Improved Hemocompatibility and Reduced Bacterial Adhesion on Superhydrophobic Titania Nanoflower Surfaces

Abstract

Thrombosis formation and bacterial infection are key challenges for blood-contacting medical devices. When blood components encounter a device’s surface, proteins are adsorbed, followed by the adhesion and activation of platelets as well as an immune response. This culminates in clot formation via the trapping of red blood cells in a fibrin matrix, which can block the device’s function and cause severe complications for the patient. In addition, bacteria may adhere to a device’s surface. This can lead to the formation of a biofilm, a protective layer for bacteria that significantly increases resistance to antibiotics. Despite years of research, no long-term solutions have been discovered to combat these issues. To impede thrombosis, patients often take antiplatelet drugs for the life of their device, which can cause excess bleeding and other complications. Patients can take antibiotics to fight bacterial infection, but these are often ineffective if biofilms are formed. Superhydrophobic surfaces show promise in reducing both thrombosis and bacterial infection on devices by impeding contact between biological components and the biomaterial. In this study, superhydrophobic titania nanoflower surfaces were successfully fabricated on a titanium alloy Ti-6Al-4V substrate with hydrothermal synthesis and vapor-phase silanization. The surface topography, surface wettability, surface chemistry, and surface crystallography of the surfaces was subsequently characterized. Surface hemocompatibility was investigated through lactate dehydrogenase (LDH) cytotoxicity analysis, blood-plasma protein adsorption, platelet and leukocyte adhesion and activation, and whole blood clotting analysis. Surface bacterial infection was characterized through Gram-positive and Gram-negative bacterial adhesion and biofilm morphology. The results indicated a reduction of protein adsorption, platelet and leukocyte adhesion and activation, bacterial adhesion, and biofilm formation as well as improved contact angle stability compared to control surfaces.

1. Introduction

Blood-contacting medical devices, such as stents and heart valves, are common treatments in modern healthcare. Every year approximately 1 million and 90,000 stent and prosthetic heart valve procedures are performed in the US, respectively [1,2]. Unfortunately, thrombosis and bacterial infection can complicate the use of these devices. Thrombosis is the process of blood clot formation and begins with the immediate deposition of blood-plasma proteins, mainly fibrinogen, on a device surface. Meanwhile, factor XII, another plasma protein, is deposited and facilitates thrombin formation via several reactions [3]. The thrombin reacts with fibrinogen to form fibrin which helps bind blood cells and activated platelets to form a clot. These clots can impede a device’s function, cause severe injury, or even death. Despite many years of research, no completely antithrombic surface has been created [4]. In addition, the threat of bacterial infection on these devices is a significant concern in the short and long-term [5-7]. Bacterial adhesion on a device surface can lead to the formation of a biofilm, a protective layer composed primarily of proteins and polysaccharides that makes treatment with antibiotics difficult [8,9]. Research also suggests that bacterial infections may lead to overactivation of the coagulation system, worsening thrombosis [10,11]. For these reasons, the development of biomaterials that can simultaneously reduce thrombosis and bacterial infection is vital.

Many different approaches have been advanced to combat thrombosis and bacterial adhesion on blood-contacting devices. The most common method to protect against thrombosis is to prescribe antiplatelet therapy with aspirin and a P2Y12 receptor inhibitor [12,13], which reduces the tendency for platelets to adhere and activate. However, the drawbacks of these drugs may include drug resistance, excess bleeding, and the need for long-term usage [14]. Another approach is through surface modification of the device. This includes heparin, an anticoagulant which can be applied to a device’s surface and acts by immobilizing thrombin [12,15]. Challenges faced with heparin coatings include difficulty binding heparin to the antithrombin molecule it acts on, and degradation of heparin due to stresses imposed within the body [16]. Other surface modifications include polyethylene glycol (PEG) coatings, albumin coatings, and endothelial cell coatings, where endothelial cells are seeded onto a surface to attempt reendothelialization of the device [16,17]. The problems with these solutions often include susceptibility to environmental factors, inefficacy in in vivo models, and difficulty finding an effective scaffold, respectively. More recently, nitric oxide (NO) releasing surfaces have been studied since NO naturally occurs in blood vessels and has both antithrombotic and antibacterial properties [18-20]. Although these surfaces have shown promise, the release of NO must be controlled and sustained long-term, which has proven difficult. Methods for reducing bacterial adhesion on surfaces have most often included the use of local or systemically released antibiotics, and silver-releasing coatings [5,21,22]. With these methods, antibiotic resistance of biofilms and damage to human cells have been significant concerns, respectively. Thus, no single antibacterial strategy has proven effective long-term and recurring infections often necessitate the complete removal and replacement of implanted devices [5,23].

Titanium and its alloys have been widely used in medical device applications for many years due to its general compatibility with bone tissues and ideal mechanical properties [24-27]. However, titanium devices are associated with many of the problems mentioned previously, and the material is not generally considered hemocompatible or antibacterial in its unaltered form. Recently, research into the nanotexture modification of the stable oxide layer (TiO₂) formed on titanium surfaces has shown promise in reducing thrombosis and bacterial adhesion [28-31]. Nanotextures can be created using different fabrication techniques such as anodization, sol-gel, or hydrothermal synthesis [32-34]. The hydrothermal synthesis technique is especially attractive because of its simplicity, low cost, and the ability to vary parameters to obtain different surface feature roughness, size, or shape. In addition to modifying nanotexture, modification of titania surface chemistry to create low surface-energy superhydrophobic nanotextured surfaces shows great potential for increasing antiadhesive properties. In recent years, researchers have shown the potential of using superhydrophobic surfaces for preventing biological fouling [35-38]. For example, superhydrophobic titania nanotubes fabricated via anodization reduce protein adsorption and platelet adhesion/activation, key processes that lead to thrombosis, as well as the adhesion of Gram-positive and Gram-negative bacteria [8,39,40]. Superhydrophobic surfaces work by combining of a rough nanotexture with low surface-energy to create the Cassie Baxter state, which provides a protective air-film layer and allows a high surface contact angle [43], severely limiting interaction between blood components/bacteria and the surface.

