 6, 2016Accepted: April 7, 2016Published: April 7, 2016

Research Article

www.acsami.org

© 2016 American Chemical Society 11162 DOI: 10.1021/acsami.6b02803ACS Appl. Mater. Interfaces 2016, 8, 11162−11178

www.acsami.org

http://dx.doi.org/10.1021/acsami.6b02803

antibiotics, and prosthesis retention have been utilized, MRSA-PJI in the early stage is difficult to treat and can affect theprognosis.9−11 Therefore, it is critical to develop practicalprophylactic strategies to mitigate early MRSA-PJI.Antimicrobial coatings are useful in early MRSA-PJI

prevention, and anti-infective surfaces can be primarilyclassified into antimicrobial-agent-releasing or contact-killingones,12−15 and the relevant animal models have mainlyconcentrated on implant and bone infection to mimic thepathogenesis of implant-related osteomyelitis.16,17 Nevertheless,early PJI caused by hospital-associated (HA) pathogens occursin the special intra-articular internal environment that iscompletely different with osteomyelitis in the bone medullarycavity.13 Furthermore, the impaired soft tissue surrounding theprosthesis makes the joint susceptible to exogenous infectionduring perioperation.18 Bacteria can be cultured from peri-prosthetic soft tissues in PJI patients,19,20 and it is essential toestablish an effective animal model for early MRSA-PJI tosystematically evaluate the intra-articular antibacterial effects ofthe coatings. The properties of antimicrobial biomaterials areusually assessed by single cultures with tissue cells or bacteria invitro. However, considering that about 30% of orthopedicsurgeries are in a bacterially contaminated state21 and early PJIsare generally induced during the insertion of prostheses,1

coculturing with both tissue cells and bacteria needs to beperformed on anti-infective coatings to better mimic thephysiological conditions.Silver nanoparticles (Ag NPs) that are immobilized on

titanium by plasma immersion ion implantation (PIII) producelong-lasting in vitro and in vivo antibiofilm activity againstsurface-adhered bacteria without producing drug-resistantbacteria and cytotoxicity.16,22,23 Nonetheless, MRSA hasrecently been found to form strong free-floating biofilm-likeaggregates in human synovial fluid (SF),24 and colonization ofS. aureus in peri-implant tissues may also be a generalphenomenon in implant infection.8,14,25 Consequently, a simplecontact-killing surface may have limited or no effects on free-floating biofilm-like aggregates or bacterial colonization in peri-implant tissues,16,26 and such antiadhesive strategy may notsatisfy clinical needs. Hence, surface functionalization shouldcombine contact-killing and release-killing activities forprophylaxis of early MRSA-PJI. Anodically oxidized titaniananotubes (TiO2−NT) have been used as the reservoir to loadantibacterial agents to provide short-term antimicrobial effectsdue to the large specific surface area, “half-open” nanostruc-tures, and excellent biocompatibility.27,28 Traditionally, owingto the lack of better alternatives, vancomycin with powerfulactivity against all staphylococci is the main treatment forMRSA infection.29 It can penetrate biofilms and decrease thenumber of bacteria.14,30 Meanwhile, Ag NPs are also the mosthopeful candidate for controlling bacterial infection due to theirbroad antimicrobial spectrum,31 and synergetic combination ofthem with antibiotics by means of the nanotechnology may be anew and powerful approach to overcome bacterial infection.

In this work, by considering the complex pathogenic processof MRSA-PJI, we design and prepare a sophisticated anti-infective coating on titanium. Ag NPs are embedded in TiO2nanotubes by PIII to attain the long term contact-killing abilityand vancomycin is loaded into the TiO2 nanotubes by vacuumextraction to provide short-term release-killing. The antibacte-rial properties against planktonic and sessile clinical epidemicMRSA are determined in vitro and the biological effects withfibroblasts and MRSA simultaneously on the surface are studiedby cocultruing in vitro. Furthermore, the antibiofilm capacity,anti-infection efficiency, and cellular biological effects in theperi-implant soft-tissues are evaluated by the intra-articularrabbit PJI model in vivo.

2. MATERIALS AND METHODS2.1. Sample Preparation and Characterization. The specimens

used in the in vitro experiments were commercial pure Ti plates (10mm × 10 mm × 1.0 mm), and in the in vivo experiments, the sampleswere Ti screws (3.5 mm in diameter and 11 mm long) with Ti washers(3.5 mm inner diameter, 10.5 mm external diameter, and 1.0 mmthick). The Ti samples were ultrasonically pretreated in an acidicsolution (HNO3:HF = 5:1 v/v) at room temperature for 10 min,ultrasonically cleaned in ethanol and deionized water several times,and dried in air. The TiO2 nanotubes were prepared by anodicoxidation of titanium in an electrolyte with 0.5 vol % hydrofluoric acidat 20 V for 30 min.32 Afterward, by performing filtered cathodic arcPIII, Ag NPs were implanted and immobilized in situ in the TiO2nanotubes. The important processing parameters are listed in Table 1.All the specimens were thoroughly cleaned in water and sterilized in75%v/v ethanol.

Vancomycin (Sigma-Aldrich, St Louis, MO, USA) was introducedto the TiO2 nanotubes by lyophilization and vacuum drying27,33 with avancomycin solution (100 mg/mL) in phosphate-buffered saline(PBS; Sigma).33 After cleaning with deionized water and drying, 20 μLof the vancomycin solution were pipetted onto the specimen surfaceand spread evenly by gentle shaking. The samples were freeze-driedunder vacuum at −75 °C for 2 h and rinsed quickly with 1 mL of PBSto remove superfluous vancomycin. The flushed fluid was collected forsubsequent analysis. Four groups were tested: TiO2 nanotubes(control, denoted as NT), Ag PIII TiO2 nanotubes (denoted as NT-Ag), vancomycin-loaded TiO2 nanotubes (denoted as NT-V), andvancomycin-loaded NT-Ag (designated as NT-Ag-V).

The surface morphology and chemistry were investigated by field-emission scanning electron microscopy (FESEM; Magellan 400, FEI,USA) equipped with energy-dispersive X-ray spectroscopy (EDS). Thecrystallinity was studied by X-ray diffraction (XRD; Ultima IV, Rigaku,Japan) with Cu Kα (λ = 1.541 Å) at a glancing angle of 1°. The phasewas identified by referencing to the standard JCPDS database. TheRaman spectra were acquired on a Raman microprobe (LabRAM,Horiba Jobin Yvon, France) equipped with an Ar ion laser at 20 mW(514.5 nm). X-ray photoelectron spectroscopy (XPS; PHI 5802,Physical Electronics Inc., Eden Prairie, MN) was employed todetermine and field-emission transmission electron microscopy(TEM; JEM-2100F, JEOL Ltd., Japan) was performed at 200 kV.Materials were scrapped from the samples and dispersed in ethanolultrasonically, and a droplet of the suspension was placed on a holeycopper grid covered with porous carbon film.

Table 1. Important Parameters in Anodic Oxidation and Plasma Immersion Ion Implantation (PIII)

anodic oxidation silver plasma immersion ion implantation (Ag PIII)

sample name voltage (V) treatment time (min) pulse duration (μs) pulse frequency (Hz) pressure (Pa) treatment time (min) voltage (kV)

NT 20 30NT-Ag 20 30 450 6 2.5 × 10−3 90 −30NT-V 20 30NT-Ag-V 20 30 450 6 2.5 × 10−3 90 −30

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b02803ACS Appl. Mater. Interfaces 2016, 8, 11162−11178

11163


2.2. Bacterial Isolate and Characterization. One epidemicclinical isolate (MRSA, ST239) originated from the inpatient withimplant infection at The Orthopedics Department of Shanghai JiaoTong University Affiliated Sixth People’s Hospital, the largestorthopedics trauma clinical medical center in eastern China. Ourstudy was approved by Shanghai Jiao Tong University Affiliated SixthPeople’s Hospital Ethics Committee (Ethical code, 2014-63).2.3. Ag and Vancomycin Release Tests. The NT-Ag and NT-

Ag-V specimens were soaked in 10 mL of phosphate buffered saline(PBS) or trypticase soy broth (TSB) for 1, 4, 7, 14, 21, and 28 days at37 °C without agitation. The leachates were collected at each timepoint for quantitative analysis by inductively coupled plasma opticalemission spectrometry (ICP-OES, Nu Instruments, Wrexham, U.K.).The protocol to determine the release of vancomycin from the NTs

followed previous studies.27,33 In brief, the NT-V and NT-Ag-Vsamples were individually immersed in 1 mL of PBS on a 24-well plateat room temperature and agitated at 70 rpm. After the specific timeintervals, a 500 μL of the sample were taken to determine the releasekinetics, and then 500 μL of fresh PBS were added to the remainingsolution each time. The extracted solutions were collected periodicallyfor up to 4 h and analyzed for vancomycin concentration by ultravioletspectrophotometry (Inesa Analytical Instrument Co., LTD, Shanghai,China) at 237 nm.33 A standard curve obtained with knownvancomycin concentrations was adopted to determine the vancomycinconcentrations in the PBS.2.4. In Vitro Cytotoxicity Evaluation. The fibroblast-like cell line

(HT1080, Shanghai Institute of Biological Science, Chinese Academyof Sciences, China) was employed to evaluate the cytotoxicity of thespecimens, and the cytotoxicity was examined by the Cell CountingKit-8 assay (CCK-8, Beyotime Bio-Tech, China). First, the fourdifferent specimens were placed on 24-well plates. Cells with a densityof 2 × 104/well and 1 mL of the growth medium composed of DMEM(Dulbecco’s modified Eagle’s medium, Gibco) supplemented with10% fetal bovine serum (FBS, Gibco) were added to the wellscontaining the samples. They were cultured for 1 and 4 days in ahumidified atmosphere of 5% CO2 at 37 °C. After each period, theculture medium was completely removed, and 1.0 mL of fresh mediumwith 10% CCK-8 was added to each well and incubated for 4 h.Finally, 100 μL of the culture medium were transferred to a 96 wellplate for detection. The absorbance was measured on an enzymelabeling instrument (BIO-TEK, ELX 800) at 450 nm.2.5. In Vitro Antibacterial Assay. In the in vitro antibacterial

tests, the clinical epidemic MRSA (ST239) was cultivated in TSBmedium overnight at 37 °C and the concentration of inocula wasadjusted to 106 colony forming units (CFU)/mL through McFarlandin TSB supplemented with 1% (w/v) glucose (TSBG) or 1% TSBGsupplemented with synovial fluid obtained from outpatients (TSBG+SF). One mL of the bacterial suspension was added to the 24-wellplates on which each foil specimen was placed and statically incubatedfor 1, 6, 12, and 24 h at 37 °C. At a given time point, the specimenswere removed from the suspension and nonadherent bacteria on thesurface were gently dislodged with PBS three times. The viable countsof planktonic bacteria in the original culture medium were determinedby serial dilution and the spread-plate method (SPM).16 The numberof CFU on the sheep blood agar (SBA) was counted according to theNational Standard of China GB/T 4789.2 protocol. In addition, eachspecimen was put into 1 mL of sterile PBS followed by ultrasonicvibration (150 W, B3500S-MT, Branson Ultrasonics Co., Shanghai,China) for 10 min to dislodge the adhered bacteria and rapid vortexmixing (Vortex Genie 2, Scientific Industries, Bohemia, NY) for 1 min.The suspension was taken and living bacteria were counted by SPM.Each test was repeated three times.The antibacterial rates (Ra) for planktonic and adherent bacteria

were calculated based on the following formulas: (1) the antibacterialratio of planktonic bacteria in the culture medium (Rap)(%) = (A-B)/A × 100%, and (2) the antibacterial ratio of adherent bacteria on thesample (Raa)(%) = (C−D)/C × 100%, where B is the mean numberof viable bacteria in the culture medium inoculated with NT, NT-Ag,NT-V, or NT-Ag-V, A is the mean number of viable bacteria in theculture medium inoculated with NT sample, D is the mean number of

viable bacteria attached on the surface of NT, NT-Ag, NT-V, or NT-Ag-V, and C is the mean number of viable bacteria attached on thesurface of NT. The number of planktonic bacteria was also estimatedby improved SPM. 100 μL of the bacteria suspension from thebacteria-specimen coculturing system was uniformly spread on theSBA and incubated at 37 °C overnight before typical pictures weretaken.

