Effect of cathodic polarization on coating doxycycline on titanium surfaces

Materials Science and Engineering C 63 (2016) 359–366

Contents lists available at ScienceDirect

Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec

Effect of cathodic polarization on coating doxycycline on titanium surfaces Sebastian Geißler, Hanna Tiainen, Håvard J. Haugen ⁎ Department of Biomaterials, Institute of Clinical Dentistry, University of Oslo, PO box 1109 Blindern, 0317 Oslo, Norway

a r t i c l e

i n f o

Article history: Received 4 December 2015 Received in revised form 29 February 2016 Accepted 1 March 2016 Available online 3 March 2016 Keywords: Titanium Doxycycline Cathodic polarisation Surface modification Antibacterial

a b s t r a c t Cathodic polarization has been reported to enhance the ability of titanium based implant materials to interact with biomolecules by forming titanium hydride at the outermost surface layer. Although this hydride layer has recently been suggested to allow the immobilization of the broad spectrum antibiotic doxycycline on titanium surfaces, the involvement of hydride in binding the biomolecule onto titanium remains poorly understood. To gain better understanding of the influence this immobilization process has on titanium surfaces, mirror-polished commercially pure titanium surfaces were cathodically polarized in the presence of doxycycline and the modified surfaces were thoroughly characterized using atomic force microscopy, electron microscopy, secondary ion mass spectrometry, and angle-resolved X-ray spectroscopy. We demonstrated that no hydride was created during the polarization process. Doxycycline was found to be attached to an oxide layer that was modified during the electrochemical process. A bacterial assay using bioluminescent Staphylococcus epidermidis Xen43 showed the ability of the coating to reduce bacterial colonization and planktonic bacterial growth. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Titanium based materials are the prevalent choice for boneanchored implants in orthopedic and dental applications. The reason for this is the well-known biocompatibility of titanium which originates from the combination of good mechanical properties, high corrosion resistance and biostability [1]. The latter two factors can be attributed to the presence of a stable native oxide layer that covers the surface making the material passive and therefore minimizing the host response [2]. However, such passive implant surfaces are susceptible to bacterial colonization [3]. The often resulting biomaterial-associated infections represent a significant risk to the establishment of host tissue integration which is crucial for the intended function of the implant [4]. A variety of surface modifications has therefore been explored in order to make the passive implant surfaces more interactive with the host tissue, and by this, influencing the bone-to-implant interface. Beside the established techniques to enhance osseointegration (e.g. grit-blasting and acid-etching, anodic oxidation, calcium phosphate coatings), biochemical methods have become a popular tool to actively influence biological processes [5,6]. The main principle of these methods is the immobilization of specific bioactive molecules such as extracellular matrix proteins, RGD-containing peptides, growth factors or antimicrobial agents onto the implant surface [5,7–12]. A covalent chemical binding of such biomolecules is mainly obtained by an ⁎ Corresponding author. E-mail address: [email protected] (H.J. Haugen).

http://dx.doi.org/10.1016/j.msec.2016.03.012 0928-4931/© 2016 Elsevier B.V. All rights reserved.

activation of the surface oxide followed by the actual attachment of the molecules [8]. One method for activating the surfaces of titanium implants was proposed by Videm et al. who reported the formation of a titanium hydride (TiH2) layer by applying cathodic polarization in organic acids [13]. This hydride layer has been shown to increase bone attachment to implants in vivo [14]. The same process has also been applied on dental abutments revealing positive effects on human gingival fibroblast proliferation [15]. Furthermore, the higher reactivity of the hydride layer has been suggested to be beneficial for attaching biomolecules onto the surface of an implant [13]. In a study by Frank et al., cathodic polarization was successfully used to bind enamel matrix derivate (EMD) onto titanium based materials in order to improve bone regeneration around implants [16]. A further approach explored the suitability of this polarization process to coat titanium-zirconium surfaces with the broad spectrum antibiotic doxycycline [17]. However, the surfaces used in this study were grit-blasted and acid-etched, resulting in very rough surfaces which already contain titanium hydride in their subsurface layer as shown by Szmukler-Moncler et al [18]. This complexity of the initial surface restricts the characterization of the modified surface properties as the effect of the modification process may be overshadowed by the original surface features. Furthermore, the complex surface morphology and high roughness limits the use of many surface specific characterization techniques that can only be applied on surfaces with low to moderate roughness. The aim of the present study was to investigate the effect of the polarization process previously reported by Walter et al. [17] on the