In this work, superhydrophobic titania nanoflowers were successfully fabricated for the first time on titanium alloy Ti-6Al-4V with the goal of improving hemocompatibility and reducing bacterial adhesion. Platelet adhesion and activation has previously been shown to be reduced on superhydrophobic titania nanoflowers grown from commercially pure titanium (CpTi) compared to a control CpTi surface [42]. However, surface interaction with individual blood components or bacteria has not been fully characterized, nor the behavior of nanoflowers grown from a Ti-6Al-4V substrate, which exhibits improved mechanical properties compared to CpTi [44]. In this study, a Ti-6Al-4V surface was first modified to create titania nanoflowers using hydrothermal synthesis, and then treated using a vapor-phase silanization technique to create a superhydrophobic surface. The material was characterized using scanning electron microscopy (SEM), contact angle goniometry, X-ray photoelectron spectroscopy (XPS), and X-ray diffraction (XRD). The surface cytotoxicity, protein adsorption, platelet/leukocyte adhesion, platelet activation, and whole blood clotting were investigated in addition to the adhesion and morphology of Gram-positive and Gram-negative bacteria. The results indicate improved antithrombic and antibacterial properties on superhydrophobic titania nanoflowers in comparison to control surfaces.

2. Materials and Methods

2.1. Fabrication of superhydrophobic titania nanoflower surfaces

Titania nanoflower surfaces were fabricated using a hydrothermal synthesis procedure on a titanium alloy, Ti-6Al-4V, as previously described in Wu et al [45]. Sections of titanium alloy TiAl6V4 were cut into 2.5 cm x 5 cm substrates and polished with 1200 grit sandpaper. The substrates were then sonicated in 100% acetone for 10 mins to remove surface contaminants. After sonication, the substrates were rinsed in de-ionized (DI) water and placed in an oxygen-plasma chamber to further remove surface contaminants. Finally, the substrates were transferred to sealed Teflon (PTFE) containers with 20 mM hydrofluoric acid (HF) solution and placed on a hot plate set to 300°C. After 8 hours, the treated substrates were rinsed with DI water, dried in nitrogen, and stored until needed. The substrates were further cut into 0.5 cm x 0.5 cm surfaces for all surface characterization and biological studies.

Superhydrophobic surfaces on titania nanoflowers were created using a simple vapor-phase silanization process with (heptadecafluoro-1,1,2,2-tetrahydrodecyl) trichlorosilane. Prior to silanization, surfaces were placed in a plasma chamber at 200 V in 10 cm³/min of oxygen gas for 15 mins to form −OH groups on the surface. Subsequently, the surfaces were placed on a glass slide adjacent to 200 μL of silane on a hot plate in an enclosed chamber. The hot plate was then heated to 120°C and left for one hour. Finally, the surfaces were rinsed in DI water and dried with nitrogen.

The following nomenclature will be used throughout this manuscript for different surfaces: Ti-6Al-4V titanium alloy (Ti), titanium alloy after silanization (Ti-s), nanoflower surface (NF), and nanoflower surface after silanization (NF-s). Both Ti and Ti-s were considered as control surfaces for all studies.

2.2. Characterization of titania nanoflower surfaces

The surface topography was characterized using a JEOL JSM-6500F Field Emission SEM. Prior to analysis, surfaces were coated with a 10 nm layer of gold to provide a conductive layer for imaging. Surfaces were imaged using 10kV at 6,500x, and 30,000x magnifications.

The surface chemistry was characterized with a PE-5800 XPS machine. Survey spectra were collected from 0 to 1100 eV and peak-fit analysis was done using Multipack and OriginLab software.

The surface crystal structure was characterized using a Shimadzu XRD 7000 maxima machine. XRD was carried out using copper (Cu) k-alpha radiation, thin film geometry with 2° incidence, a speed of 1°/min, and a range of 20 to 80 degrees.

The static contact-angle and roll-off angles of surfaces were measured using a Ramé-Hart Model 250 goniometer connected to a camera [46]. The surfaces were placed on the goniometer’s stage where a 10 μL droplet of water was placed on top. Using the DROPimage software connected to the goniometer, the contact angle between the droplet and surface was measured. Roll-off angle was measured by tilting the stage with droplet still on the surface until it slid.

The surface stability was evaluated by incubating surfaces in 2 mL of phosphate-buffered saline (PBS) at 37°C and 5% CO₂ on a shaker plate set to 100 rpm and characterizing them using contact angle and surface chemistry analysis. A control set of surfaces was also exposed to ambient air at 37°C and 5% CO₂ and evaluated similarly. The static contact-angles of the surfaces were measured initially and then subsequently every 7 days over a 4-week period using the method for characterizing surface wettability described above. XPS survey scans were taken initially and after the 4-week period using the method to characterize the surface chemistry described above.

2.3. Surface sterilization and incubation in biological studies

Different characterization techniques were used to evaluate the hemocompatibility and bacterial adhesion on surfaces. The surfaces were prepared the same way for all the characterization techniques except for the whole blood clotting study. Surfaces were sterilized under UV light for 30 mins and rinsed with sterile PBS 3 times. Sterilized surfaces in hemocompatibility studies were placed in a 48-well plate and incubated in 400 μL of 100μg/mL PBS protein solution or platelet rich plasma (PRP) for 2 hours at 37°C and 5% CO₂ on a shaker plate set to 100 rpm. Sterilized surfaces in bacterial studies were placed in a 48-well plate and incubated in 500 μL of bacterial solution for either 6 or 24 hours at 37°C. All solutions were aspirated after incubation and rinsed 3 times with PBS.

2.4. Protein adsorption onto surfaces

The adsorption of blood serum proteins, fibrinogen and albumin, on surfaces was characterized using XPS. After incubation in protein solution for 2 hrs and rinsing with PBS, the surfaces were rinsed 2 times with DI water to remove any un-adsorbed proteins. Survey spectra collected from 0 to 1100 eV and resolution N1s scans from 395 to 410eV. Peak-fit analysis was done using Multipack and OriginLab software.