2.6. Evaluation of in Vitro Bacterial Biofilm by SEM andCLSM. To assess the antiadhesive effect, the biofilms on the foils werequalitatively examined by scanning electron microscopy (SEM; JEOLJSM-6310LV, Japan) and confocal laser scanning microscopy (CLSM510 meta; Zeiss, Germany). The samples with adherent bacteria werefixed with 2.5% glutaraldehyde overnight at 4 °C and dehydrated witha series of gradient alcohol (50, 70, 80, 90, 95, and 100 v/v %). Afterdehydration, the specimens were critical-point dried, coated with goldin a sputter coater, and examined by SEM. In the CLSM test, thespecimens were placed on a 24-well plate, stained with 1 mL ofcombined dye (LIVE/DEAD BacLight bacteria viability kits,Invitrogen) for 20 min, and imaged by CLSM. The living bacteriawith intact membranes were stained green whereas dead ones withimpaired membranes were stained red.

2.7. Longevity of the Antiadhesive Ability on NT-Ag. In orderto evaluate the long-term antiadhesive effect of Ag−PIII, NT and NT-Ag were incubated separately in sterile PBS for 28 days. At given timepoints (1, 7, 14, 21, and 28 d), 1 mL of the bacteria suspension (106

CFU/mL) was put on the specimens overnight at 37 °C. In this assay,each Raa value of the adhered bacteria on the surface was calculated asaforementioned, and the procedure was repeated three times.

2.8. Bacteria-Tissue Cells Cocultures. According to the previousreport,34 in vitro coculture was carried out with MRSA (ST239) andfibroblast cells on the specimens. Twenty μL of the bacterialsuspensions (104 CFU/mL) were distributed on the samples andincubated at 37 °C for 2h. Afterward, fibroblast cells suspended in themodified culture medium which included the cell growth medium and2% (v/v) TSB were seeded on the bacteria-contaminated samples witha density of 2 × 104 cells/well on a new 24-well plate. MRSA (ST239)and fibroblast cells were cocultured in a humidified 5% CO2 at 37 °Cfor 3 days. Finally, the fibroblast cells were fixed in 4%paraformaldehyde, imaged by SEM after serial dehydration, dried,gold coated, and examined by fluorescence microscopy (LeicaMicrosystems LTD, Germany) after staining with TRITC-phalloidinand DAPI. The quantitative results were expressed as the surfacecoverage by fibroblast cells in the presence of adherent bacteria andanalyzed according to the aforementioned reference. All the tests wereperformed in triplicate.

2.9. Animal Surgery. 2.9.1. New Rabbit PJI Model. The animalexperimental protocol was approved by the Animal Care and EthicsCommittee of Sixth Peoples Hospital affiliated to Shanghai Jiao TongUniversity, School of Medicine. 32 New Zealand White rabbitsweighing an average of 3.2 kg (2.5−4.0 kg) were used. The rabbit PJImodel was described previously35,36 and Zhai et al.35 found that 1 mLof inoculum with 104 CFUs S. epidermidis injected into the knee jointformed a stable PJI model. Considering that the virulence of MRSAsurpassed that of the S. epidermidis, 104 CFUs could be the appropriatedose to make the improved PJI model. The rabbits were randomlydivided into 4 groups (8 in each group, NT, NT-Ag, NT-V, and NT-Ag-V).

Surgery was performed under general anesthesia with 3%pentobarbital (1 mL/kg). After shaving the hairs around left knee,the lower extremity was sterilized and draped with sterile surgicaldrapes. A 3 cm longitudinal skin incision was made over the lateralaspect of the left knee. Arthrotomy was implemented fully exposingthe lateral femoral condyle and lateral collateral ligament (LCL). Witha 3.5 mm sterile drill bit, a hole parallel to the coronal axis was drilledin the lateral femoral condyle anterior to the LCL. A metal screw andmetal washer were implanted into the knee joint. The joint capsule,deep fascia, and skin were closed by 4−0/3−0 monofilament nylonsuture. After surgery, 100 μL of inoculum with 104 CFUs of MRSAwere injected into the knee joint near the washer. Every animalreceived acetaminophen 30 mg/kg/day for analgesia for 3 days after



11164


operation. After surgery, the body temperature and weight of eachrabbit were recorded and the rabbits were not found to have fever orweight loss.2.9.2. Gross Observation and Radiographic Evaluation of

Inflammatory Reaction. Fifteen days after surgery, high-resolutionlateral radiographs were acquired when the rabbits were underanesthesia with 3% pentobarbital. In the following days, the rabbitswere euthanized with an overdose of pentobarbital and the skin of theknee joint and joint capsule were successively sectioned under sterileconditions. The inflammatory condition was graded on an improvedfour point scale.36 The grade one articular cavity included no purulenteffusion and necrotic tissue. The synovial fluid was clear and the screwtail and washer were covered by synovial tissue. The grade twoarticular cavity showed mild inflammation with little necrotic tissueand no obvious pus. The synovial fluid was slightly muddy and thescrew tail and washer were covered by synovial tissue. The grade threearticular cavity exhibited moderate inflammation with necrotic tissueand more purulence. The synovial fluid was turbid and the screw tailand washer were covered by infectious synovial membrane andnecrotic tissue. The grade four articular cavity showed severeinflammation with white sticky pus and a large number of necrotictissue. The joint fluid was not visible and joint capsule was full ofpurulence. The screw tail and washer were covered by white stickytissue. The gross appearance of each treated knee was given a score asfollows: 0 = grade one, 1 = grade two, 2 = grade three, and 3 = gradefour. Three independent observers assessed the inflammatory reactionand calculated the corresponding score in a blind fashion.2.9.3. CFU Counting of Metal Washer and Surrounding Tissues In

Vivo. After harvesting from the knee joint, the washers were washedgently three times with normal saline (NS) to dislodge the planktonicbacteria and then immersed in tubes containing 1 mL of NS. Thesamples were sonicated at 150 kHz for 15 min and vortexed rapidly for1 min in order to break the biofilm aggregates and detach the adheredbacteria. The peri-prosthetic soft tissues were weighed andhomogenized using a high-speed homogenizer (Jingxin IndustrialLimited Company, Shanghai, China) in tubes with 1 mL of NS. Thesuspensions and homogenates were serially diluted from10−1 to10−4 inNS and 100 μL of the diluent were spread on SBA in duplicates. Theplates were incubated overnight at 37 °C and the viable bacterialcolonies were counted as mean ± standard deviation (S.D.) andexpressed by log10CFU/per metal washer and log10CFU/g of peri-implant soft tissue.2.9.4. SEM of Metal Washer In Vivo. According to previous

protocols,35,37 the retrieved washers were rinsed three times with PBSand preserved in 2.5% glutaraldehyde for 1 day. After dehydration,drying, and coating with gold−palladium, the metal washers wereexamined by SEM (JEOL JSM-6310LV, Japan). The bacterialexperiment and SEM analysis were conducted in a blind fashion(operators not knowing the sample history).2.9.5. Histological Analysis. The experimental knee joints were

isolated and peri-implant soft tissues were dissected. The tissues werefixed in 10% neutral buffered formalin for 72 h, dehydrated in serialalcohol solutions, and infiltrated with xylene and paraffin beforeparaffin embedding. Sagittal sections of the soft tissues were cut intosections (approximately 5 μm thickness) by a sledge microtome(Leica, Hamburg, Germany), deparaffinised in xylene, and stained withMayer’s Hematoxylin and Eosin (H&E) and May-Grunewald Giemsafor histological evaluation using an optical microscope.2.10. Statistical Analysis. Each experiment was repeated three

times, and the results were expressed as means ± standard deviations.The one-way analysis of variance (ANOVA) was combined with aStudent−Newman−Keuls (SNK) post hoc test to determine thestatistical significance. The differences in the numbers of CFU wereanalyzed by the two-tailed Mann−Whitney rank sum test. In all thetests, P values of <0.05 or <0.01 were considered statisticallysignificant.

3. RESULTS3.1. Sample Preparation and Characterization. In a

typical process as shown in Figure 1, the precleaned Ti

underwent anodic oxidation leading to in situ growth of theTiO2 nanotube arrays on the surface. After PIII, Agnanoparticles were immobilized in the TiO2 nanotubes andthe samples were subsequently loaded with vancomycin byvacuum extraction. As shown in Figure 2a-b, a homogeneousnanoporous topography composed of sparsely distributed TiO2nanotubes appears on the Ti surface after anodic oxidation,with a tube diameter of less than 100 nm and tube spacing ofless than 50 nm from each other. This type of nanotube arraysis different from the closely distributed nanotubes because theyhave a larger specific surface area boding well for loading ofantibiotics. The inset in Figure 2b shows the typicalmorphology of an individual TiO2 nanotube. Figure 2c exhibitsthe secondary electron (SE) image of the Ag PIII TiO2nanotubes (NT-Ag). The pristine nanotube structure issputtered to some extent during Ag PIII and no significantAg nanoparticles are observed from the nanotube surface,although there are a few individual ones (indicated by redarrow). The backscattered electron (BSE) image acquired fromthe same region is displayed in Figure 2d.Tiny Ag nanoparticlesare observed on the nanotube surface (white dots) indicated bynot only the red arrow but also the green one. The imagebefore Ag nanoparticles “breaking through the soil” is captured.Here, a vacuum extraction method is implemented to loadvancomycin into the TiO2 nanotubes with/without Ag PIII.With regard to the NT-V sample and as shown in Figure 2e, theTiO2 nanotubes are covered/loaded with a homogeneous layerof vancomycin showing a topography different from that of theTiO2 nanotubes. This is further evidenced by the high-magnification image of the vancomycin-loaded TiO2 nanotubesin Figure 2f, in which the purple arrow shows that the tubespace is filled with vancomycin. This is corroborated by thehigher magnification inset. A layer of vancomycin is alsovacuum-extracted onto the inner wall of the TiO2 nanotubes(indicated by purple arrows). With regard to the vancomycin-loaded Ag-implanted TiO2 nanotubes (i.e., NT-Ag-V), most ofthe Ag nanoparticles are not on the nanotubes surface in the SEimage, despite the presence of a few individual exposednanoparticles (red arrow). The corresponding BSE image

Figure 1. Schematic illustrating the versatile surface design by thethree-step approach: (i) Anodic oxidation to prepare TiO2 nanotubesarray, (ii) plasma immersion ion implantation to embed Agnanoparticles into TiO2 nanotubes, and (iii) vacuum extraction toload vancomycin into the TiO2 nanotubes.