360

S. Geißler et al. / Materials Science and Engineering C 63 (2016) 359–366

chemical surface properties of titanium. The main objective was to examine the potential role of hydride formation in binding doxycycline onto titanium surfaces during this polarization process. Using a variety of characterization techniques on mirror-polished surfaces, we assessed the events occurring during the polarization process and their effect on the biomolecule coating. 2. Materials and methods 2.1. Sample preparation Grade IV commercially pure titanium (Ti) was used in this study. Disc-shaped samples with a diameter of 6.2 mm and a height of 2 mm were mirror-polished and washed as described before [19]. For the doxycycline coating, the same cathodic polarization process as described by Walter et al. was used [17]. Briefly, the samples were cathodically polarized in a 2 M acetic acid/sodium acetate buffer (Sigma Aldrich, Oslo, Norway) with a pH of 3 containing 1 mg/ml doxycycline hyclate (AK Scientific Inc., Union City, CA, USA). A current density of 1 mA/cm2 was applied for 3 h. After the process, the samples were rinsed with deionized (DI) water and dried with nitrogen gas. Unmodified samples were used as a first control group (“Control”). Samples polarized without doxycycline served as a second control group (“Polarized”). A release study was performed by immersing doxycycline coated samples in DI water for 10 h (“Doxycycline imm.”). 2.2. Field emission scanning electron microscopy Images of the surface micro- and nanostructure were attained using a field emission scanning electron microscope (FE-SEM; S-4800, Hitachi, Tokyo, Japan). The samples were sputter-coated with platinum (Cressington 308R, Watford, UK) prior to imaging. Acceleration voltage was set to 5 kV and the samples were imaged at a working distance between 8.6 mm and 8.8 mm. 2.3. Atomic force microscopy Further surface characterization and visualization was conducted using an atomic force microscope (AFM; MFP-3D-SA, Asylum Research, Santa Barbara, CA, USA) in combination with OMCL-AC240TS cantilevers (Olympus Corporation, Tokyo, Japan). Scanning mode (contact or AC), scanning rate, set point and integral gain were adjusted accordingly in order to obtain a good following of the trace and retrace signal. Surface parameters were measured on 5 spots per sample (10 μm × 10 μm) and 3 samples per group (n = 15). Obtained datasets were analyzed using the open source software Gwyddion. Height and amplitude signal graphs were overlaid for an optimal visualization of the surfaces. Furthermore, an area measuring 2.5 μm × 2.5 μm was scanned for obtaining three-dimensional (3D) images of the surfaces. 2.4. Transmission electron microscopy Transmission electron microscopy (TEM) was used to visualize the surfaces in cross section. The samples were prepared by two different routes. The doxycycline coated sample was prepared by a Helios NanoLab DualBeam focused ion beam scanning electron microscope (FIB-SEM, FEI, Hillsboro, OR, USA). First, a platinum protection layer was deposited with electron beam assisted deposition to avoid Ga+ ion beam damage of the sample surface. Ga+ ion beam assisted deposition was then used to deposit further platinum and carbon protection layers. After this, the sample was thinned at 30 kV ion beam acceleration voltage, followed by a final thinning step at 5 kV to minimize surface damage of the TEM sample. The control sample and the polarized sample were prepared by mechanical tripod wedge polishing as described by Eberg et al. down to a thickness of 5–10 μm [20]. The final thinning was performed by FIB-SEM as described above. TEM was performed

on a double Cs corrected (probe- and image-corrected) cold-FEG JEOL ARM200F (JEOL, Tokyo, Japan) operated at 200 kV. 2.5. Fluorescence microscopy A confocal laser scanning microscope (Leica TCS SP8, Leica Microsystems, Wetzlar, Germany) equipped with a 10 × Leica HC PL APO objective was used for imaging the sample surfaces. Fluorescence images were obtained using an excitation wavelength of 405 nm while the signal was detected in the range between 440 nm and 560 nm. 2.6. X-ray photoelectron spectroscopy The elemental composition on the sample surfaces was analyzed by X-ray photoelectron spectroscopy (XPS; Axis UltraDLD XP spectrometer, Kratos Analytical, Manchester, UK) using monochromatic Al Kα X-rays (hν = 1486.69 eV). The measurement area was 300 μm × 700 μm. Survey spectra were recorded between 1100 eV and 0 eV binding energy. Detail spectra were recorded in the energy regions of Ti2p, O1s and C1s. Angle-resolved XPS was performed on a Thetaprobe XP spectrometer (Thermo Scientific, Waltham, MA, USA) using monochromatic Al Kα radiation (hν = 1486.69 eV). The spectra were acquired in parallel angle-resolved mode over an analysis area of approximately 400 μm in diameter. Spectra analysis was conducted using CasaXPS (Casa Software Ltd, Teignmouth, UK). 2.7. Secondary ion mass spectrometry The depth profiles of the isotopes 1H and 12C were measured by secondary ion mass spectrometry (SIMS; IMS 7f, CAMECA, Gennevilliers, France). The samples were bombarded by a 15 keV primary ion beam (Cs+) and emitted secondary ions were collected from the central part (67 μm × 67 μm) of a crater measuring 150 μm × 150 μm. The isotope concentration distribution was indicated by count intensity (c/s) of secondary ions as a function of sputter time. 2.8. Fourier transform infrared spectroscopy A PerkinElmer Spectrum 400 FT-IR/FT-NIR spectrometer (PerkinElmer, Waltham, MA, USA) was used to investigate the presence of doxycycline on the sample surface. Samples were measured with a diffuse reflectance accessory in the mid infrared range between 4000 cm−1 and 450 cm−1. Per sample, 16 scans with a resolution of 4 cm− 1 were conducted. The background spectrum was collected from an unmodified sample. 2.9. Bacteria assay The antibacterial properties of the modified Ti surfaces were assessed by culturing bioluminescent Staphylococcus epidermidis Xen43 on non-polished polarized and doxycycline coated discs. Doxycycline samples were tested both immediately after coating and after immersion in DI water for 24 h. S. epidermidis Xen43 were cultured in tryptic soy broth (TSB) at 37 °C in aerobic atmosphere. When optical density (OD600) reached 0.4–0.5, the bacteria were diluted 1:100 in TBS and incubated overnight. Sample discs (n = 16) were placed in 96-well opaque microplates (OptiPlate-96, PerkinElmer, Waltham, USA) and inoculated with 150 μl of the overnight bacteria suspension diluted 1:100 in TSB (OD600 = ~ 0.05). The well plates were sealed with transparent adhesive seals (TopSeal™ A-Plus, PerkinElmer, Waltham, MA, USA) and incubated at 37 °C in a multi-detection plate reader (Synergy HT, BioTek, Winooski, VT, USA). Luminescence was measured every 15 min for 16 h, after which the sample discs were removed from the wells and prepared for SEM imaging. Bacteria on the