2.5. Isolation of platelet-rich plasma (PRP)

Whole human blood was extracted from healthy adult individuals using 10 mL Vacuum Tubes containing the anticoagulant ethylenediaminetetraacetic acid (EDTA). The procedure for blood drawing from healthy donors was approved by Colorado State University Institutional Review Board and performed by a trained phlebotomist at the Colorado State University Health Center. Following a protocol described elsewhere, the first tube was discarded as waste, accounting for locally activated platelets due to needle insertion [47]. To separate the PRP from the whole blood, the tubes were centrifuged at 150 g for 15 mins. The PRP was then pooled into a single tube and used within 2 hrs. All hemocompatibility studies discussed below were performed with at least three healthy donors. To avoid donor-to-donor variability, the results presented are only from one donor. However, similar trends were observed for all the donors, which indicates the reproducibility of the results.

2.6. Cytotoxicity of surfaces

The cytotoxicity of surfaces was characterized using a commercially available LDH assay. After incubation in PRP, 100 μL of surface-exposed PRP was removed and added to wells in a 96-well plate. Subsequently, 100 μL of LDH solution was added to each well and incubated 30 mins at room temperature. Finally, the mixed solution was read at 490 nm using a plate reader to determine the amount of LDH released by the platelets and leukocytes.

2.7. Platelet and leukocyte adhesion on surfaces

Platelet and leukocyte adhesion on surfaces was characterized using a Zeiss Axio Imager.A2 fluorescence microscope. After incubation in PRP and rinsing in PBS, the surfaces were incubated with a 2 μM calcein-AM PBS solution for 20 mins in a dark atmosphere to avoid light exposure. The stain solution was then aspirated, and the surfaces rinsed 2 times in PBS before fluorescence imaging. Platelet and leukocyte adhesion area coverage was calculated using ImageJ software.

2.8. Identification of platelets and leukocytes on surfaces

Identification of platelets and leukocytes adhered on surfaces was characterized using fluorescence microscopy. Rhodamine-phalloidin was used to stain the cytoskeleton protein, actin, which will stain red in both platelets and leukocytes; while DAPI (4’,6-diamidino-2-phenylindole) was used to stain the cell nuclei, which will only stain blue in leukocytes. After incubation in PRP and rinsing in PBS, surfaces were incubated in a fixative of 3.7% formaldehyde PBS solution for 15 mins and rinsed 3 times with PBS. Subsequently, the surfaces were incubated in a permeative of 1% Triton X in PBS solution for 3 mins and rinsed 2 more times with PBS. 400 μL of rhodamine-phalloidin in PBS (1:200 concentration) was added to each well and allowed to incubate for 20 mins. After 20 mins, 42 μL of DAPI was added to each well and incubated for 5 additional minutes. Finally, the solution was aspirated, and the surfaces were rinsed 2 times with PBS before being imaged under a fluorescence microscope. Platelet and leukocyte adhesion area coverage and the number of cell nuclei were calculated using Image J software.

2.9. Platelet activation on surfaces

Platelet activation on surfaces was characterized using SEM imaging. After incubation in PRP and rinsing in PBS, surfaces were incubated for 45 mins in a primary fixative solution comprised of 3% glutaraldehyde, 0.1 M sodium cacodylate, and 0.1 M sucrose in DI water. The surfaces were then moved to a glass dish and incubated for 10 mins in a buffer solution of 0.1 M sodium cacodylate, and 0.1 M sucrose in DI water. Lastly, the surfaces were moved to a separate glass dish and dehydrated in solutions of 35, 50, 70, and 100 % ethanol in DI water for 10 mins each before a final dehydration in a solution of hexamethyldisilazane for 10 mins. The surfaces were coated with 10 nm of gold and imaged using SEM at 5 kV. The platelets visualized using SEM were classified into three different categories based on their morphologies:

• Unactivated (U): Platelets were essentially circular with no dendrite extension to the surface.

• Partially activated (P): Platelets displayed a few, short dendrites attached to the surface.

• Fully activated (F): Platelets displayed significant dendrite formation and attachment to the surface.

2.10. Whole blood clotting on surfaces

Whole blood clotting on surfaces was evaluated using a plate reader. The surfaces were sterilized under UV light for 30 mins and rinsed with sterile PBS 3 times. Sterilized surfaces were placed in a 24-well plate and 10 μL of whole blood droplets were placed on the surface. The blood was allowed to clot for 15, 30, or 45 mins. After the specific time interval, 1 mL of DI water was added to wells. The plate was then placed on a shaker plate at 100 rpm for 1 minute and allowed to rest for 5 mins. 200 μL of the un-clotted blood and DI water solution was then transferred to a 96-well plate and read at 540 nm.

2.11. Preparation of bacteria cultures

Bacterial strains of Escherichia coli (E. coli) and Staphylococcus aureus (S. aureus) were used to characterize the behavior of Gram-negative and Gram-positive bacteria on surfaces, respectively. Bacterial colonies were introduced into 5 mL of trypsin soy broth (TSB) used as a bacterial growth media and vortexed for 10 seconds to properly mix the bacteria within the media before incubation for 6 hours at 37°C. After incubation, 200 μL of solution was added to a 96-well plate. Three dilutions of the solution were made to create different bacterial concentrations, and the plate was subsequently read at a wavelength of 562 nm in a plate reader to determine optical density. A dilution was made until the solution obtained an average optical density of 0.52, indicating a concentration of 10⁹ bacterial cells/mL TSB solution. Once this was achieved, the solution was further diluted with TSB to obtain a 10⁶ bacterial cells/mL TSB solution.

2.12. Bacteria adhesion on surfaces

Bacteria adhesion on surfaces was characterized using fluorescence microscopy. After incubation and rinsing in PBS surfaces were moved to a new 48-well plate. A fluorescence stain was made using a commercially available live/dead bacteria staining assay with SYTO® 9 and propidium iodide. In a dark environment, 500 μL of stain solution was added to each well and incubated for 20 mins at room temperature. The stain solution was removed, and the surfaces rinsed with PBS before being fixed in a 3.7% formaldehyde PBS solution for 15 mins. Finally, the fixative was aspirated, and the surfaces rinsed in PBS before imaging under a Zeiss fluorescence microscope. Bacteria adhesion area coverage was calculated using Image J software.