11165


shows these tiny Ag nanoparticles (green arrow) which areconfirmed by the high-magnification images in the correspond-ing insets. Meanwhile, the TiO2 nanotubes are covered by ahomogeneous layer of vancomycin, even on the tube inner wall(purple arrow).The TiO2 nanotubes are split to investigate the cross

sections. Figure 3a shows the cross-section of the pristine TiO2nanotubes showing a bamboo-like morphology. Microareaelemental analysis is conducted by EDS (indicated by yellowarrow) and only Ti, O, and C are detected (Figure 3b,c). Thecross-sectional SE image of NT-Ag is displayed in Figure 3d.Combining the corresponding BSE image (Figure 3e), there areboth embedded Ag nanoparticles (green arrow) and exposedAg nanoparticles (red arrow) in/on the TiO2 nanotubes. EDSshows the presence of Ag (Figure 3c,f). Concerning sampleNT-V, the obvious topography contrast of the tube inner wall(Figure 3g) implies successful loading of vancomycin. Theelemental contents after the vancomycin treatment increase (C,3.48 at. % to 13.40 at. %; O, 36.95 at. % to 45.90 at. %; Cl, 0 to0.32 at. %), accompanied by the corresponding decrease in theTi concentrations from 59.57 at% to 39.45 at%. With regard toNT-Ag-V, the vacuum extraction process enables vancomycinto flow into the TiO2 nanotubes (indicated by purple arrow inFigure 3j). The inner wall of the tube is coated with a layer ofvancomycin (indicated by yellow arrow in Figure 3k). Based on

Figure 3i,l, there is no appreciable difference in the amounts ofvancomycin difference between NT-V and NT-Ag-V.Figure 4a shows the XRD patterns of samples NT, NT-Ag,

NT-V, and NT-Ag-V. A layer with the amorphous TiO2nanotubes array on the Ti surface is indicated by the diffractionpeaks assigned to anatase and rutile phases and Ag PIII doesnot alter the phase structure. After vacuum extraction to loadvancomycin, new diffraction peaks emerge from the XRDpatterns of NT-V and NT-Ag-V and can be indexed to thevancomycin phase in the TiO2 nanotubes. As shown in Figure4b, the Raman bands at 150 and 612 cm−1 of the insetcorrespond to the surface anatase and rutile phases,respectively. After vancomycin introduction, new Ramanbands appear from the spectra of NT-V and NT-Ag-V. TheRaman bands at 876 and 986 cm−1 are assigned to the C−Csymmetrical stretch and C−C antisymmetrical stretch,respectively. The 1125 cm−1 Raman band is associated withthe CH2 symmetrical twist and that at 1236 cm−1 correspondsto C−N bending. The 1606 cm−1 Raman band originates fromstretching of aromatic rings.38,39 All these Raman bands arecharacteristic of vancomycin indicating successful incorporationof vancomycin into TiO2 nanotubes. In comparison, thevancomycin molecules show stronger Raman bands on NT-Agthan NT, with a Raman intensity ratio of around 4 due tosurface-enhanced Raman scattering (SERS) from the Agnanoparticles.40 XPS is carried out to investigate the elementalchemical states. The Ti 2p XPS spectrum shows two peaks at464.6 and 458.8 eV (Figure 4c) characteristic of Ti 2p1/2 and Ti2p3/2 in TiO2, respectively. Figure 4d shows the Ag 3d doubletsat 368.3 eV (Ag 3d5/2) and 374.3 eV (Ag 3d3/2) with a spinenergy separation of 6.0 eV stemming from metallic silver(Ag0). After vancomycin introduction, the C 1s XPS spectrumshows three peaks at 288.4, 286.6, and 285.0 eV, which areassigned to imide C, amide − C, and C−C/C-H, respectively,as shown in Figure 4e-f. Figure 4g-h show that N 1s containstwo subpeaks at 401.0 and 400.0 eV corresponding N−C3 andN−C/N-CO, respectively.41 Figure 4i shows the schematicillustration of loading and interactions of vancomycin moleculesin the Ag nanoparticles modified TiO2 nanotubes. Based onthese characterization results, a layer containing TiO2 nano-tubes, Ag nanoparticles, and vancomycin is fabricated.

3.2. Drug Loading/Release and Cytotoxicity Evalua-tion. The loading efficiency of vancomycin in NT-V and NT-Ag-V is evaluated according to a previous method.27 After aninitial wash, more than 80% of vancomycin is retained in thetwo groups of NTs as shown in Figure 5a. Release ofvancomycin from the two kinds of NTs is investigated andshown in Figure 5c-d. As reported previously,27,33 the releaseprocess has two time phases: high initial release and slowsubsequent release. In the first half an hour, over 50% ofvancomycin is released from NT-V and NT-Ag-V and in thenext 0.5−4 h, these two groups show a relatively slow andconstant elution process. As shown in Figure 5c-d, theconcentrations of vancomycin released are approximately 9.16(NT-V) and 8.09 (NT-Ag-V) μg/mL at 240 min, which are stillhigher than those from the MIC of MRSA in our study (1 μg/mL). NT-Ag and NT-Ag-V are immersed in PBS or TSB forone month but no detectable silver ion (Ag+) release isobserved (below the detection limit of ICP-OES). Figure 5bshows that compared to NT (control), NT-Ag, NT-V, and NT-Ag-V have no cytotoxicity.

3.3. In Vitro Antibacterial Property Against PlanktonicBacteria. The antibacterial rate against planktonic bacteria

Figure 2. SEM images: (a,b) SE images of NT; (c,d) SE image of NT-Ag (c) and BSE image of the same region (d), with the correspondinghigh-magnification insets; (e,f) SE images of NT-V with a high-magnification inset; (g,h) SE image of NT-Ag-V (g) and BSE image ofthe same region (h), with the corresponding high-magnification insets.Note: SE, secondary electron; BSE, backscattered electron; greenarrow, embedded Ag nanoparticles; red arrow, exposed Ag nano-particles; purple arrow, vancomycin.



11166


(Rap) is presented in Figures 6 and S1. With respect to the Rapvalue, although the culture medium is TSBG (Figure S1a) orTSBG+SF (Figure 6a) and compared with NT and NT-Agwhich exhibit little or no lethality against planktonic bacteria inthe culturing period, NT-V and NT-Ag-V have strongerantibacterial ability with the Rap values exceeding 90%. Theantibacterial effects of different samples against planktonicbacteria are qualitatively investigated by SPM as shown inFigure 6b and S1b. During the culturing period, the amount ofbacteria in NT and NT-Ag increase but there is no obviousstatistical significance between them. In comparison, NT-V andNT-Ag-V show fewer bacteria at every time point indicatingthat the latter two samples have better antibacterial ability.3.4. In Vitro Antiadhesive Property Against Sessile

Bacteria. In the short and long-term experiments, theantibacterial rates against adherent bacteria (Raa) are presentedin Figures 7 and S2. For the short-run Raa value, in contrastwith NT, the modified samples (NT-Ag, NT-V and NT-Ag-V)have excellent antiadhesive properties. In TSBG+SF (Figure7a), the Raa values vary from 90%, 90%, and 92% at 1 h to 66%,

98%, and 99% at 24 h, respectively. In TSBG (Figure S2a), theRaa values of NT-Ag, NT-V, and NT-Ag-V range from 88%,89%, and 91% at 1 h to 62%, 98%, and 99% at 24 h,respectively. The antiadhesive effect of NT-Ag becomes weakerwith time. Compared to NT, NT-V and NT-Ag-V show thestrongest antiadhesive activity during short-term incuba-tion.The antiadhesive ability of NT-Ag at different time pointsduring 28 days is assessed. As shown in Figures S2b and 7b, theRaa values of NT-Ag in TSBG and TSBG+SF are around 62%and 66%, respectively, demonstrating that Ag−PIII haspersistent antiadhesive activity and the effect is independentof silver ion release.As shown in Figures 8 and S3, the aforementioned

antiadhered trend in the short term is verified by CLSM andSEM Figures 8a and S3a display the top-views of bacterialadhesion projected by CLSM on the sample surface. As timeelapses, the green fluorescent intensity (viable bacteria) fromNT and NT-Ag increases but the increase from NT-Ag is lessthan that on NT while it decreases from NT-V and NT-Ag-V.The red fluorescent intensity (dead bacteria) from the sample

Figure 3. Cross-sectional morphology and composition by SEM and EDS: (a) SE image of NT with the EDS spectrum of the spot indicated by theyellow arrow (b); (d,e) SE and BSE images of NT-Ag with the EDS spectrum of the spot indicated by the yellow arrow (f); (g) SE image of NT-Vwith the EDS spectrum of spot indicated by the yellow arrow (h); (j,k) SE images of NT-Ag-V with the EDS spectrum of the spot indicated by theyellow arrow spot (l); (c, i) EDS results of the cross sections of various samples. Note: green arrow, embedded Ag nanoparticles; red arrow, exposedAg nanoparticles; purple arrow, vancomycin.



11167


surface shows an opposite tendency. Figures 8b and S3b showthe similar bacterial adhesion trend by SEM. As the test isextended from 1 to 24 h, little bacteria colonization is found onNT-V and NT-Ag-V. In contrast, the number of bacteria on NTincreases and finally a sea of bacteria conglomerates intocolonies to form a recalcitrant biofilm. Although NT-Ag has theantiadhesive ability, some bacteria are observed from thesurface at the end of the experiment. Additionally, the growthstatus of MRSA in the TSBG+SF is different from that inTSBG and there are some fibrous proteins on NT-V and NT-Ag-V after 12 and 24 h.3.5. In Vitro Cell-Bacteria Cocultures. Compared to

single-cultures with either bacteria or cells, our cells-bacteriacoculture better simulates the clinical conditions of early PJI.The adhesion, spreading, and growth behavior of fibroblast cellsare assayed by immunocyto-stained fluorescent image (ISFI)and SEM. As shown in Figure 9a, SEM reveals the cellmorphology on the different surfaces. After 3 days, instead of

adherent cells, clustered bacteria considered as biofilms arepresent on NT, suggesting that it is almost impossible for cellsto survive in this hostile environment. The number of bacteriaon the surface of NT-Ag decreases significantly compared toNT and a few odd-shape cells surrounded by bacteria are foundon the surface. However, robust cells survive well on NT-V andNT-Ag-V, especially NT-Ag-V on which the cells stretch betterwith a few filopodia and establish good intercellularconnections. As shown in the fluorescent images (Figure 9b),NT shows no viable cells and a few cells without filopodiaextensions on NT-Ag sample have a spherical and wizenedmorphology. NT-V has more fibroblasts adhesion to the surfacethan the former two groups and the largest number of cells witha polygonal shape is observed from NT-Ag-V. As shown inFigure 9c, after 3 days, there is no cell surface coverage on NT,indicating that fibroblast cells cannot survive in the presence ofadhering MRSA. NT-Ag shows little surface coverage and thepercentage reaches 12.7%. In contrast, the percentages of

Figure 4. Surface analysis results: (a) Thin-film XRD patterns; (b) Raman spectra; (c-h) XPS core-level spectra for Ti 2p (c), Ag 3d (d), C 1s (e-f)and N 1s (g-h); (i) Simplified schematic of TiO2-resorcinol/Ag-resorcinol coordination, partly adapted from ref 70, with granted copyrightpermission from John Wiley and Sons (License Number: 3841351095289).