S. Geißler et al. / Materials Science and Engineering C 63 (2016) 359–366

sample surfaces were fixed with 2.5% glutaraldehyde in 0.1 M Sørensen phosphate buffer. The samples were then rinsed with DI water, dehydrated with 30% ethanol, sputtered with platinum and examined by SEM (TM-3030, Hitachi, Tokyo, Japan). Additional colour overlaying on SEM micrographs was performed with Adobe Photoshop CS6 to highlight the adhered bacteria on the Ti surfaces. 2.10. Statistical analyses Topographical parameters obtained by AFM were compared between the different groups using a One-Way ANOVA in SigmaPlot 13.0 (Systat Software, San José, CA, USA). Pairwise multiple comparisons were performed using the Holm-Sidak method. Values are presented as mean ± standard deviation. 3. Results and discussion In this study, we investigated the effect of the cathodic polarization process previously reported by Walter et al. [17] on the chemical surface properties of titanium and the potential role of hydride formation in binding doxycycline onto titanium surfaces. Mirror-polished commercially pure titanium surfaces were chosen to enable the use of surfacespecific analysis techniques, and to avoid misinterpretation due to surface roughness and presence of pre-existing titanium hydride on the sample surface. Visualizing the surfaces is often the first step in a characterization procedure because it gives a quick first overview over changes induced by a surface modification. In the present study, we applied the two most commonly used methods, SEM and AFM, to obtain images of our titanium surfaces (Fig. 1). The mirror-polished control sample exhibited a

361

smooth surface with some defects which most likely originated from the polishing process. After 3 h of polarization, the surface was covered by spherical structures with a diameter of 20–30 nm. Interestingly, these spheres often seemed to be aligned along small scratches on the surface (Fig. SI 1). Already Frank et al. detected the formation of nanostructures on titanium and titanium-zirconium polarized for 1 h at similar current densities [21]. They correlated these findings with the observations made by Videm et al. who saw the creation of titanium hydride nodules during cathodic polarization [13]. Those observations, however, were made on surfaces polarized with a higher current density (6 mA/cm2) and for much longer time (over 20 h) in a different buffer system. Comparing the AFM images of our polarized sample with the ones made by Videm et al., the surface structures they obtained were quite different both in shape and size [13]. Generally, the images obtained by AFM were comparable to the SEM images. However, more details of the actual surface were visible, e.g. small structures were distinguishable within the defects on the control surface. This was not surprising since the AFM scans were conducted with the cantilever in direct contact to the surface while the electrons in the SEM have a higher interaction volume within the material. Moreover, AFM allows for the readout of surface-specific parameters such as roughness values. The polarization was not found to alter the surface roughness significantly with Sa values of 3.00 ± 0.56 nm compared to 2.75 ± 0.72 nm on the control surface. The appearance of the surface changed considerably when doxycycline was added to the buffer during polarization. Distinct micro- and nano-scale structures were observed and large agglomerates were irregularly distributed over the surface. Tilting the sample by 52° in the SEM revealed that these micro- and nano-scale structures were sticking out from the surface (Fig. SI 2). AFM scans in contact mode

Fig. 1. Visualization of the different surfaces by SEM and AFM. 3D images (2.5 × 2.5 μm2) were created using the data obtained by AFM. Note the different colour index scale on the 3D image of the doxycycline sample.

362

S. Geißler et al. / Materials Science and Engineering C 63 (2016) 359–366

did not lead to satisfying results due to the more complex surface structure and increased roughness. Hence, these surfaces were scanned in AC mode. Obtained Sa values confirmed a significant increase in roughness compared to the two other surfaces (29.13 ± 12.10 nm, p ˂ 0.001). A qualitative overview over the analyzed samples is given by means of 3D visualization of the surfaces in Fig. 1. Drawing a first conclusion, it was clear that the polarization itself had an influence on the surface nano-structure and that the surface was completely altered when doxycycline was present in the process. On the basis of these results, we were able to specify the research aims to finding answers to the following questions: (i) Which changes does the polarization process induce to the outermost surface layer? What are the observed spherical structures composed of? As stated in the introduction, the idea of the polarization process was to create a reactive hydride layer which is then used to attach the biomolecule. (ii) What are the observed micro- and nano-scale structures on the doxycycline sample?