2.13. Bacteria morphology on surfaces

Bacteria morphology on surfaces was characterized using SEM imaging. After incubation in PRP and rinsing in PBS, surfaces were incubated for 45 mins in a primary fixative solution comprised of 3% glutaraldehyde, 0.1 M sodium cacodylate, and 0.1 M sucrose in DI water. The surfaces were then moved to a glass dish and incubated for 10 mins in a buffer solution of 0.1 M sodium cacodylate, and 0.1 M sucrose in DI water. Lastly, the surfaces were moved to a separate glass dish and dehydrated in solutions of 35, 50, 70, and 100 % Ethanol in DI water for 10 mins each before a final dehydration in a solution of hexamethyldisilazane for 10 mins. The surfaces were coated with 10 nm of gold and imaged using SEM at 5 kV.

2.14. Statistical analysis

Surface characterizations with SEM were done using at least 3 different surfaces for each surface type at 5 different locations on the surface (n_min=15). Surface characterizations contact angle were done with at least 3 different surfaces for each surfaces type at 3 different locations (n_min=9). All hemocompatibility characterizations mentioned were completed at least 3 times with PRP taken from 3 different individuals (n_min=9). However, the results shown in each study are only from one individual since there is a large amount of variability in platelet count among donors. Data from all quantitative results were checked for normality and homogeneity of variance prior to evaluation with ANOVA and Tukey tests. Data were compared with appropriate groups and results were considered statistically significant with a p-value ≤ 0.05. Analysis was done using JMP and OriginLab software.

3. Results and Discussion

3.1. Surface characterization

In this study, superhydrophobic titania nanoflower surfaces was fabricated on titanium alloy Ti-6Al-4V substrates. This alloy has been extensively used in bone-related biomedical implants due to its high yield strength in comparison to CpTi and biocompatibility with bone tissues [48-51]. In addition, Ti-6Al-4V shows a high elastic modulus, excellent corrosion resistance, and improved fatigue strength [52]. However, Ti-6Al-4V has not been thoroughly investigated with relation to hemocompatible titania nanosurfaces or superhydrophobic coatings and thus was chosen for this research. NF-s was fabricated using two primary chemical processes: hydrothermal synthesis and vapor-phase silanization. The hydrothermal synthesis process is outlined completely by Wu et al. [45] and is described by the following chemical equations. Initially, the Ti from the Ti-6Al-4V substrate reacts with HF to form H₂TiF₆.

$$\mathsf{Ti}+6\mathrm{HF}\to {\mathsf{H}}_2{\mathsf{TiF}}_6+2{\mathsf{H}}_2\uparrow$$ (1)

The H₂TiF₆ produced from this reaction then reacts with water molecules to create Ti(OH)₄.

$${\mathsf{H}}_2{\mathsf{TiF}}_6+4{\mathsf{H}}_2\mathsf{O}\to {\mathrm{Ti}(\mathrm{OH})}_4+6\mathsf{HF}$$ (2)

The Ti(OH)₄ becomes TiO₂, nucleates, and continues to form more TiO₂ under reaction conditions.

$$\mathrm{Ti}(\mathrm{OH})4\to \mathrm{TiO}2+2H2O$$ (3)

The HF in solution begins to etch the formed TiO₂ while H₂TiF₆ diffuses to the surface of the TiO₂ nanoparticles, where more TiO₂ is continually created and deposited on top of itself [32].

$${\mathsf{TiO}}_2+6\mathsf{HF}\to {\mathsf{H}}_2{\mathsf{TiF}}_6+2{\mathsf{H}}_2\mathsf{O}$$ (4)

Once the synthesis of NF was complete, vapor-phase silanization was utilized to form a superhydrophobic surface. This process begins by treating the NF surfaces in an oxygen plasma chamber to form hydroxyl (OH⁻) groups on the surface. The silicon within the silane forms strong covalent R-Si-O bonds with these groups in order to attach itself to the substrate [53]. The combination of low-surface energy −CF₂ and −CF₃ functional groups bonded to the silicon and the rough nanoflower texture allow the surface to obtain high contact angles via the Cassie-Baxter state [54]. Either this state or the Wenzel state is observed when a droplet is placed onto a surface. In the Wenzel state, the fluid molecule is fully in contact with the surface features. Surfaces in this state do not achieve contact angles of >150°, which are indicative of a superhydrophobic surface. In the Cassie-Baxter state, fluid molecules are only in contact with a small fraction of the surface features and an air-film layer trapped between these features [54]. This significantly increases contact angle and reduces the liquid-surface interfacial area, which can reduce surface blood-clot formation and bacterial adhesion.

SEM was used to characterize the surface topography of titania nanoflower surfaces. The results for Ti indicate the expected topography (figure 1). The surface was relatively flat but showed streaks and scratches due to surface imperfections and processing. The results for NF showed many clusters of nanoflower formation uniformly distributed throughout the surface. Modifying both surfaces with silane to create Ti-s and NF-s did not change surface morphology upon SEM inspection. Average nanoflower diameter was determined by analyzing SEM images in ImageJ software. This was done by measuring a straight line drawn across the longest distance on the feature. The results indicate an average diameter of 823.6 ± 163.6 nm for the nanoflowers and no significant difference in the size of nanoflowers on NF-s when compared to NF.

Figure 1:

Representative SEM images of control and titania nanoflower surfaces fabricated using hydrothermal synthesis and vapor-phase silanization.

Contact angle and roll-off angle measurements were used to characterize surface wettability (table 1). Roll-off angles were found to be 29 ± 2.3° for Ti-s and 3 ± 0.5° for NF-s, while water droplets did not roll off Ti and NF. Ti, Ti-s, and NF, indicated the Wenzel state due to their relatively low contact angles and wettability. Ti showed contact angles of <90° indicating the expected hydrophilic surface while Ti-s displayed contact angles of >90° indicating a hydrophobic surface. The silane coating lowered the surface energy and allowed it to obtain hydrophobic qualities; however, the lack of nanotexturing impeded the development of high contact angles typical of the Cassie-Baxter state. NF showed contact angles of approximately 0°, indicating a superhydrophilic surface. This was observed through complete surface wetting upon placing a droplet of water on the surface. Although the mechanisms of superhydrophilic surfaces are still under debate, the effect is generally apparent on rough surfaces, such as titania nanotextures, with large effective surface area compared to projected surface area [55]. Contact angle measurements on NF-s of >150° and roll-off angles of <10° confirmed the successful fabrication of a superhydrophobic surface. This indicates the presence of an air-film layer trapped between the surface features, and thus the Cassie-Baxter state.