11168


surface coverage on NT-V and NT-Ag-V are 27.8% and 47.1%,respectively, and in comparison with the other three groups,NT-Ag-V shows the most significant statistical difference.These results demonstrate that only in the presence ofsufficient anti-infective agents can cells win the race againstattacking bacteria. This can be observed from NT-Ag-V which

indicates the superiority of the combined release-killingvancomycin and contact-killing Ag nanoparticles with TiO2

nanotubes platform as illustrated in Figure 9d.3.6. In Vivo Antibacterial Performance Evaluation. In

our model, the effect of different samples is assessed byobserving the inflammatory symptoms of the knee joint and

Figure 5. Vancomycin loading, release, and sample cytotoxicity: (a) Loading efficiency of vancomycin in the nanotubes; (b) Cytotoxicity offibroblast cells cultured with various samples for 1 and 4 days: (c-d) Release kinetics of vancomycin from NT-V and NT-Ag-V, respectively.

Figure 6. Antibacterial efficacy against planktonic bacteria in vitro after 1, 6, 12, and 24 h: (a) Antibacterial rates of planktonic bacteria (Rap) for fourspecimens in the TSBG+SF medium. ***P < 0.001. (b) Representative photos of recultivated planktonic bacteria colonies on SBA for fourspecimens in the TSBG+SF medium.



11169


biofilm formation on the metal washers, counting the quantityof bacteria on the washers and peri-implant soft tissues, andevaluating the knee joint imageological features and histo-pathological changes.3.6.1. Radiographical and Gross Evaluation for Peri-

Implant Soft Tissues and CFU Counting of Soft Tissues. After2 weeks, the rabbits are examined by X-ray, and theradiographic characteristics of the peri-implant tissue arerevealed (Figure 10a). The soft tissue swelling of NT andNT-Ag is more severe than that of NT-V and NT-Ag-V. Theinflammatory response of the knee is also evaluated. As shownin Figure 10b, the trend of the inflammatory reaction inarticular cavity is the same as the imaged features. Compared toNT and NT-Ag showing thick and white suppuration coveringthe whole joint cavity, NT-V and NT-Ag-V show little or noserosanguineous exudation and necrosis tissue. The averagegross scores of NT, NT-Ag, NT-V, and NT-Ag-V are 2.63 ±0.52, 2.13 ± 0.64, 0.5 ± 0.53, and 0.25 ± 0.46, respectively(Figure 10c). NT and NT-Ag are not statistically significant andthere is also no significant difference between NT-V and NT-Ag-V, but there are differences among NT, NT-Ag, NT-V, andNT-Ag-V (P < 0.05). The trend is NT ≈ NT-Ag > NT-V ≈NT-Ag-V and consistent with that determined by radiology.In addition, soft tissues surrounding the metal washers are

harvested for bacterial load derivation from the homogenates of

soft tissues. With respect to the average CFU counts per gramof peri-washer soft tissues, the trend is as follows: NT ≈ NT-Ag> NT-V ≈ NT-Ag-V (Figure 10d). The differences are notstatistically significant between NT and NT-Ag as well asbetween NT-V and NT-Ag-V but statistically significant amongother groups.

3.6.2. Evaluation of in Vivo Biofilm on the Metal Washersand CFU Counting of Washers. In the knee PJI model, themodified metal washers are removed and the in vivo biofilmformation and bacteria number are determined. According toFigure 11a, there is obvious in vivo biofilm formation on theNT and NT-Ag washers, whereas no biofilm can be observedfrom the NT-V and NT-Ag-V washers. SEM is conducted tofurther examine biofilm formation and bacterial aggregationfrom random regions. As shown in Figure 11b, NT showsbiofilms and dense clustered bacteria. NT-Ag shows reducedbiofilm formation and several bacteria are detected. In contrast,NT-V and NT-Ag-V show no biofilms and different quantitiesof fibroblast cells adhere and grow on the washers. The celladhesion, growth, and accumulation reveal better outcomefrom NT-Ag-V. The bacteria detected by SEM is quantified bysubsequent bacterial cultures. Figure 11c shows the quantifiedbacterial counts from the sonicated eluates of the washers foreach group. The average counts of CFU per washer display thefollowing trend: NT > NT-Ag > NT-V ≈ NT-Ag-V. There isno statistical difference between group NT-V and NT-Ag-V butthere are statistical differences among the other groups.

3.6.3. Histological Evaluation. The morphological changeof the peri-implant soft tissues is assessed by H&E staining andthe bacterial residue is evaluated by Giemsa staining. As shownin Figure 12, the typical feature of soft tissue infection is shownby NT including significant acute inflammation, exudation,synovial tissue necrosis, and neutrophils infiltration into tissues.Moreover, there are many bacteria in the slices of Giemsastaining. NT-Ag shows a relatively mild inflammation reactionand a certain amount of neutrophils through H&E staining, anda few bacteria are observed from the slices of Giemsa staining.However, on NT-V and NT-Ag-V, there is little or no sign ofinfection and the amount of bacteria decreases significantly.There is no significant difference between the two groups.Furthermore, according to the evaluation of soft-tissuecondition, there are fresh granulation tissues with maturefibroblasts and newly formed vessels on NT-Ag-V and NT-V,

Figure 7. Quantitative antibacterial evaluation for sessile bacteria in vitro in the short term and long-term tests: (a) During the course of 24 h, theantibacterial rates of sessile bacteria (Raa) of the four specimens in the TSBG+SF medium. (b) Over the course of 28 days, the antibacterial rates ofsessile bacteria (Raa) for the four specimens in the TSBG+SF medium. **P < 0.01, ***P < 0.001.

Figure 8. Qualitative antibacterial evaluation of the sessile bacteria invitro after 1, 6, 12, and 24 h: (a) In the TSBG+SF medium, fluorescentimages illustrating biofilm formation on the four samples after stainingwith the Baclight live/dead bacteria. Magnification is ×200 and thescale bar is 30 μm. (b) In the TSBG+SF medium, the SEMmorphology showing biofilm formation on the four samples.Magnification is ×2000 and the scale bar is 10 μm.



11170


indicating that the survival state of fibroblast cells improve inthe presence of released antimicrobial agents. As shown inFigure 10b and Figure 12d3-d4, NT-Ag-V shows more thickand robust fibrotic capsules with fibroblast cells and angio-genesis in the granulation tissues than NT-V (Figures 10b and12c3-c4), implying NT-Ag-V promotes early restoration of softtissues around the prosthesis.

4. DISCUSSION

A qualitative relationship between the Ag nanoparticles andrelease behavior of Ag ions is first established. The classicalnucleation theory describes the nucleation and growth behaviorof Ag nanoparticles in nano-TiO2.

42 When the Ag concen-tration exceeds the solubility limit in nano-TiO2, the system

relaxes via nucleation and growth of Ag nanoparticles. DuringAg PIII, energetic Ag ions bombard the nano-TiO2 withsufficient energy and come to rest at a distance from thesurface. The energy lost of the incoming ions is partiallytransformed into atomic scale heating (ASH), which plays asignificant role in Ag cluster evolution.43 Here, the total energygain by an individual Ag ion in Ag PIII is determined by

= + + +E E QeV E Eik bias p

.43 Considering the mean charge state Q ≈ 244 and the meanwork function of nano-TiO2 W ≈ 5 eV,45 the kinetic energy Ei

is calculated to 2.55 eV by the equation.46,47

Figure 9. Antibacterial evaluation in the bacteria/cells coculture model: (a) SEM morphology of the attached bacteria and fibroblasts on the foursamples. The images below have higher magnification than the areas in the red boxes. The blue arrow and green arrow mark the bacteria andfibroblast, respectively. (b) Fluorescent images of fibroblasts on the four samples after staining with DAPI (blue) and TRITC-phalloidin (red). (c)Surface coverage of the four different surfaces after adhesion, spreading, and growth of fibroblast cells for 3 days. **P < 0.01, ***P < 0.001. (d)Schematic illustration of the process of bacteria, fibroblasts, and samples.



11171


∑=− −− +=

−

EW Q j

Q j22( ) 1

8( ) 2i

j

Q

0

1

The energy gain components of the total energy of anindividual Ag ion reaching the nano-TiO2 are listed in Table 2.The total energy gain of an individual Ag ion over a wide biasvoltage (Vbias) is much greater than the minimum displacementenergy in the range of 10−40 eV.43 Hence, during Ag PIII, thestationary Ag atoms under the nano-TiO2 surface can beremoved and dislodged. Accordingly, the Ag PIII process canbe divided into two regimes. At a relatively small bias voltage,for example, Vbias = 0.5/1.0 kV (Table S1), the implantingdepth of Ag is shallow and Ag nanoparticles can migrate to thesurface with ease. In this case, the longer Ag PIII is carried outon the nano-TiO2, the more Ag nanoparticles gather on thesurface. They gradually undergo Ostwald ripening, coalescefrom being discrete to continuous, and even form a nanothickAg thin film on the nano-TiO2 surface (Figure 13a). At arelatively high bias is used in Ag PIII, for instance, 30 kV asused in our experiments here, since the penetration depth islarger, the Ag nanoparticles have a long path to migrate out ofthe interior of nano-TiO2. The majority of the Ag nanoparticlesare embedded in the interior of nano-TiO2 beneath the surface,

although some of them can still migrate into the surface. As aconsequence, Ag ions cannot be released quickly from the Ag-embedded TiO2 and the released amount is smaller than 10ppb48,49 which is less than the minimum bactericidalconcentration of 100 ppb.50 Prolonged Ag+ leaching rendersthe release-killing ability of the Ag-implanted surface whilelimited Ag+ release contributes to a contact-killing as illustratedin Figure 13b. To further verify the above theoretical analysisand give a direct experimental evidence, high resolution TEManalysis was carried out to investigate the presence of Agnanoparticles on/in TiO2 nanotubes. As shown in Figure 13c,in a randomly selected area, prevalent nanoparticles can be seenalong the nanotube wall from the inside to the nanotube top,which include the partially exposed ones (the minority,indicated by “1”) and the embedded ones (the majority,indicated by “2”). Furthermore, the corresponding elementalanalysis results of microregions “1” and “2” by EDS were givenin Figure 13g, which demonstrated that the above nanoparticleswere Ag nanoparticles. Likewise, Figure 13d−f also demon-strated the existence of Ag nanoparticles, consistent with theabove analysis (Please see supplementary details). In this case,direct contact of adherent bacteria with Ag nanoparticles-embedded TiO2 nanotubes will cause the disruption of their

Figure 10. Systematic evaluation of the peri-implant soft tissues in vivo in the four groups: (a) X-rays images of rabbits inserted with various implantsat 15 days post surgery. The red oval marks soft tissue swelling and there are few or no infective signs in groups NT-V and NT-Ag-V. (b) Generalobservation of peri-implant soft tissues inflammation from the four groups. (c) Gross scores of peri-implant soft tissues inflammation in four groups.***P < 0.001. (d) Survival of bacteria in the soft tissues surrounding the washer. The results are expressed as the actual amounts of CFUs retrieved.The horizontal line demonstrates the median value and each group includes eight animals. *P < 0.05, **P < 0.01, ***P < 0.001.