For this reason, cross sections of the surface-near areas were prepared and imaged by TEM. The control sample was covered by a homogeneous oxide layer with a thickness of approximately 5 nm (Fig. 2A and B). Angle-resolved XPS revealed that the outermost layer of this oxide consisted mainly of TiO2 changing gradually into Ti2O3 before reaching the bulk Ti metal (Fig. 3A). This gradual change of oxidation states is characteristic of naturally grown Ti oxide layers [2]. Videm et al. detected first hydride growth after approximately 1 h of polarization [13]. However, we could not observe any hydride formation on the sample polarized for 3 h. Instead, an irregular growth of the oxide layer and the formation of oxygen-rich pockets were seen (Fig. 2C and D). Moreover, the spherical structures seen on the SEM and AFM images

were not detected. We suppose therefore that they were removed during the preparation of the TEM sections. However, taking the observed growth of the oxide layer into consideration, as well as the fact that the spheres were often aligned along surface defects, it is likely that these structures were composed of Ti oxide formed during the polarization, preferably at defect sites within the existing oxide layer. This assumption was supported by the results obtained by angle-resolved XPS which showed an increase in the TiO2 layer thickness on the polarized sample compared to the control sample (Fig. 3B). The formation of oxide nanostructures on modified titanium and titanium alloy surfaces has previously been observed by Wennerberg et al [22]. They reported that liquid storage of acid-etched samples resulted in a reorganization of the outermost oxide layer and the formation of well defined TiO2 nanostructures. Thus, it seemed like the polarization process with the used conditions had a similar but accelerated effect as the liquid storage. In other studies, the evaluation of hydride formation was mainly based on 1H isotope depth profiles obtained by SIMS. Neither Frank et al. nor Walter et al. detected significantly elevated hydrogen levels after the process [17,21]. However, those observations were made on acid-etched surfaces which already exhibited increased hydrogen contents [18]. The authors suggested therefore a possible remodeling of the hydride layer. In contrast, Xing et al. saw increased hydrogen levels after polarizing polished Ti surfaces in different organic acids [15]. The largest differences, however, were observed for high current densities (15 mA/cm2) and polarization times of 5 h. Furthermore, elevated hydrogen contents were reported after using the process for coating titanium based materials with EMD [16]. However, these samples were not compared to surfaces that were only polarized without EMD. Thus, no conclusions about the impact of the polarization process itself can be made. When we performed the same SIMS measurements on our samples, the hydrogen profile of the polarized sample was comparable to the profile of an unmodified sample (Fig. 4). This confirmed the

Fig. 2. Cross-sectional TEM images of the outermost surface layers. A) Pt and C protection layers were deposited to avoid ion beam damage during sample preparation. B) The unmodified sample (Control) was covered by a homogeneous 4–5 nm thick oxide layer. C) The polarized sample exhibited a thick inhomogeneous oxide layer. The highlighted area (red box) is magnified in D). E) After the doxycycline coating, an organic layer was observed between the bulk material and the Pt protection layer. F) When magnifying the highlighted area in E), it could be seen that the organic layer was on top of an oxide layer.

S. Geißler et al. / Materials Science and Engineering C 63 (2016) 359–366

Fig. 3. Composition of the first 8 nm of the surface as determined by angle-resolved XPS for the control sample (A) and the polarized sample (B). The native oxide layer on the control sample exhibited a gradual change from TiO2 to titanium metal. The polarized sample showed an increased thickness of the TiO2 layer. Some organic contamination was observed on both samples. The detected silicon and copper residues were likely to come from the silica polishing solution and the copper clips used in the polarization setup.

preceding observations that the polarization did not induce hydrogen absorption into the material. The reason for the growth of oxide instead of hydride remains unclear. In theory, hydrogen ions in the acidic electrolyte adsorb to the negatively charged Ti surface during cathodic polarization and begin to dissolve the passivating oxide layer covering the surface [23]. Once

363

the surface becomes free of the passivating oxide film, hydrogen absorption into the material occurs rapidly until the material surface is completely covered by hydride [24]. However, the polarization conditions play a significant role in the dissolution of the passivating oxide layer as well as the hydrogen absorption reaction itself. While Videm et al. [13] did not see a pH influence on hydride formation in the range from 1.3 to 5.5, other authors have detected a clear pH dependence. Phillips et al. stated that increasing pH values of the solutions significantly decreased the rate of hydrogen absorption [25]. Solutions with a pH of 3, as used in our study, resulted in only limited absorption. Further studies have shown that for solutions with a pH of 3 or higher, the surfaces are likely to be passivated [24,26]. Since the samples in the study by Walter et al. [17] and in our study had not been pickled prior to the polarization process to remove the native oxide layer, this passivating oxide may have hindered reactions at the titanium cathode. Moreover, hydride formation on cathodically polarized titanium surfaces is dependent on both current density and polarization time [13,15,27]. The used parameters for attaching biomolecules to the surfaces (1 mA/cm2 and 3 h in the present study; 0.49–1.65 mA/cm2 and 1 h [16]; 0.88 mA/cm2 and 75 min [17]) might have been too low for an actual hydride build-up. However, degradation of the doxycycline molecule was already observed during the polarization process at 1 mA/cm2 current density used in the present study (Supporting information). Thus, altering the polarization parameters to create titanium hydride by either increasing the current density or polarization time can be expected to result in further unwanted degradation of the biomolecule. In addition, the polarization was performed at room temperature and not at elevated temperatures [13]. This might also have influenced occurring reactions since increasing temperature even from 25 °C to 40 °C has been shown to significantly increase hydrogen adsorption into titanium [25]. Furthermore, the compartments for anode and cathode were not separated in this setup. However, in studies where hydride build-up was observed, the compartments were always separated and in some cases the anode compartment was even deoxygenated [13,24,25,27]. Thus, we believe that oxygen originating from the anode could have interfered with the hydrogen evolution reaction at the cathode [27]. In summary, the presence of a passivating oxide layer which could not be removed by the mildly acidic conditions, short polarization times and low current densities, low solution temperatures and the fact that the electrode compartments were neither separated nor deoxygenated, are factors which in combination can have hindered the formation of titanium hydride. Knowing that the build-up of a titanium hydride layer could not have been involved in the attachment of the biomolecules onto the