Table 1:

Measured static contact angle of different surfaces from goniometry.

	Contact Angle	Roll-off Angle
Ti	67.1 ± 4.6°	N/A
Ti-s	119.9 ± 2.8°	29.3 ± 2.3°
NF	~0°	N/A
NF-s	156.4 ± 3.8°	3.0 ± 0.5°

XPS scans were used to characterize the surface chemistry. XPS scans indicate relative atomic composition on a surface by measuring atomic binding energy. The results indicated O1s (~529 eV), Ti2p_2/3 (~458.5 eV), and C1s (~284.8 eV) peaks on all surfaces (figure 2). F1s (~689 eV) peaks and trace amounts of silicon in the form of Si2p (~99.4 eV) peaks were visible on Ti-s and NF-s due to silanization. A smaller F1s peak was seen on the NF surface because of the presence of HF in the hydrothermal synthesis process. The O1s peak increase and subsequent Ti2p_2/3 peak decrease seen on NF was caused by oxidation from the hydrothermal synthesis process. The C1s peak was highest on Ti because of contamination present in the environment and on the surface. After silanization, there was a change of the shape of the C1s peaks on Ti-s and NF-s due to the −CF₂ and −CF₃ functional groups contained within the silane compound.

Figure 2:

XPS survey scans for Ti, Ti-s, NF, and NF-s surfaces.

XRD was used to characterize the surface crystalline structure of different surfaces. Characterizing crystalline structure is useful for understanding the effects of hydrothermal synthesis and silanization on the titanium alloy surface. The results indicate the presence of rutile and anatase crystal phases on all surfaces (figure 3). The rutile phase is visible in the range of 35-40°, 53°, and 75-80°, while the anatase phase is visible around 63° and 71°. Both Ti and Ti-s surfaces also showed anatase at around 27°. The rutile phase is known to be the most stable phase and the anatase phase has been shown to be biocompatible with several different types of cells [56-58]. Overall, minimal changes in crystalline structure were seen after silanization or hydrothermal synthesis.

Figure 3:

XRD scans of Ti, Ti-s, NF, and NF-s surfaces.

Surface stability plays a vital role in internal medical devices because the device may be implanted in a patient for many years. Surfaces composed of superhydrophobic nanotextures present a challenge in this area. If the air-film layer necessary to maintain the Cassie-Baxter state is compromised, then the surface will lose its superhydrophobic nature, which will increase interaction between blood components, bacteria, and the surface, leading to complications [43]. Past research has shown air-film layers to be stable on superhydrophobic surfaces for at least 50 days under controlled parameters [59] and Sabino et al. showed titania nanotubes to be stable when incubated in PBS for 4 weeks under physiologically similar conditions [40]. In this work, stability was characterized after incubating surfaces in air or PBS under physiologically similar conditions for 4 weeks. The results indicate that after 4 weeks in air, Ti contact angles significantly increased from 67.1 ± 4.6° to 87.8 ± 5.6° (31%) while Ti-s, NF, and NF-s contact angles remained as they were as-fabricated (p≤0.05) (figure 4A). This increase was not expected and was due to contamination on Ti. After 4 weeks in PBS, Ti-s showed a statistically significant decrease in contact angle of 121.8 ± 7.0° to 94.3 ± 5.1° (23%) while Ti, NF, and NF-s contact angles did not significantly change (p≤0.05) (figure 4B). Furthermore, NF-s maintained superhydrophobic contact angles of >150° after 4-week incubation in both cases, which was significantly higher than all other surfaces (p≤0.05).

Figure 4:

Contact angle measurements on Ti, Ti-s, NF, and NF-s surfaces incubated in air (A) and PBS (B) over a 4-week period. Ti contact angle significantly reduced after 4-week incubation in air and Ti-s contact angle significantly reduced after 4-week incubation in PBS (* → p≤0.05). NF-s contact angles remained >150° after 4-week incubation in both air and PBS (p≤0.05).

XPS survey scans (data not shown) were taken after the 4-week period for surfaces incubated in air or PBS. Quantized results for the survey scans were tabulated in table 2. The results indicate silicon and fluorine elemental composition on NF-s decreased after 4-week incubation in either air or PBS, suggesting a degradation of the silane coating that allows for a superhydrophobic surface. High resolution XPS C1s scans of the surfaces were taken to further investigate the degree of this degradation by observing the −CF₂ and −CF₃ functional groups that are characteristic of the silane. These scans indicate C-C (~284.8 eV) and C-O (~286.5 eV) functional groups were found on all surfaces (figure 5), and O-C=O (~288.5 eV) was detected on Ti, Ti-s, and NF. As expected, −CF₂ (~292.0 eV) and −CF₃ (~293-294 eV) functional groups were found on Ti-s and NF-s. The results obtained after 4-week incubation in PBS are of primary importance for evaluating surface stability because they model what may happen when implants are in contact with liquid media after an extended period. Upon observing the −CF₂ and −CF₃ functional groups on Ti-s, a reduction in peak intensity is evident on the surface incubated in PBS in comparison to the as-fabricated and air-incubated surfaces. In contrast, these functional groups remained in similar quantities on NF-s under all conditions. These results explain why such a steep decrease in contact angle was seen after 4-week incubation on Ti-s in contrast to NF-s (figure 4B). Overall, the results suggest the nanotexture of NF-s allowed the surface to maintain the air-film layer, allowing high contact angles (>150°) and thus surface stability to remain.

Table 2:

Elemental composition for as-fabricated, 4-week air, and 4-week PBS surfaces taken from XPS survey scans.