11172


normal behaviors contributed by electron transfer and freeradical formation.51,52

A growing number of TJAs accelerate the development of PJIso that their prevention is a continuous and arduouschallenge.13,14 Implant-associated infections are commonlycaused by surface-adhered bacteria that form thick andmultilayered biofilms.53,54 In terms of pathogens, comparedwith S. epidermidis which produces less virulence factors, S.aureus is more toxogenic and once infected, multiple virulencefactors of MRSA enable it to invade and undermine the hosttissues and the infection tends to be more destructive.53,55

However, although MRSA invades host tissues surroundingprostheses or forms a biofilm on implant surfaces, it is crucialthat MRSA should survive in the local surgical site and thepostoperative impaired host immune defense cannot resist thestrong etiological agent. If MRSA has enough time and space tocolonize the prosthesis and peri-prosthesis tissues and attackthem, early PJI will be difficult to eradicate. Therefore, weshould take advantage of a high concentration of antibacterialagents to sterilize as many bacteria as possible by means ofrelease from the antimicrobial coatings. However, a smallamount of residual bacteria still have the chance to populate theimplant surface. It has been demonstrated that S. aureus biofilmformation can be induced by subminimal inhibitory concen-trations of some antibiotics such as cephalothin56 andvancomycin,57 and this is thought to arise from a global generegulation in response to cell stress.58 Hence, an antiadhesivesurface is also needed. With regard to systematic prophylaxis ofearly MRSA-PJI, a single release-killing or contact-sterilizingsurface is not good enough and the most efficacious method is

to construct an implant coating that combines a controlledrelease system that kills nonadherent bacteria during the short-term and a permanent antiadhesive layer that inhibits bacterialcolonization in the long term.Clinically, a 6-h postimplantation period is the “decisive

period” to prevent bacterial adhesion.59,60 For the prophylaxisof surgical site infections (SSIs), a guiding documentsummarizing the clinical practice has been published by theU.S. Department of Health and Human Services andpreoperative intravenous administration of antimicrobial agentsshould be provided for a few hours after the surgery.61

However, local delivery with fewer side effects is superior tosystematic administration which brings small topical drugconcentrations but high intravenous concentrations and mayinduce systemic toxicity.14 Certainly, our anti-infective coatingsplay a good role in the prophylaxis of early MRSA-PJI. Asshown in Figures 6-8 and S1−S3, the released vancomycin cankill planktonic and sessile MRSA in the early stage. As timegoes on, the quantity of released vancomycin diminishes(Figure 5c-d) and the leached vancomycin can be graduallymetabolized in vivo. With regard to the adhered bacteria, NT-Ag shows some antibacterial effects during the course of 24 hand the immobilized Ag NPs can also provide long-term andstable antiadhesive effects up to 28 days. Specifically, unlike thecoatings prepared by Cheng et al.17 and Mei et al.62 which havegood antibacterial properties based on the release of silver ions,there is insignificant Ag+ leaching from NT-Ag thus minimizingthe cytotoxicity induced by Ag+.In our in vitro study, we first evaluate the antibacterial

properties ex vivo with human SF. With regard to the choice of

Figure 11. Systematic evaluation of the metal washers in vivo in the four groups: (a) General observation of in vivo biofilm formation on the fourwashers. (b) Representative SEM images of the attached bacteria on the four washers. The images in the blue boxes are higher magnification of theareas in the red boxes. (c) Survival of bacteria on the washes. The results are expressed as the actual amounts of CFUs retrieved and the horizontalline demonstrates the median value. Each group includes eight animals. *P < 0.05, **P < 0.01, ***P < 0.001.



11173


experimental strains, previous studies for reproducible resultsu s e t h e we l l - k nown s t a nd a r d s t r a i n (MRSA ,ATCC433000),17,22 but such strain cannot represent theclinical prevalent strain type. Campoccia et al. has suggestedthat the reference strains should be of the major clinicalprevailing strain type63 and the selection should be based onupdated studies of molecular epidemiology.64 In our experi-ments, for better simulating clinical conditions, we choose theclinical MRSA (ST239), an epidemic HA-MRSA in China,65

and it enhances the validity of our experiment. As for theanimal model that closely models early MRSA-PJI, we candirectly visualize in vivo biofilm formation, find the intravitalbiofilm-forming characteristics of clinical MRSA, and detectbacterial accumulation on the washers by SEM. Moreover, ourprotocols also have the following two advantages. One is thatMRSA is introduced directly into the knee joint when theincision is closed, but previous models import bacteria directlyinto the bone marrow cavity prior to biomaterial insertion orimmerse implants in a bacterial culture suspension prior tointra-articular implantation.66 The second one is that it includesthe essential fixed type presented in total knee arthroplasty.

The use of in vitro coculturing model is also in keeping withthe clinical situation that intraoperative bacterial contaminationof primary TJA is common,67 and the dissected periprostheticsoft-tissue with the impaired host immune defense makes thejoint susceptible to bacteria.13,18 Gristina has suggested thatthere is a race to the implant surface between the host cellsfrom the peri-implant tissue and invading bacteria.68 Themodified surface can afford the projection of cell-to-substratumevents68 and for the host, mammalian cells can rapidly inhabitthe surface prior to incursive pathogens,26 which will greatlydecrease PJI. As shown in Figure 9, in contrast with NT, NT-Ag has a certain degree of antibacterial activity and a fewirregularly shaped fibroblasts adhere to the surface. Incomparison, NT-V and NT-Ag-V both show prominentantimicrobial effects, but adhesion and growth of fibroblastson NT-Ag-V is better than that on the other three surfaces. Inconcert with the coculturing results, it is also found that NT-Ag-V in vivo presents the best peri-implant soft tissues growthwithout infective inflammation(Figure 10b) and demonstratesburly fibroblasts on the washer (Figure 11b4). During culturing,there is a fierce competition between MRSA and fibroblast cellsin the race to the implant surface, and fibroblasts cannot surviveor adhere to the surface when the number of bacteria is high.Although NT-Ag possesses the antibacterial ability againstadhered MRSA, it cannot kill the floating bacteria so thatfibroblast cells cannot anchor on the surface. Nevertheless, asthe amount of bacteria is reduced by the action of released

Figure 12. Histological evaluation of the peri-implant soft tissues at 15 days postsurgery with the red arrow showing the interface of the implant andperi-implant soft tissue: Histological slices in a sagittal plane after H&E staining and Giemsa staining. Group NT: Overview image (a1) and close-upviews (a2-a3) show severe synovial tissue necrosis and lots of neutrophils infiltrated into tissues. A large number of bacteria (arrowheads) are alsopresented in the soft tissues (a4). Group NT-Ag: Overview image (b1) and close-up views (b2-b3) reveal relatively alleviated synovial tissue necrosisand certain amount of neutrophils. Many bacteria reside in the soft tissues (b4). Group NT-V and NT-Ag-V: Overview image (c1 and d1) and close-up views (c2-c3, d2-d3) display no obvious histopathological signs of infection and fresh granulation tissues formation with mature fibroblasts andnewly formed vessels. Bacteria cannot be found from the tissues (c4 and d4).

Table 2. Energy Gain of an Individual Silver Ion during PIII

Ek (eV) QeVbias (eV) Ei (eV) Ec (eV) Ee (eV) EQ (eV) Vbias (kV)

69.00a 6.00 × 104 2.55 2.95a ∼0b 29.1a 30.0aFrom ref 43. bFrom ref 69.



11174


vancomycin, fibroblast cells have the opportunity to stick to theNT-Ag-V surface to resist bacteria. The intricately modifiedsurface not only kills planktonic bacteria directly and preventsbacterial adhesion, but also better supports the cell-to-substratum events to reduce infection, which suggests thesynergetic superiority of the combined release-killing vanco-mycin and contact-killing Ag nanoparticles with TiO2 nano-tubes platform. A robust fibrous interface around the implantsalso serves as a phylactic barrier to keep implants frommicroorganic invasion. It is of vital importance that earlypostoperative rehabilitation and the gradually recuperativeimmune system, albeit attenuated in the peri-implant environ-ment, can protect the space from bacterial colonization. Thecombination of release-killing and contact-killing properties onthe designed surface is illustrated in Figure 14a, which also

shows the trap-killing ability contributed by Ag nanoparticlesmaking surface charge shift positively to electrostatically attractnegatively charged bacteria. Figure 14b summarizes theplausible antibacterial and antibiofilm mechanisms on thetitanium surface in a cells-bacteria coexisting system.

5. CONCLUSIONS

Silver nanoparticles are incorporated into TiO2 nanotubes byplasma immersion ion implantation (PIII) to produce acontact-killing antimicrobial surface with limited silver ionrelease. Vancomycin is subsequently loaded into the nanotubesby vacuum extraction and lyophilization to establish a release-killing surface. The release profile of vancomycin exhibits twotime phases: initially large release followed by slow subsequentrelease, but no detectable silver ions are lixiviated from the Ag−

Figure 13. (a) Schematic illustrating the basic physical processes (from left to right) involved in the generation of Ag nanoparticles (purple balls) innano-TiO2 by Ag plasma immersion ion implantation (Ag PIII). (b) Schematic illustrating the contact-killing mode and release-killing mode of Agnanoparticles. (c−f) TEM images showing the existence of Ag nanoparticles in the TiO2 nanotubes together with the corresponding EDS resultsacquired from spots “1” to “8” (g).



11175


PIII nanotubes during a long immersion time period. Asverified by our in vitro and in vivo models, the vancomycin-loaded Ag-implanted TiO2 nanotubes show excellent anti-bacterial activity against both planktonic and sessile bacteriaduring the 24 h culturing period and bacterial adhesion can beinhibited up to 28 days. This bodes well for prophylaxis of earlyMRSA-PJI during the perioperative period and perhaps evenlate-developing PJI. The dual-function surface can assistfibroblasts to combat bacteria in the coculture. The combinedcontact-killing and release-killing strategy provides new insightsinto surface modification of antibacterial biomaterials. Thesuperficial nanoarchitecture is well established as a bondbetween the inorganic bactericide (Ag NPs) and organicantibiotic (vancomycin) to obtain a total victory in the battle ofPJI. Owing to the dual antibacterial properties and cell-assistingfunctions, the materials and method are very attractive to theprevention of PJI and promotion of soft tissue reparationespecially for prosthesis implants.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acsami.6b02803.

Additional materials as discussed in the text. (PDF)

■ AUTHOR INFORMATIONCorresponding Authors*(X.Z.) Tel.: +86 21 6436 9183. Fax: +86 21 6470 1363. E-mail: [email protected].*(X.L.) Tel.: +86 21 5241 2409. Fax: +86 21 5241 2409. E-mail: [email protected].*(H.S.) Tel.: +86 21 6436 9183. Fax: +86 21 6470 1363. E-mail: [email protected].