Fig. 4. Intensity profiles of 1H and 12C for the different treatment groups as determined by SIMS.

364

S. Geißler et al. / Materials Science and Engineering C 63 (2016) 359–366

Table 1 XPS element distribution of titanium, oxygen, carbon and nitrogen. Element

Polarized (at.%)

Doxycycline (at.%)

Doxycycline imm. (at.%)

Ti2p O1s C1s N1s

14.37 47.86 30.25 0.00

1.22 23.84 69.20 4.80

3.88 27.32 63.93 2.69

surfaces, the next step was to investigate what had happened to doxycycline during the process. By means of the cross-sectional images, the structures observed by SEM and AFM on the doxycycline sample could be identified as an organic layer covering the surface (Fig. 2E and F). The thickness of this organic layer was not homogeneous varying from only few nanometers up to 50 nm. Since no hydride formation was seen, the elevated 1H intensity values detected on the doxycycline sample by SIMS (Fig. 4) must therefore have originated from the organic layer. This could also explain the increased hydrogen content observed by Frank et al. on EMD coated surfaces [16]. The high 12C isotope intensity on the doxycycline sample proved the organic nature of the covering layer. Similar oxygen-rich areas as observed on the polarized surface were found in between the bulk material and the organic layer and oxygen seemed to have diffused along grain-boundaries (Fig. SI 3). Due to the rough and inhomogeneous character of the surface, it was not possible to get reasonable results by angle-resolved XPS for these samples. For this reason, the elemental composition of the surfaces was determined by standard XPS. For all surfaces, mainly titanium, oxygen, carbon and nitrogen were detected (Table 1). While the Ti2p signal was reduced over 90% on the doxycycline sample compared to the polarized sample, a more than two-fold increase in the C1s signal was seen. This result again confirmed the covering of the titanium substrate by an organic, carbon rich layer. Furthermore, the detection of nitrogen on the doxycycline sample indicated the presence of doxycycline molecules on the surface. Looking at the XPS detail spectra (Fig. 5) and the peak compositions (Table SI 1), the surface of the polarized sample consisted mainly of TiO2, TiOH and adventitious carbon contamination. The rather high carbon signal (30.25 at.%) is likely to come from the acetate buffer used in the polarization process. In contrast, the oxygen signal on the doxycycline sample derived primarily from organic compounds such as\\C_O and \\C\\OH. Combined with the higher fraction of phenyl carbon, this was a sign for the presence of doxycycline within the organic layer. The XPS results were in agreement with the findings on considerably rougher surfaces [17]. Walter et al. furthermore confirmed the presence and intactness of doxycycline on the surfaces by FTIR [17]. When we applied the same method to our doxycycline sample, we

could only detect very weak signals with a maximum band intensity of b2% transmittance (Fig. SI 4). It was impossible to get any useful information about the chemical composition from such spectra. As a simple test for the stability of the organic layer, the doxycycline sample was immersed in DI water for 10 h and XPS analyses were repeated. The increase in the Ti2p and O1s signal and the decrease in the C1s and N1s signal compared to the non-immersed sample indicated that parts of the organic layer were removed (Table 1). This was also seen in the XPS detail spectra (Fig. 5) where an increased TiO2 peak and increased TiO2 contribution to the oxygen peak were observed. Moreover, the intensity of \\C\\OH contribution to the carbon signal was decreased. However, the chemical composition of the immersed sample was still different from the polarized sample indicating a remaining organic layer on the surface. To further confirm this observation, the doxycycline sample was imaged by fluorescence microscopy before and after the water immersion. Tetracyclines such as doxycycline are known to produce weak fluorescence [28]. Indeed, we were able to observe fluorescent structures on the surfaces which completely vanished after the immersion (images not shown). Similar observations have been made by Xing et al. who coated dental abutment materials with doxycycline using cathodic polarization [29]. They reported a burst release of doxycycline in phosphate-buffered saline (PBS) at 37 °C within the first 24 h. After this initial release, doxycycline was found to remain present at the surface for up to 2 weeks (as determined by the presence of nitrogen on the surface) [29]. However, the long-term antibacterial effect of this remaining coating has not been tested. All the above obtained results indicate that even though there was no hydride formed by the used polarization, and hydride was therefore not involved in the binding of doxycycline, there was still an organic layer containing the biomolecules attached to the surface. However, it seems like the active compounds of this coating were only weakly bound to the surface and were instantly released by simply immersing the samples in water. The described long-term effect of the doxycycline coating on bone formation [17] and a potential long-term antibacterial effect [29] must therefore be regarded critically. To test our hypothesis about the limited long-term availability of doxycycline on the surfaces, the antibacterial effect of the samples was investigated before and after immersion by culturing S. epidermidis Xen43 on the sample surfaces. S. epidermidis is a biofilm-forming bacterial strain that has been shown to form pathogenic biofilms on implant surfaces in vivo [30–32]. Together with Staphylococcus aureus, S. epidermidis is the most common cause of biofilm-associated infections [30]. Using a bioluminescent modification of S. epidermidis, it was possible to obtain information about the metabolic activity of the bacteria by detecting the emission of light [33]. Representative luminescence