		F%	O%	Ti%	C%	Si%
As Fabricated	Ti	0.00	50.67	26.60	22.76	0.00
	Ti-s	53.20	19.96	16.78	9.56	0.56
	NF	7.92	61.44	13.46	17.17	0.00
	NF-s	45.23	22.21	6.99	23.56	2.01
Air	Ti	0.00	40.28	15.09	44.63	0.00
	Ti-s	53.61	20.41	11.41	13.87	0.69
	NF	7.02	55.86	14.67	22.45	0.00
	NF-s	37.50	29.39	6.53	25.02	1.55
PBS	Ti	0.00	53.93	20.76	25.32	0.00
	Ti-s	42.65	22.29	9.60	24.16	1.31
	NF	4.37	64.83	15.33	15.47	0.00
	NF-s	40.98	28.09	7.52	22.78	0.63

Figure 5:

High resolution C1s XPS scans of as-fabricated surfaces and surfaces incubated in air and PBS for 4 weeks.

3.2. Characterization of hemocompatibility

Before a surface can be implemented in a biomedical device, it must be shown to be non-cytotoxic. In this study, a commercially available LDH assay was used to determine the cytotoxicity of all surfaces. LDH is an enzyme released when the plasma membranes of cells and platelets are damaged [60]. The assay works by using LDH to catalyze several reactions that result in the formation of formazan, which can be read for absorbance in a plate reader and is proportional to LDH release. Results show all surfaces in comparison to a negative (−) and positive (+) control (figure 6). The negative control in this study was just PRP in a well, and the positive control for this study was PRP lysed in a well to release maximum amount of LDH from platelets and leukocytes. All surfaces showed negligible cytotoxicity compared to the positive control, and no significant difference in cytotoxicity was found between the surfaces and negative control.

Figure 6:

Cytotoxicity results from a commercially available LDH assay. All surfaces showed minimal cytotoxicity compared to the positive control (p≤0.05).

Medical device-induced thrombosis is a common problem with blood-contacting biomedical devices such as stents, heart valves, and catheters, among others. The very beginning stages of thrombosis on a surface involve the deposition and adsorption of blood-plasma proteins that form a foundation for inducing future platelet, leukocyte, and blood cell attachment [3]. The primary protein involved in thrombosis is fibrinogen. When it attaches to a surface, it is central to binding platelet integrin receptors which leads to platelet activation and aggregation [61]. In contrast, albumin, the most prominent blood-plasma protein, behaves passively and inhibits protein adhesion which can indirectly reduce platelet activation and adhesion [62]. Fibrinogen’s shape is characterized as long and narrow while the shape of albumin is globular, which leads to easier unfolding and attachment on titania nano-textured surfaces [39]. Hence, it was expected that fibrinogen would adsorb more readily onto surfaces than albumin. In this study, surfaces were incubated in a fibrinogen or albumin solution. Protein adsorption was characterized using XPS scans by detecting the presence of the nitrogen N1s peak (figure 7). The N1s peak is characteristic of protein adsorption because it is not present on any of the surfaces prior to protein adsorption [39]. As expected, the results indicate the highest percentage of elemental nitrogen on Ti for surfaces incubated in fibrinogen (Table 3). This was followed by NF, Ti-s, and NF-s. For surfaces incubated in albumin, Ti was also found to have the highest adsorption followed by Ti-s, NF, and NF-s. The results show reduced adsorption of both proteins on NF-s due to its superhydrophobic nature. The reduced adsorption of fibrinogen on NF-s suggests a reduction of subsequent platelet adhesion and activation.

Figure 7:

High resolution N1s XPS scans indicate the presence of protein on the surfaces.

Table 3:

Elemental nitrogen percentage on different surfaces taken from XPS survey scans.

	Fibrinogen	Albumin
Ti	12.73%	5.50%
Ti-s	10.39%	4.7%
NF	11.29%	2.87%
NF-s	0.95%	0.72%

Following protein adsorption, platelets adhere and then activate on an artificial surface [3]. Activated platelets aid in modulating the activation of leukocytes in the blood stream, which leads to the formation of platelet-leukocyte aggregates [63]. Leukocytes can also promote platelet/leukocyte activation themselves and the blocking of coagulation inhibitors [12]. Therefore, a surface that prevents adhesion of platelets and leukocytes is ideal for reducing future activation and thrombosis. In this study, fluorescence microscopy was used to characterize platelet and leukocyte adhesion on surfaces after incubation in isolated PRP. The resulting images show platelets and leukocytes uniformly distributed on Ti, Ti-s, and NF (figure 8A). By contrast, NF-s showed platelets and leukocytes sparsely adhered across the surface. Quantification of the images indicates no significant difference between Ti and NF platelet and leukocyte adhesion (figure 8B). Ti-s showed a significant reduction when compared to both Ti and NF (p≤0.05) and NF-s had the least platelet and leukocyte adhesion compared to all surfaces (p≤0.05).

Figure 8:

(A) Representative fluorescence microscopy images of calcein-AM-stained platelets and leukocytes on different surfaces. (B) Platelet and leukocyte adhesion area as a percentage on different surfaces (calcein-AM stain). NF-s showed significantly lower platelets and leukocytes adhesion compare to all the surfaces (* → p≤0.05). Ti-s showed significantly lower platelets and leukocytes adhesion compare to Ti and NF (** → p≤0.05)

Calcein-AM staining do not differentiate between platelet and leukocyte adhesion. For this reason, another staining was performed, which utilized rhodamine-phalloidin and DAPI stains to distinguish the two blood components. Rhodamine-phalloidin stains all cytoskeletons red while DAPI solely stains the nuclei of cells blue. Since platelets do not have a nucleus, only leukocytes are stained by DAPI. The resulting rhodamine-phalloidin images show platelets and leukocytes uniformly distributed on Ti, Ti-s, and NF (figure 9A). By contrast, NF-s showed platelets and leukocytes sparsely adhered across the surface. Quantification of these images indicates no significant difference between Ti, Ti-s, and NF platelet and leukocyte adhesion (figure 9B). NF-s showed a significant reduction in adhesion when compared to all other surfaces (p≤0.05), which agrees with calcein-AM adhesion results. The DAPI images show leukocytes present on Ti, Ti-s, and NF (figure 9A) with fewer leukocytes visible on NF-s. Quantification of these images indicates a significant reduction between Ti and all other surfaces (p≤0.05) (figure 9C). NF-s was shown to have a significantly lower number of leukocytes than all other surfaces as well (p≤0.05).