Author Contributions∇These authors contributed equally to this work.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors acknowledge the financial support from theNational Basic Research Program of China (973 Program,2012CB933600), National Natural Science Foundation ofChina (Grant Nos. 81271962, 81472109, and 81472108),National Science Fund for Distinguished Young Scholars(51525207), Shanghai Committee of Science and Technology,China (15441904900 and 14XD1403900), and Hong KongResearch Grants Council (RGC) General Research Funds(GRF) No. CityU 11301215.

■ REFERENCES(1) Zimmerli, W.; Trampuz, A.; Ochsner, P. E. Prosthetic-JointInfections. N. Engl. J. Med. 2004, 351, 1645−1654.(2) Shirwaiker, R. A.; Springer, B. D.; Spangehl, M. J.; Garrigues, G.E.; Lowenberg, D. W.; Garras, D. N.; Yoo, J. U.; Pottinger, P. S. AClinical Perspective on Musculoskeletal Infection Treatment Strategiesand Challenges. J. Am. Acad. Orthop Sur 2015, 23 (Suppl), S44−S54.(3) Kurtz, S.; Ong, K.; Lau, E.; Mowat, F.; Halpern, M. Projections ofPrimary and Revision Hip and Knee Arthroplasty in the United Statesfrom 2005 to 2030. J. Bone Joint Surg. Am. 2007, 89, 780−785.(4) Fitzgerald, R. H., Jr.; Nolan, D. R.; Ilstrup, D. M.; Van Scoy, R. E.;Washington, J. A., 2nd; Coventry, M. B. Deep Wound SepsisFollowing Total Hip Arthroplasty. J. Bone Joint Surg Am. 1977, 59,847−855.(5) Parvizi, J.; Azzam, K.; Ghanem, E.; Austin, M. S.; Rothman, R. H.Periprosthetic Infection Due to Resistant Staphylococci: SeriousProblems on the Horizon. Clin. Orthop. Relat. Res. 2009, 467, 1732−1739.(6) Walls, R. J.; Roche, S. J.; O’Rourke, A.; McCabe, J. P. Surgical SiteInfection with Methicillin-Resistant Staphylococcus Aureus afterPrimary Total Hip Replacement. J. Bone Jt. Surg., Br. Vol. 2008, 90,292−298.(7) Arciola, C. R.; Campoccia, D.; Speziale, P.; Montanaro, L.;Costerton, J. W. Biofilm Formation in Staphylococcus ImplantInfections. A Review of Molecular Mechanisms and Implications forBiofilm-Resistant Materials. Biomaterials 2012, 33, 5967−5982.(8) Arciola, C. R.; Visai, L.; Testoni, F.; Arciola, S.; Campoccia, D.;Speziale, P.; Montanaro, L. Concise Survey of Staphylococcus AureusVirulence Factors That Promote Adhesion and Damage to Peri-Implant Tissues. Int. J. Artif. Organs 2011, 34, 771−780.(9) Aggarwal, V. K.; Bakhshi, H.; Ecker, N. U.; Parvizi, J.; Gehrke, T.;Kendoff, D. Organism Profile in Periprosthetic Joint Infection:Pathogens Differ at Two Arthroplasty Infection Referral Centers inEurope and in the United States. J. Knee Surg 2014, 27, 399−406.(10) Cobo, J.; Miguel, L. G.; Euba, G.; Rodriguez, D.; Garcia-Lechuz,J. M.; Riera, M.; Falgueras, L.; Palomino, J.; Benito, N.; del Toro, M.D.; Pigrau, C.; Ariza, J. Early Prosthetic Joint Infection: Outcomes withDebridement and Implant Retention Followed by Antibiotic Therapy.Clin. Microbiol. Infect. 2011, 17, 1632−1637.(11) Soriano, A.; Garcia, S.; Bori, G.; Almela, M.; Gallart, X.; Macule,F.; Sierra, J.; Martinez, J. A.; Suso, S.; Mensa, J. Treatment of Acute

Figure 14. (a) Schematic illustrating the antibacterial actions of thedesigned surface boasting a combination of release-killing, contact-killing, and trap-killing capabilities. (b) Schematic illustrating thepossible antibacterial and antibiofilm mechanisms on the titaniumsurface in a cell-bacteria coexisting system.



11176

http://pubs.acs.org

http://pubs.acs.org/doi/abs/10.1021/acsami.6b02803

http://pubs.acs.org/doi/suppl/10.1021/acsami.6b02803/suppl_file/am6b02803_si_001.pdf

mailto:[email protected]




Post-Surgical Infection of Joint Arthroplasty. Clin. Microbiol. Infect.2006, 12, 930−933.(12) Neoh, K. G.; Kang, E. T. Combating Bacterial Colonization onMetals via Polymer Coatings: Relevance to Marine and MedicalApplications. ACS Appl. Mater. Interfaces 2011, 3, 2808−2819.(13) Gallo, J.; Holinka, M.; Moucha, C. S. Antibacterial SurfaceTreatment for Orthopaedic Implants. Int. J. Mol. Sci. 2014, 15, 13849−13880.(14) ter Boo, G. J.; Grijpma, D. W.; Moriarty, T. F.; Richards, R. G.;Eglin, D. Antimicrobial Delivery Systems for Local InfectionProphylaxis in Orthopedic- and Trauma Surgery. Biomaterials 2015,52, 113−125.(15) Vasilev, K.; Sah, V.; Anselme, K.; Ndi, C.; Mateescu, M.;Dollmann, B.; Martinek, P.; Ys, H.; Ploux, L.; Griesser, H. J. TunableAntibacterial Coatings That Support Mammalian Cell Growth. NanoLett. 2010, 10, 202−207.(16) Qin, H.; Cao, H.; Zhao, Y.; Zhu, C.; Cheng, T.; Wang, Q.; Peng,X.; Cheng, M.; Wang, J.; Jin, G.; Jiang, Y.; Zhang, X.; Liu, X.; Chu, P.K. In Vitro and in Vivo Anti-Biofilm Effects of Silver NanoparticlesImmobilized on Titanium. Biomaterials 2014, 35, 9114−9125.(17) Cheng, H.; Li, Y.; Huo, K.; Gao, B.; Xiong, W. Long-Lasting inVivo and in Vitro Antibacterial Ability of Nanostructured TitaniaCoating Incorporated with Silver Nanoparticles. J. Biomed. Mater. Res.,Part A 2014, 102, 3488−3499.(18) Sendi, P.; Zimmerli, W. Challenges in Periprosthetic Knee-JointInfection. Int. J. Artif. Organs 2011, 34, 947−956.(19) Font-Vizcarra, L.; Garcia, S.; Martinez-Pastor, J. C.; Sierra, J. M.;Soriano, A. Blood Culture Flasks for Culturing Synovial Fluid inProsthetic Joint Infections. Clin. Orthop. Relat. Res. 2010, 468, 2238−2243.(20) Sendi, P.; Zimmerli, W. Diagnosis of Periprosthetic JointInfections in Clinical Practice. Int. J. Artif. Organs 2012, 35, 913−922.(21) Gurkan, I.; Wenz, J. F. Perioperative Infection Control: AnUpdate for Patient Safety in Orthopedic Surgery. Orthopedics 2006, 29,329−339 ; quiz 340−1..(22) Jin, G.; Qin, H.; Cao, H.; Qiao, Y.; Zhao, Y.; Peng, X.; Zhang,X.; Liu, X.; Chu, P. K. Zn/Ag Micro-Galvanic Couples Formed onTitanium and Osseointegration Effects in the Presence of S. aureus.Biomaterials 2015, 65, 22−31.(23) Zhao, Y.; Cao, H.; Qin, H.; Cheng, T.; Qian, S.; Cheng, M.;Peng, X.; Wang, J.; Zhang, Y.; Jin, G.; Zhang, X.; Liu, X.; Chu, P. K.Balancing the Osteogenic and Antibacterial Properties of Titanium byCodoping of Mg and Ag: An in Vitro and in Vivo Study. ACS Appl.Mater. Interfaces 2015, 7, 17826−17836.(24) Dastgheyb, S.; Parvizi, J.; Shapiro, I. M.; Hickok, N. J.; Otto, M.Effect of Biofilms on Recalcitrance of Staphylococcal Joint Infection toAntibiotic Treatment. J. Infect. Dis. 2015, 211, 641−650.(25) Broekhuizen, C. A.; de Boer, L.; Schipper, K.; Jones, C. D.;Quadir, S.; Vandenbroucke-Grauls, C. M.; Zaat, S. A. StaphylococcusEpidermidis Is Cleared from Biomaterial Implants but Persists in Peri-Implant Tissue in Mice despite Rifampicin/Vancomycin Treatment. J.Biomed. Mater. Res., Part A 2008, 85, 498−505.(26) Hickok, N. J.; Shapiro, I. M. Immobilized Antibiotics to PreventOrthopaedic Implant Infections. Adv. Drug Delivery Rev. 2012, 64,1165−1176.(27) Popat, K. C.; Eltgroth, M.; Latempa, T. J.; Grimes, C. A.; Desai,T. A. Decreased Staphylococcus Epidermis Adhesion and IncreasedOsteoblast Functionality on Antibiotic-Loaded Titania Nanotubes.Biomaterials 2007, 28, 4880−4888.(28) Xin, Y.; Jiang, J.; Huo, K.; Hu, T.; Chu, P. K. Bioactive SrTiO3

Nanotube Arrays: Strontium Delivery Platform on Ti-BasedOsteoporotic Bone Implants. ACS Nano 2009, 3, 3228−3234.(29) Schweizer, M. L.; Chiang, H. Y.; Septimus, E.; Moody, J.; Braun,B.; Hafner, J.; Ward, M. A.; Hickok, J.; Perencevich, E. N.; Diekema,D. J.; Richards, C. L.; Cavanaugh, J. E.; Perlin, J. B.; Herwaldt, L. A.Association of a Bundled Intervention with Surgical Site Infectionsamong Patients Undergoing Cardiac, Hip, or Knee Surgery. Jama2015, 313, 2162−2171.