Fig. 5. XPS detail spectra of Ti2p, O1s and C1s. The doxycycline sample (green) is compared to a polarized sample (red) and a doxycycline sample which was immersed for 10 h in DI water (blue, doxycycline imm.).

S. Geißler et al. / Materials Science and Engineering C 63 (2016) 359–366

Fig. 6. Representative luminescence curves obtained for the three different groups during 16 h of incubation at 37 °C.

profiles over a time period of 16 h are shown in Fig. 6. An initial peak in luminescence was observed for all samples and all different groups, including the positive control (bacteria cultured in empty wells). Since this high luminescence level in the beginning has also been seen by other authors and has been assigned to the luminescence of Lux proteins formed before the actual incubation, it was not included in the analysis [34]. For further analysis, we chose a luminescence value of 20,000 as a threshold for bacterial contamination since all samples with values above this threshold exhibited turbid media on top of the samples, a sign for high amounts of bacteria in the media. The observed luminescence originated therefore mainly from the growth of planktonic bacteria in the bacteria suspension. All polarized samples exhibited the largest increase in luminescence reaching the maximum after approximately 4 h. After that, a decrease in luminescence was observed when the bacteria reached the stationary phase. This decrease was likely to be caused by the lack of oxygen and nutrients in the media since the well plates were sealed and there was no media exchange during the 16 h of incubation, resulting in a decrease in metabolic activity [35]. However, once luminescence exceeded a value of 20,000, the sample was regarded contaminated even though luminescence values again went below this threshold afterwards.

365

Using this classification (Table SI 2), all polarized samples were contaminated already after 4 h of incubation. In contrast, the doxycycline coated samples could inhibit the growth of bacteria inoculated on the sample surface as luminescence remained below 20,000 throughout the 16 h of incubation. Doxycycline is known to be taken up by bacteria and to inhibit protein biosynthesis [36]. Therefore, doxycycline must have been released from the surfaces and have interfered with bacterial growth resulting in low luminescence values. The doxycycline coated samples that had been immersed in DI water prior to bacteria inoculation showed a clear reduction in the antibacterial effect as seen by the large increase in luminescence (N 20,000) for the majority of the samples. After 16 h of incubation, 13 out of 16 samples were contaminated (Table SI 2). This proved that most of the functional doxycycline had been released during the 24 h water immersion. The remaining organic coating, however, was still able to curb the bacteria growth in comparison to the polarized samples as seen by the lower and broader luminescence peak indicating a slower growth rate. Hence, there were either still low amounts of doxycycline coming off the surface or some bacteria must have been killed when coming in contact with the surface. Finally, the different surfaces were examined for bacterial contamination by SEM. Interconnected clusters of bacteria were found on the surfaces of the polarized samples (Fig. 7 A and B). Assuming that sufficient oxygen and nutrients would be provided over a longer period of time, the surface would probably end up being completely covered by a biofilm. In contrast, the amount of bacteria was significantly reduced on the doxycycline samples (Fig. 7 C and D). Some bacteria exhibited a collapsed shape indicating that they were killed on contact with the surface. The size of the bacteria was generally larger compared to the ones on the polarized surfaces. This could be a sign for unsuccessful attempts to proliferate and a resulting stress response. Considerably fewer bacteria compared to the polarized samples were also observed on the immersed doxycycline surfaces (Fig. 7 E and F). While the surfaces still seemed to resist biofilm formation to some extent, no dead bacteria were seen. Combined with the observed increased growth of planktonic bacteria in the media, these surfaces were clearly less effective in their antibacterial activity compared to the doxycycline samples. Since the antibacterial properties of the coating appeared to be mostly related to the doxycycline molecules released from the surface, the bactericidal effect of the doxycycline coating would be limited to the initial burst release period. Although the remaining adherent doxycycline layer showed some potential in curbing biofilm formation (Fig. 7 E and F),

Fig. 7. Low and high magnification SEM images of S. epidermidis Xen43 on the surfaces of the different groups (all scale bars 10 μm). A) and B) Polarized. C) and D) Doxycycline. E) and F) Doxycycline imm.