Figure 9:

(A) Representative fluorescence microscopy images of rhodamine-phalloidin and DAPI stained platelets and leukocytes on different surfaces. (B) Platelet and leukocyte adhesion area as a percentage on different surfaces (rhodamine-phalloidin stain) and (C) Leukocytes per mm² (DAPI stain). NF-s showed significantly lower platelet and leukocyte adhesion compare to all the surfaces (* → p≤0.05). Ti showed significantly higher leukocyte adhesion compare to Ti-S (** → p≤0.05) and NF (*** → p≤0.05).

As expected, the results from all platelet and leukocyte adhesion characterizations (figure 8, 9) indicated more adhesion on Ti, Ti-s, and NF in comparison to NF-s. The change of surface nanotexture between Ti and NF only lowered adhesion in the case of leukocytes (figure 9A), but not platelets (figure 8B, 9B). The effect of lowering surface energy between Ti and Ti-s appeared to reduce overall adhesion to a larger degree. However, the combination of the nanotexture and lowered surface energy on NF-s significantly reduced platelet and leukocyte adhesion. This is due to the surface’s superhydrophobic properties decreasing interaction between PRP and the surface features, suggesting NF-s may reduce the body’s coagulation and immune response.

When platelets are activated they change morphologies by forming dendrites that attach to the surface and other platelets [64]. This activation is a prerequisite to blood-clotting and full thrombosis [3]. Platelet activation was characterized through SEM imagery. The results show large networks of platelets adhered and activated on Ti, Ti-s, and NF (figure 10A). NF-s did not show these networks and very few platelets were found on the surface. Before quantification, the platelets adhered to the surface were categorized as unactivated, partially activated, or fully activated based on criteria outlined in section 2.9. No statistical difference was observed in the number of unactivated, partially activated, or fully activated platelets between Ti and Ti-s (figure 10B). Furthermore, neither unactivated or partially activated platelets were found on NF or NF-s. There was a statistically significant increase in fully activated platelets on NF compared to all surfaces (p≤0.05), which may be due to the high surface roughness allowing for more surface area for platelets to attach and form dendrites. Finally, NF-s showed the least amount of fully activated platelets compared to all other surfaces (p≤0.05), which was expected considering its reduced protein adsorption and platelet adhesion.

Figure 10:

(A) Representative SEM images of platelet activation on various surfaces. (B) Platelet count on different surfaces. NF-s was shown to have the lowest number of fully activated platelets compare to all the other surfaces (* → p≤0.05). NF was shown to have the highest number of fully activated platelets compare to all all the other surfaces (* → p≤0.05).

Reduction of whole blood clotting is important for long-term hemocompatibility. Within the coagulation cascade, formation of a fibrin matrix follows platelet activation, which traps red blood cells and forms a clot [40,47]. Clot formation can result in a failure of the device and injury to the patient, such as stroke. In this study, whole blood clotting was characterized by measuring free hemoglobin with an absorption plate reader after 15, 30, and 45-minute clotting periods. Blood was allowed to clot on surfaces for the given amount of time, after which DI water was added to the surface to dissolve and lyse whole blood cells that were not trapped in the fibrin matrix, releasing hemoglobin into the solution [65]. Hence, higher absorbance values indicated more free hemoglobin in solution and thus less blood clotting on the surface (figure 11). Both Ti and NF showed considerable blood clotting. After 30 mins on Ti, there was significantly more clotting on the surface. No significant changes were seen for Ti-s, NF, or NF-s in clotting over 45 mins. After 45 mins, Ti-s and NF-s indicated the lowest amount of clotting of the surfaces. This was expected based on results for NF-s from the aforementioned hemocompatibility characterizations and further suggests that NF-s may improve hemocompatibility of a blood-contacting biomedical device.

Figure 11:

Whole blood clotting indicated by free hemoglobin absorbance measured after 15, 30, and 45 mins. There was significant reduction in free hemoglobin absorbance on Ti from 15 to 45min (# and & → p≤0.05). There was significant difference between free hemoglobin absorbance on Ti-s and NF (* → p≤0.05), NF and NF-s (** → p≤0.05), Ti and Ti-s (*** → p≤0.05) and Ti and NF-s (**** → p≤0.05) after 45 mins.

3.3. Bacterial studies

Bacterial infection presents another issue for blood-contacting biomedical devices. Bloodstream infections are most common in patients with catheters and can lead to sepsis or even death [66]. Bacterial colonies grown on implant surfaces often lead to biofilm formation, which can protect the bacteria from antibiotics [8]. Infectious bacteria can be divided into two different types: Gram-positive and Gram-negative. These two bacterial types differ in several ways. One prominent difference is in cell-wall thickness and composition, where Gram-positive bacteria generally have thick peptidoglycan layers on top of a phospholipid cytoplasmic membrane, and Gram-negative bacteria have thin peptidoglycan layers in between two inner and outer cytoplasmic membrane [67]. In this study, Staphylococcus aureus and Escherichia coli were chosen to characterize the behavior of Gram-positive and Gram-negative bacterial on surfaces, respectively. S. aureus is a Gram-positive bacterial strain commonly associated with infections on medical devices and can even be found on human skin [8]. E. coli, while not often associated with medical device infections, is similar in structure to P. aeruginosa, a more prevalent infectious bacterial strain [19], and thus served as a model for the behavior of Gram-negative bacteria. Since both bacterial types may require different treatments [6], it is essential to characterize their respective behaviors on a biomedical device surface.

Fluorescence microscopy was used to characterize adhesion of bacteria on all surfaces. A commercially available live/dead bacteria assay containing two stains (Syto-9 and propidium iodide) was used to differentiate between bacteria that were alive after incubation and those that were dead. Syto-9 stains both living and dead bacteria, while propidium iodide only stains dead bacteria. Syto-9 appears green in fluorescence images and propidium iodide appears red. Images taken of surfaces incubated in both bacterial types generally showed more living bacteria than dead bacteria. Images of surfaces incubated in S. aureus indicate bacterial adhesion onto Ti, Ti-s, and NF after 6 hrs and reduced adhesion on NF-s (figure 12A). After 24 hrs, increased bacterial adhesion is observed on all surfaces compared to 6 hrs, however, there is lower overall adhesion on NF-s. The quantified results indicate that both live and dead bacterial adhesion increased between 6 and 24 hrs for NF (p≤0.05) (figure 12B, 12C). No significant difference was observed between 6 and 24 hrs on Ti, Ti-s and NF-s for live and dead S. aureus adhesion. Live bacterial adhesion after 24 hrs was shown to be the least on NF-s compared to all other surfaces (p≤0.05). Dead bacterial adhesion after 24 hrs was highest on NF and no significant difference was found amongst Ti, Ti-s, or NF-s (p≤0.05).