(30) Wilcox, M. H.; Kite, P.; Mills, K.; Sugden, S. In SituMeasurement of Linezolid and Vancomycin Concentrations inIntravascular Catheter-Associated Biofilm. J. Antimicrob. Chemother.2001, 47, 171−175.(31) Eckhardt, S.; Brunetto, P. S.; Gagnon, J.; Priebe, M.; Giese, B.;Fromm, K. M. Nanobio Silver: Its Interactions with Peptides andBacteria, and its Uses in Medicine. Chem. Rev. 2013, 113, 4708−4754.(32) Mei, S.; Zhao, L.; Wang, W.; Ma, Q.; Zhang, Y. BiomimeticTitanium Alloy with Sparsely Distributed Nanotubes Could EnhanceOsteoblast Functions. Adv. Eng. Mater. 2012, 14, B166−B174.(33) Zhang, H.; Sun, Y.; Tian, A.; Xue, X. X.; Wang, L.; Alquhali, A.;Bai, X. Improved Antibacterial Activity and Biocompatibility onVancomycin-Loaded TiO2 Nanotubes: in Vivo and in Vitro Studies.Int. J. Nanomed. 2013, 8, 4379−4389.(34) Yue, C.; Kuijer, R.; Kaper, H. J.; van der Mei, H. C.; Busscher,H. J. Simultaneous Interaction of Bacteria and Tissue Cells withPhotocatalytically Activated, Anodized Titanium Surfaces. Biomaterials2014, 35, 2580−2587.(35) Zhai, H.; Pan, J.; Pang, E.; Bai, B. Lavage with Allicin inCombination with Vancomycin Inhibits Biofilm Formation byStaphylococcus Epidermidis in a Rabbit Model of Prosthetic JointInfection. PLoS One 2014, 9, e102760.(36) Craig, M. R.; Poelstra, K. A.; Sherrell, J. C.; Kwon, M. S.; Belzile,E. L.; Brown, T. E. A Novel Total Knee Arthroplasty Infection Modelin Rabbits. J. Orthop. Res. 2005, 23, 1100−1104.(37) Fujimura, S.; Sato, T.; Kikuchi, T.; Zaini, J.; Gomi, K.;Watanabe, A. Efficacy of Clarithromycin plus Vancomycin in Micewith Implant-Related Infection Caused by Biofilm-Forming Staph-ylococcus Aureus. J. Orthop. Sci. 2009, 14, 658−661.(38) Chis, V.; Venter, M. M.; Leopold, N.; Cozar, O. Raman, Surface-Enhanced Raman Scattering and DFT Study of Para-Nitro-Aniline.Vib. Spectrosc. 2008, 48, 210−214.(39) Ku, S. H.; Ryu, J.; Hong, S. K.; Lee, H.; Park, C. B. GeneralFunctionalization Route for Cell Adhesion on Non-Wetting Surfaces.Biomaterials 2010, 31, 2535−2541.(40) Wang, T.; Zhang, Z.; Liao, F.; Cai, Q.; Li, Y.; Lee, S.-T.; Shao,M. The Effect of Dielectric Constants on Noble Metal/SemiconductorSERS Enhancement: FDTD Simulation and Experiment Validation ofAg/Ge and Ag/Si Substrates. Sci. Rep. 2014, 4, 4052.(41) Ciampi, S.; Bocking, T.; Kilian, K. A.; James, M.; Harper, J. B.;Gooding, J. J. Functionalization of Acetylene-Terminated Monolayerson Si(100) Surfaces: A Click Chemistry Approach. Langmuir 2007, 23,9320−9329.(42) Stepanov, A. L. Applications of Ion Implantation forModification of TiO2: A Review. Rev. Adv. Mater. Sci. 2012, 30,150−165.(43) Anders, A. Atomic Scale Heating in Cathodic Arc PlasmaDeposition. Appl. Phys. Lett. 2002, 80, 1100−1102.(44) Brown, I. G. Vacuum Arc Ion Sources. Rev. Sci. Instrum. 1994,65, 3061−3081.(45) Xiong, G.; Shao, R.; Droubay, T. C.; Joly, A. G.; Beck, K. M.;Chambers, S. A.; Hess, W. P. Photoemission Electron Microscopy ofTiO2 Anatase Films Embedded with Rutile Nanocrystals. Adv. Funct.Mater. 2007, 17, 2133−2138.(46) Burgdorfer, J.; Meyer, F. Image Acceleration of MultiplyCharged Ions by Metallic Surfaces. Phys. Rev. A: At., Mol., Opt. Phys.1993, 47, R20−R22.(47) Cao, H.; Qiao, Y.; Liu, X.; Lu, T.; Cui, T.; Meng, F.; Chu, P. K.Electron Storage Mediated Dark Antibacterial Action of Bound SilverNanoparticles: Smaller Is not Always Better. Acta Biomater. 2013, 9,5100−5110.(48) Huang, R.; Han, Y.; Lu, S. Enhanced Osteoblast Functions andBactericidal Effect of Ca and Ag Dual-Ion Implanted Surface Layers onNanograined Titanium Alloys. J. Mater. Chem. B 2014, 2, 4531−4543.(49) Zheng, Y.; Li, J.; Liu, X.; Sun, J. Antimicrobial and OsteogenicEffect of Ag-Implanted Titanium with a Nanostructured Surface. Int. J.Nanomed. 2012, 7, 875−884.(50) Zhou, W.; Li, W.; Wang, J.-Q.; Qu, Y.; Yang, Y.; Xie, Y.; Zhang,K.; Wang, L.; Fu, H.; Zhao, D. Ordered Mesoporous Black TiO2 as



11177


Highly Efficient Hydrogen Evolution Photocatalyst. J. Am. Chem. Soc.2014, 136, 9280−9283.(51) Pang, M.; Hu, J.; Zeng, H. C. Synthesis, Morphological Control,and Antibacterial Properties of Hollow/Solid Ag2S/Ag Heterodimers.J. Am. Chem. Soc. 2010, 132, 10771−10785.(52) Cao, H.; Qiao, Y.; Meng, F.; Liu, X. Spacing-DependentAntimicrobial Efficacy of Immobilized Silver Nanoparticles. J. Phys.Chem. Lett. 2014, 5, 743−748.(53) Gordon, R. J.; Lowy, F. D. Pathogenesis of Methicillin-ResistantStaphylococcus Aureus Infection. Clin. Infect. Dis. 2008, 46, S350−S359.(54) Del Pozo, J. L.; Patel, R. Clinical Practice. Infection Associatedwith Prosthetic Joints. N. Engl. J. Med. 2009, 361, 787−794.(55) von Eiff, C.; Peters, G.; Heilmann, C. Pathogenesis of InfectionsDue to Coagulase-Negative Staphylococci. Lancet Infect. Dis. 2002, 2,677−685.(56) Subrt, N.; Mesak, L. R.; Davies, J. Modulation of VirulenceGene Expression by Cell Wall Active Antibiotics in StaphylococcusAureus. J. Antimicrob. Chemother. 2011, 66, 979−984.(57) Mirani, Z. A.; Jamil, N. Effect of Sub-Lethal Doses ofVancomycin and Oxacillin on Biofilm Formation by VancomycinIntermediate Resistant Staphylococcus Aureus. J. Basic Microbiol. 2011,51, 191−195.(58) Kaplan, J. B. Antibiotic-Induced Biofilm Formation. Int. J. Artif.Organs 2011, 34, 737−751.(59) Poelstra, K. A.; Barekzi, N. A.; Rediske, A. M.; Felts, A. G.;Slunt, J. B.; Grainger, D. W. Prophylactic Treatment of Gram-Positiveand Gram-Negative Abdominal Implant Infections Using LocallyDelivered Polyclonal Antibodies. J. Biomed. Mater. Res. 2002, 60, 206−215.(60) Hetrick, E. M.; Schoenfisch, M. H. Reducing Implant-RelatedInfections: Active Release Strategies. Chem. Soc. Rev. 2006, 35, 780−789.(61) Mangram, A. J.; Horan, T. C.; Pearson, M. L.; Silver, L. C.;Jarvis, W. R. Guideline for Prevention of Surgical Site Infection, 1999.Centers for Disease Control and Prevention (CDC) HospitalInfection Control Practices Advisory Committee. Am. J. Infect. Control1999, 27, 97−132 ; quiz 133−4; discussion 96..(62) Mei, S.; Wang, H.; Wang, W.; Tong, L.; Pan, H.; Ruan, C.; Ma,Q.; Liu, M.; Yang, H.; Zhang, L.; Cheng, Y.; Zhang, Y.; Zhao, L.; Chu,P. K. Antibacterial Effects and Biocompatibility of Titanium Surfaceswith Graded Silver Incorporation in Titania Nanotubes. Biomaterials2014, 35, 4255−4265.(63) Campoccia, D.; Montanaro, L.; Moriarty, T. F.; Richards, R. G.;Ravaioli, S.; Arciola, C. R. The Selection of Appropriate BacterialStrains in Preclinical Evaluation of Infection-Resistant Biomaterials.Int. J. Artif. Organs 2008, 31, 841−847.(64) Montanaro, L.; Campoccia, D.; Arciola, C. R. Advancements inMolecular Epidemiology of Implant Infections and Future Perspec-tives. Biomaterials 2007, 28, 5155−5168.(65) Li, M.; Du, X.; Villaruz, A. E.; Diep, B. A.; Wang, D.; Song, Y.;Tian, Y.; Hu, J.; Yu, F.; Lu, Y.; Otto, M. MRSA Epidemic Linked to aQuickly Spreading Colonization and Virulence Determinant. Nat. Med.2012, 18, 816−819.(66) Stavrakis, A. I.; Niska, J. A.; Loftin, A. H.; Billi, F.; Bernthal, N.M. Understanding Infection: A Primer on Animal Models ofPeriprosthetic Joint Infection. Sci. World J. 2013, 2013, 925906.(67) Jonsson, E. O.; Johannesdottir, H.; Robertsson, O.; Mogensen,B. Bacterial Contamination of the Wound During Primary Total Hipand Knee Replacement. Median 13 Years of Follow-Up of 90Replacements. Acta Orthop. 2014, 85, 159−164.(68) Gristina, A. G. Biomaterial-Centered Infection: MicrobialAdhesion versus Tissue Integration. Science (Washington, DC, U. S.)1987, 237, 1588−1595.(69) Anders, A. Cathodic Arcs: From Fractal Spots to EnergeticCondensation; Springer Science Business Media, LLC: New York,2008.(70) Anderson, T. H.; Yu, J.; Estrada, A.; Hammer, M. U.; Waite, J.H.; Israelachvili, J. N. The Contribution of DOPA to Substrate−

Peptide Adhesion and Internal Cohesion of Mussel-Inspired SyntheticPeptide Films. Adv. Funct. Mater. 2010, 20, 4196−4205.



11178


S-1

Supporting Information

Antibacterial Surface Design of Titanium-Based Biomaterials for

Enhanced Bacteria-Killing and Cell-Assisting Functions Against

Periprosthetic Joint Infection

Jiaxing Wang a,1, Jinhua Li b,e,1, Shi Qian b,1, Geyong Guo a, Qiaojie Wang a, Jin Tang d,

Hao Shen a,*, Xuanyong Liu b,*, Xianlong Zhang a,* and Paul K. Chu c

a Department of Orthopaedics, Shanghai Jiao Tong University Affiliated Sixth

People’s Hospital, Shanghai Jiao Tong University, Shanghai 200233, China b State Key Laboratory of High Performance Ceramics and Superfine Microstructure,

Shanghai Institute of Ceramics, Chinese Academy of Sciences, Shanghai 200050,

China c Department of Physics and Materials Science, City University of Hong Kong, Tat

Chee Avenue, Kowloon, Hong Kong, China d Department of Clinical Laboratory, Shanghai Jiao Tong University Affiliated Sixth

People’s Hospital, Shanghai Jiao Tong University, Shanghai 200233, China e University of Chinese Academy of Sciences, Beijing 100049, China

* Corresponding Authors:

Prof. Xianlong Zhang

Tel.: +86 21 6436 9183; Fax: +86 21 6470 1363.