366

S. Geißler et al. / Materials Science and Engineering C 63 (2016) 359–366

the antibacterial effect can be expected to be further reduced in vivo as proteins from physiological fluids such as blood or saliva will adsorb onto the implant surface, and therefore bacteria are less likely to be in contact with the strongly adherent doxycycline molecules. 4. Conclusions In summary, it was shown that the cathodic polarization process, which has previously been used to attach biomolecules to implant surfaces, was not effective in creating a hydride layer on mirror-polished commercially pure titanium surfaces. Applying a current density of 1 mA/cm2 for a time period of 3 h resulted only in a modification of the existing titanium oxide layer. Even though it was demonstrated that titanium hydride was not involved in biomolecule binding processes, doxycycline was found to be adhered to the remodeled oxide layer. The antibacterial effect of this coating was investigated by a new method where bioluminescent S. epidermidis Xen43 were cultured on the sample surfaces. The doxycycline modified surfaces exhibited reduced bacterial colonization and successfully inhibited growth of planktonic bacteria. However, this antibacterial effect was only of short duration as most of the active molecules were released within the first 24 h. The remaining strongly adhered coating showed a clear reduction in antibacterial efficiency. Acknowledgement The authors acknowledge Martin F. Sunding, Lord Wengenroth, Per Erik Vullum, and Susana Villa Gonzalez for their technical assistance with XPS, AFM, FIB-TEM, and LC-MS/MS characterization, and Jessica Lönn-Stensrud for the valuable advice on the bacterial assay. This work was supported by the Research Council of Norway (grant 230258) and by the NORTEM project (grant 197405). Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.msec.2016.03.012. References [1] D.F. Williams, Titanium for Medical Applications, in: D.M. Brunette, P. Tengvall, M. Textor, P. Thomsen (Eds.), Titanium in Medicine, Springer-Verlag, Berlin Heidelberg 2001, pp. 13–24. [2] M. Textor, C. Sittig, V. Frauchiger, S. Tosatti, D.M. Brunette, Properties and Biological Significance of Natural Oxide Films on Titanium and its Alloys, in: D.M. Brunette, P. Tengvall, M. Textor, P. Thomsen (Eds.), Titanium in Medicine, Springer-Verlag, Berlin Heidelberg 2001, pp. 171–230. [3] J.W. Costerton, P.S. Stewart, E.P. Greenberg, Bacterial biofilms: a common cause of persistent infections, Science 284 (1999) 1318–1322. [4] H.J. Busscher, H.C. van der Mei, G. Subbiahdoss, P.C. Jutte, J.J. van den Dungen, S.A. Zaat, M.J. Schultz, D.W. Grainger, Biomaterial-associated infection: locating the finish line in the race for the surface, Sci. Transl. Med. 4 (2012) 153rv110. [5] X. Chen, Y. Li, C. Aparicio, Biofunctional Coatings for Dental Implants, in: S. Nazarpour (Ed.), Thin Films and Coatings in Biology, Springer, Dordrecht Heidelberg New York London 2013, pp. 105–143. [6] J. Lausmaa, Mechanical, Thermal, Chemical and Electrochemical Surface Treatment of Titanium, in: D.M. Brunette, P. Tengvall, M. Textor, P. Thomsen (Eds.), Titanium in Medicine, Springer-Verlag, Berlin Heidelberg 2001, pp. 231–266. [7] D.A. Puleo, A. Nanci, Understanding and controlling the bone-implant interface, Biomaterials 20 (1999) 2311–2321. [8] S.J. Xiao, G. Kenausis, M. Textor, Biochemical Modification of Titanium Surfaces, in: D.M. Brunette, P. Tengvall, M. Textor, P. Thomsen (Eds.), Titanium in Medicine, Springer-Verlag, Berlin Heidelberg 2001, pp. 417–455. [9] S.J. Xiao, M. Textor, N.D. Spencer, H. Sigrist, Covalent attachment of cell-adhesive, (Arg-Gly-Asp)-containing peptides to titanium surfaces, Langmuir 14 (1998) 5507–5516.