Figure 12:

(A) Representative fluorescence images of S. aureus on different surfaces. Bacterial cell adhesion area percentage for live (B) and dead (C) S. aureus after 6 and 24 hrs. NF-s showed the lowest amount of live S. aureus adhesion after 24 hrs compared to all other surfaces (* → p≤0.05). NF showed the highest amount of live and dead S. aureus adhesion after 24 hrs compared to all other surfaces (** → p≤0.05).

Images of surfaces incubated in E. coli indicate bacterial adhesion onto Ti, Ti-s, and NF after 6 hrs and reduced adhesion on NF-s (figure 13A). After 24 hrs, increased bacterial adhesion is observed on all surfaces relative to 6 hrs, however, adhesion is still comparatively less on NF-s. The quantified results indicate that live bacterial adhesion increased between 6 and 24 hrs on Ti-s (p≤0.05) (figure 13B, 13C). Live bacterial adhesion did not significantly change after 24 hrs for Ti, NF, and NF-s and was shown to be the least on NF-s compared to all surfaces (p≤0.05). No significant difference was found on Ti, Ti-s, or NF-s for dead bacterial adhesion after 24 hrs compared to 6 hrs. NF indicated a significant decrease in bacterial adhesion between 6 and 24 hrs (p≤0.05). No significant difference was found amongst Ti, Ti-s, or NF-s after 24 hrs incubation (p≤0.05).

Figure 13:

(A) Representative fluorescence images of E. coli on different surfaces. Bacterial cell adhesion area percentage for live (B) and dead (C) E. coli after 6 and 24 hrs. NF-s showed the lowest amount of live E. coli adhesion after 24 hrs compared to all other surfaces (* → p≤0.05).

In general, dead bacterial adhesion was comparable between Gram-positive and Gram-negative bacteria. The bactericidal effects of the surfaces can be interpreted as minimal since the number of dead bacteria was either much less than or equal to the number of live bacteria on every surface. Similar to the results from hemocompatibility characterizations, the findings suggest reduced liquid-surface interaction on superhydrophobic NF-s inhibited the adhesion of both Gram-positive and Gram-negative bacterial colonies.

Following bacterial adhesion, biofilms form and create a network that provides protection bacterial cells. Certain genes that code for multidrug resistance can more readily be shared between bacterial plasmids within the film, increasing antibiotic resistance [66,68]. Since there is often not an effective solution to destroying a biofilm, the best solution is to prevent its formation [9,66], which may be achieved by altering the biomaterial’s surface. In this work, biofilm formation was characterized using SEM images. The results for S. aureus showed more adhesion after 24 hrs compared to 6 hrs on all surfaces (figure 14A). After 6 hrs, colonies of bacteria began to form on Ti, Ti-s, and NF. After 24 hrs, larger colonies are formed on these surfaces and biofilm formation appears to begin on Ti and NF. After 6 hrs no bacteria were visible on the NF-s surface. After 24 hrs, small colonies began to form, but were scattered sparsely and only seen on damaged parts of the surface. The results for E. coli showed more adhesion after 24 hrs compared to 6 hrs on Ti, Ti-s and NF (figure 14B). After 6 hrs, colonies of bacteria began to form on Ti, Ti-s, and NF. After 24 hrs, larger colonies are formed on these surfaces and biofilm formation appears to begin on Ti-s. After 6 hrs and 24 hrs, no bacteria were visible on the NF-s surface. The results suggest that the superhydrophobic NF-s surface helped to inhibit adhesion and biofilm formation of both Gram-positive and Gram-negative bacteria.

Figure 14:

Representative SEM images of S. aureus (A) and E. coli (B) on different surfaces.

Overall, NF-s significantly reduced adhesion and activation of blood components and bacteria on the surface. This is due to the superhydrophobic nature of the surface under Cassie Baxter state that effectively prevented any significant contact of liquid with the surface. The cassie Baxter state is achieved due to the combination of roughness and nanotopography of titania nanoflower surfaces and modification of the surfaces with a low surface-energy molecule, resulting in high water contact angles and formation of a barrier air-film layer preventing minimal liquid contact with the surface. The lack of these features on Ti, Ti-s, and NF resulted in surfaces that did not prevent biological response similar to that on NF-s.

4. Conclusions

In this work, superhydrophobic titania nanoflowers were fabricated via simple hydrothermal synthesis and silanization procedures and were shown to remain superhydrophobic after 4-week incubation in PBS. Hemocompatibility was evaluated by measuring surface cytotoxicity, fibrinogen and albumin adsorption, platelet/leukocyte adhesion and activation, and whole blood clotting. The surface did not show short-term cytotoxic effects and indicated reduced protein adsorption compared to control surfaces. This reduced adsorption impeded subsequent platelet and leukocyte adhesion. Furthermore, bacterial adhesion and biofilm formation were characterized with Gram-positive and Gram-negative bacteria. The results indicated a significant reduction of live adhesion and no biofilm formation after 24-hour incubation in either bacterial strain on the surface compared to control surfaces. In summary, superhydrophobic titania nanoflowers indicated improved hemocompatibility and reduced bacterial adhesion compared to both non-textured and unmodified Ti-6Al-4V surfaces. Future work should be directed towards improving the durability of these surfaces and understanding their interaction with biological components on a long-term scale.

Highlights:

• Superhydrophobic titania nanoflowers (NF-s) fabricated using hydrothermal process

• NF-s surfaces prevented platelet adhesion and activation

• NF-s surfaces reduced adhesion of both Gram-positive and Gram-negative bacteria