E-mail: [email protected]

Prof. Xuanyong Liu

Tel.: +86 21 5241 2409; Fax: +86 21 5241 2409.

E-mail: [email protected]

Prof. Hao Shen

Tel.: +86 21 6436 9183; Fax: +86 21 6470 1363.

E-mail: [email protected] 1 These authors contributed equally to this work.

S-2

S-1. Supporting Information

Figure S1. Antibacterial efficacy against planktonic bacteria in vitro after 1, 6, 12,

and 24 hours: (a) Antibacterial rates of planktonic bacteria (Rap) for four specimens

in the TSBG medium. ***P<0.001. (b) Representative photos of recultivated

planktonic bacteria colonies on sheep blood agar for four specimens in the TSBG

medium.

S-3

Figure S2. Quantitative antibacterial evaluation for sessile bacteria in vitro in the

short-term and long-term tests: (a) During the course of 24 hours, the antibacterial

rates of sessile bacteria (Raa) of the four specimens in the TSBG medium. (b) Over

the course of 28 days, the antibacterial rates of sessile bacteria (Raa) for the four

specimens in the TSBG medium. **P<0.01, ***P<0.001.

S-4

Figure S3. Qualitative antibacterial evaluation of the sessile bacteria in vitro after 1, 6,

12, and 24 hours: (a) In the TSBG medium, fluorescent images illustrating biofilm

formation on the four samples after staining with the Baclight live/dead bacteria.

Magnification is ×200 and the scale bar is 30 µm. (b) In the TSBG medium, the SEM

morphology showing biofilm formation on the four samples. Magnification is ×2000

and the scale bar is 10 µm.

S-5

S-2. Supplementary Analysis

Herein, we firstly try to build a qualitative relationship between the presence form

of Ag nanoparticles and the release behavior of Ag ions. Back to the source, firstly,

the classical nucleation theory can give us a qualitative comprehension of the

nucleation and growth behavior of Ag nanoparticles in nano-TiO21. The equation,

32 4 ln43

ππ γ∆ = −m

r RT SG rV

, expresses the factors that determine the excess free

energy (∆G) during crystallization process2, including cluster radius (r), cluster

surface tension (γ), ideal gas constant (R=8.314 Jmol-1K-1), absolute temperature (T),

molar volume of bulk crystal (Vm). Besides, “S” represents the ratio of solute

concentrations at saturation and equilibrium conditions. According to this, under

supersaturated condition (S>1), “r” increase will lead to “∆G” decrease to stabilize

the clusters, with a nucleus critical radius (r*) determined by * 2ln

γ= mV

rRT S

.

From this, ascending “lnS” will induce descending “r*”, which indicates that upon Ag

atoms concentration exceeding the solubility limit in nano-TiO2, a system will relax

by Ag nanoparticles nucleation and growth.

During Ag PIII process, energetic Ag plasma arriving at nano-TiO2 have sufficient

energy to come to rest under the surface. Since they deliver kinetic and potential

energy, blocking by nano-TiO2 transforms their energy into the atomic scale heating

(ASH), which plays a significant role in Ag clusters evolution3. Here, three

acceleration zones are taken into consideration for cathodic-arc Ag ions, including:

(1) Zone I near cathode spot forms the Ag ion kinetic energy Ek; (2) Zone II in

space-charge sheath between Ag plasma and nano-TiO2 surface endows Ag ion of

charge state Q with the kinetic energy QeVbias (e, elementary charge) by applying

negative bias voltage (Vbias) to nano-TiO2 before impacting the surface. (3) Zone III in

nanoscale vicinity of nano-TiO2 produces image charge acceleration and the kinetic

energy Ei. Meanwhile, Ag ion owns substantial potential energy Ep, determined by the

equation = + +p c e QE E E E , where Ec, Ee, EQ represent cohesive energy, excitation

energy and ionization energy, respectively3. The ionization energy EQ of the above

S-6

triply charged Ag ion is the summation of three ionization steps as 1

0

−

=

=∑Q

i

Q Q

i

E E3. As a

result, the total energy gain of an individual Ag ion during Ag PIII is determined by

the equation, = + + +k bias i pE E QeV E E3. Considering the mean charge state Q≈24

and the mean work function of nano-TiO2 W≈5 eV 5, the kinetic energy Ei was

calculated to 2.55 eV by the equation 1

0

2( ) 1

2 8( ) 2

−

=

− −=

− +∑Q

i

j

W Q jE

Q j

6. According to

references and calculations, the energy gain components of total energy of an

individual Ag ion upon reaching nano-TiO2 were listed in Table 2. From this table,

the total energy gain of an individual Ag ion over a wide bias voltage (Vbias) is much

greater than the minimum displacement energy in the range of 10~40 eV3. This means

that, during Ag PIII, the stationary Ag atoms under nano-TiO2 surface can be

removed and dislodged out of interior. On this basis, the Ag PIII process can be

divided into two cases. Under a relatively low bias voltage, for example, Vbias=0.5/1.0

kV (Table S1), the implanting depth of Ag plasma is shallow under the surface of

nano-TiO2. As a result, Ag nanoparticles can migrate to the surface with ease. In this

case, the longer Ag PIII lasts on nano-TiO2, the more Ag nanoparticles gather on the

surface. Then they gradually undergo Ostwald ripening, coalescence from discrete to

continuous, even to form a nano-thick Ag thin film on nano-TiO2 surface (Figure

13a). The majority of exposed Ag nanoparticles (even Ag thin film) on surface make

it easy to release Ag ions (Ag+) from nano-TiO2 in a physiological pH environment.

So, recently, Mei et al7 designed the Ag-implanted TiO2 nanotubes by low bias

voltage Ag PIII technique (0.5/1.0 kV) that can release Ag+ from surface to kill

bacteria. The second case is to adopt a relatively high bias voltage for Ag PIII, for

instance, 30 kV used here. In this case, for one thing, the Ag nanoparticles have a long

path to migrate out of the interior of nano-TiO2; for another, the majority of Ag

nanoparticles are embedded in the interior of nano-TiO2 beneath the surface, although

some of them can migrate onto the surface in a budding form. As a consequence, Ag

ions cannot be released from the Ag-embedded TiO2 or the release amount is lower

S-7

than 10 ppb8, 9, much less than the minimum bactericidal concentration of 100 ppb10.

All in all, feasible Ag+ leaching can impart a release-killing ability to the

Ag-implanted surface while limited Ag+ releasing can contribute to a contact-killing

property of the Ag-embedded surface via a direct interaction mode, as illustrated in

Figure 13b.

Randomly selected another area, as seen in Figure 13d, it clearly shows a set of

lattice fringes, in which the lattice fringe spacing of 2.04 Å can be indexed as the (200)

plane of Ag nanoparticles8, further evidenced by the corresponding fast Fourier

transform (FFT) electron diffraction pattern of the inset. Moreover, the EDS results

further confirmed the existence of Ag nanoparticles in regions “3” and “4” (Figure

13g). Likewise, the majority of embedded Ag nanoparticles came to eyes in another

random field, as seen in Figure 13e. Together with the lattice fringe spacing of 2.04 Å

of the partially exposed nanoparticle and the corresponding EDS results of regions “5”

and “6”, they further demonstrated the presence and distribution of Ag nanoparticles

within TiO2 nanotubes. More interestingly, in Figure 13f, the twin crystal of Ag

nanoparticles was formed when the interior tiny ones migrating to surface during PIII.

The lattice fringe spacing of 2.35 Å was assigned to the (111) plane of the partially

exposed Ag nanoparticles8, and the EDS analysis of regions “7” and “8” further

evidenced the existence of Ag nanoparticles. As a result, the tiny nanoparticles in dark

color, including the embedded ones and the partially exposed ones, were attributed to

the Ag nanoparticles generated in Ag PIII process. The size development and

distribution of Ag nanoparticles from the inside to the top along the TiO2 nanotubes

accorded well with the above theoretical analysis of Ag nanoparticles formation

during Ag PIII process.

S-8

Table S1. Energy gain of an individual silver ion during PIII.

Ek (eV) QeVbias (eV) Ei (eV) Ec (eV) Ee (eV) EQ (eV) Vbias (kV)

69.00α 0.20E4 2.55 2.95α 0≈ β 29.1α 1.0

69.00α 0.10E4 2.55 2.95α 0≈ β 29.1α 0.5

Notes: α From reference 3, β From reference 11.

S-9

REFERENCES

(1) Stepanov, A. L. Applications of Ion Implantation for Modification of TiO2: A

Review. Rev. Adv. Mater. Sci. 2012, 30, 150-165.

(2) Christian, J. W. The Theory of Transformations in Metals and Alloys. UK:

Elsevier Science Ltd 2002.

(3) Anders, A. Atomic Scale Heating in Cathodic Arc Plasma Deposition. Appl. Phys.

Lett. 2002, 80, 1100-1102.

(4) Brown, I. G. Vacuum Arc Ion Sources. Rev. Sci. Instrum. 1994, 65, 3061-3081.

(5) Xiong, G.; Shao, R.; Droubay, T. C.; Joly, A. G.; Beck, K. M.; Chambers, S. A.;

Hess, W. P. Photoemission Electron Microscopy of TiO2 Anatase Films Embedded

with Rutile Nanocrystals. Adv. Funct. Mater. 2007, 17, 2133-2138.

(6) Burgdörfer, J.; Meyer, F. Image Acceleration of Multiply Charged Ions by

Metallic Surfaces. Phys. Rev. A 1993, 47, R20-R22.

(7) Mei, S.; Wang, H.; Wang, W.; Tong, L.; Pan, H.; Ruan, C.; Ma, Q.; Liu, M.; Yang,

H.; Zhang, L.; Cheng, Y.; Zhang, Y.; Zhao, L.; Chu, P. K. Antibacterial Effects and

Biocompatibility of Titanium Surfaces with Graded Silver Incorporation in Titania

Nanotubes. Biomaterials 2014, 35, 4255-4265.

(8) Huang, R.; Han, Y.; Lu, S. Enhanced Osteoblast Functions and Bactericidal Effect

of Ca and Ag Dual-Ion Implanted Surface Layers on Nanograined Titanium Alloys. J.

Mater. Chem. B 2014, 2, 4531-4543.

(9) Zheng, Y.; Li, J.; Liu, X.; Sun, J. Antimicrobial and Osteogenic Effect of

Ag-Implanted Titanium with a Nanostructured Surface. Int. J. Nanomed. 2011, 7,

875-884.

(10)Monteiro, D. R.; Gorup, L. F.; Takamiya, A. S.; Ruvollo-Filho, A. C.; Camargo, E.

R. d.; Barbosa, D. B. The Growing Importance of Materials That Prevent Microbial

Adhesion: Antimicrobial Effect of Medical Devices Containing Silver. Int. J.

Antimicrob. Agents 2009, 34, 103-110.

(11) Anders, A. Cathodic Arcs: From Fractal Spots to Energetic Condensation. USA:

Springer Science Business Media, LLC 2008.

Antibacterial Surface Design of Titanium-Based Biomaterials ...Antibacterial Surface Design of...

Documents

Transcript of Antibacterial Surface Design of Titanium-Based Biomaterials ...Antibacterial Surface Design of...