[10] M. Morra, Biochemical modification of titanium surfaces: peptides and ECM proteins, Eur. Cell. Mater. 12 (2006) 1–15. [11] L.Z. Zhao, P.K. Chu, Y.M. Zhang, Z.F. Wu, Antibacterial coatings on titanium implants, J. Biomed. Mater. Res. B 91B (2009) 470–480. [12] F. Costa, I.F. Carvalho, R.C. Montelaro, P. Gomes, M.C.L. Martins, Covalent immobilization of antimicrobial peptides (AMPs) onto biomaterial surfaces, Acta Biomater. 7 (2011) 1431–1440. [13] K. Videm, S. Lamolle, M. Monjo, J.E. Ellingsen, S.P. Lyngstadaas, H.J. Haugen, Hydride formation on titanium surfaces by cathodic polarization, Appl. Surf. Sci. 255 (2008) 3011–3015. [14] J.E. Ellingsen, K. Videm, L. Opsahl, H.J. Ronold, Implants with Modified Surfaces for Increased Biocompatibility, and Method for Production thereof, Astra Tech AB, 2003. [15] R. Xing, L. Salou, S. Taxt-Lamolle, J.E. Reseland, S.P. Lyngstadaas, H.J. Haugen, Surface hydride on titanium by cathodic polarization promotes human gingival fibroblast growth, J. Biomed. Mater. Res. A 102 (2014) 1389–1398. [16] M.J. Frank, M.S. Walter, M. Rubert, B. Thiede, M. Monjo, J.E. Reseland, H.J. Haugen, S.P. Lyngstadaas, Cathodic polarization coats titanium based implant materials with enamel matrix derivate (EMD), Materials 7 (2014) 2210–2228. [17] M.S. Walter, M.J. Frank, M. Satue, M. Monjo, H.J. Ronold, S.P. Lyngstadaas, H.J. Haugen, Bioactive implant surface with electrochemically bound doxycycline promotes bone formation markers in vitro and in vivo, Dent. Mater. 30 (2014) 200–214. [18] S. Szmukler-Moncler, M. Bischof, R. Nedir, M. Ermrich, Titanium hydride and hydrogen concentration in acid-etched commercially pure titanium and titanium alloy implants: a comparative analysis of five implant systems, Clin. Oral Implants Res. 21 (2010) 944–950. [19] S.F. Lamolle, M. Monjo, S.P. Lyngstadaas, J.E. Ellingsen, H.J. Haugen, Titanium implant surface modification by cathodic reduction in hydrofluoric acid: surface characterization and in vivo performance, J. Biomed. Mater. Res. A 88A (2009) 581–588. [20] E. Eberg, A.F. Monsen, T. Tybell, A.T.J. van Helvoort, R. Holmestad, Comparison of TEM specimen preparation of perovskite thin films by tripod polishing and conventional ion milling, J. Electron Microsc. 57 (2008) 175–179. [21] M.J. Frank, M.S. Walter, M.M. Bucko, E. Pamula, S.P. Lyngstadaas, H.J. Haugen, Polarization of modified titanium and titanium-zirconium creates nano-structures while hydride formation is modulated, Appl. Surf. Sci. 282 (2013) 7–16. [22] A. Wennerberg, L.M. Svanborg, S. Berner, M. Andersson, Spontaneously formed nanostructures on titanium surfaces, Clin. Oral Implants Res. 24 (2013) 203–209. [23] Y. Zeng, J.J. Noel, P.R. Norton, D.W. Shoesmith, Hydrogen transport through thin titanium oxides, J. Electroanal. Chem. 649 (2010) 277–285. [24] L. Yan, S. Ramamurthy, J.J. Noel, D.W. Shoesmith, Hydrogen absorption into alpha titanium in acidic solutions, Electrochim. Acta 52 (2006) 1169–1181. [25] I.I. Phillips, P. Poole, L.L. Shreir, Hydride formation during cathodic polarization of Ti.2. Effect of temperature and pH of solution on hydride growth, Corros. Sci. 14 (1974) 533–542. [26] T. Okada, Factors influencing the cathodic charging efficiency of hydrogen by modified titanium electrodes, Electrochim. Acta 28 (1983) 1113–1120. [27] I.I. Phillips, P. Poole, L.L. Shreir, Hydride formation during cathodic polarization of Ti.1. Effect of current-density on kinetics of growth and composition of hydride, Corros. Sci. 12 (1972) 855–866. [28] H.-T. Lu, Y. Jiang, H.-B. Li, F. Chen, M.-H. Wong, Simultaneous determination of oxytetracycline, doxycycline, tetracycline and chlortetracycline in tetracycline antibiotics by high-performance liquid chromatography with fluorescence detection, Chromatographia 60 (2004) 259–264. [29] R. Xing, I.L. Witsø, D. Jugowiec, H. Tiainen, M. Shabestari, S.P. Lyngstadaas, J. Lönn-Stensrud, H.J. Haugen, Antibacterial effect of doxycycline-coated dental abutment surfaces, Biomed. Mater. 10 (2015) 055003. [30] M. Otto, Staphylococcal Biofilms, in: T. Romeo (Ed.), Bacterial Biofilms, Springer, Berlin Heidelberg 2008, pp. 207–228. [31] C.R. Arciola, D. Campoccia, P. Speziale, L. Montanaro, J.W. Costerton, Biofilm formation in Staphylococcus implant infections. A review of molecular mechanisms and implications for biofilm-resistant materials, Biomaterials 33 (2012) 5967–5982. [32] C. Vuong, M. Otto, Staphylococcus epidermidis Infections, Microbes Infect. 4 (2002) 481–489. [33] C. Vuong, S. Kocianova, J. Yu, J.L. Kadurugamuwa, M. Otto, Development of real-time in vivo imaging of device-related Staphylococcus epidermidis infection in mice and influence of animal immune status on susceptibility to infection, J. Infect. Dis. 198 (2008) 258–261. [34] S.N.A. Qazi, S.E. Harrison, T. Self, P. Williams, P.J. Hill, Real-time monitoring of intracellular Staphylococcus aureus replication, J. Bacteriol. 186 (2004) 1065–1077. [35] J.L. Kadurugamuwa, L. Sin, E. Albert, J. Yu, K. Francis, M. DeBoer, M. Rubin, C. Bellinger-Kawahara, T.R. Parr, P.R. Contag, Direct continuous method for monitoring biofilm infection in a mouse model, Infect. Immun. 71 (2003) 882–890. [36] B.A. Cunha, C.M. Sibley, A.M. Ristuccia, Doxycycline, Ther. Drug Monit. 4 (1982) 115–135.