1. Introduction
Titanium alloys have been widely used in the medical industry for surgical implants due to their favourable mechanical properties, corrosion resistance, passivity and biomedical properties [
1,
2,
3,
4]. In recent times, there has been a large research focus on fabricating nanostructured surfaces as a means of reducing bacterial adhesion on medical implants. Physical (micro and nanoscale topography) and chemical surface properties have been found to influence bacterial adhesion and attachment [
5,
6].
Factors such as the body’s acceptance of the foreign material, osseointegration and bone regrowth, influence implant integration. Surface topography is said to have a direct effect on the bone’s biological response, as micro-rough implants have shown to be superior to smooth implants [
7,
8]. The cells’ ability to adhere to the surface of the implant is influenced by surface properties, charge distribution and material chemistry [
9]. Surface modification techniques alter the surface roughness of the implant, to mimic that of natural bone, thereby improving the implant’s integration into the body [
10,
11,
12]. Studies have found that inducing porous structures, with morphologies similar to that of the natural extracellular matrix, benefits vascularization, improves osseointegration and bone regrowth, transports nutrients, and increases biocompatibility and implant lifespan [
2,
9,
10,
13,
14,
15]. Similarly, nanostructures improve cell adhesion by providing pathways for a continuous supply of the nutrients and body fluids necessary for cell growth [
16].
Surface structure, morphology and wettability play key roles in the successful osseointegration of implants [
7,
11,
17,
18]. Wettability is considered one of the most important surface properties that affect the biological response to an implant [
19]. It can affect bacterial adhesion and biofilm formation, protein adsorption, and the rate of osseointegration in vivo. Hydrophilic surfaces have a tendency to improve early stages of cell adhesion, proliferation, differentiation and bone mineralization, compared to hydrophobic surfaces [
7,
19,
20,
21,
22,
23,
24,
25,
26]. While the antibacterial properties of TiO¬
2 has been established, research is limited in establishing the mechanical properties of bactericidal surfaces.
In our group’s previous work, we found that nanofibers approximately 300 nm in length improved human osteoblast cell growth over 24 h compared to flat Ti-6Al-4V, and simultaneously accelerated
Staphylococcus aureus (
S. aureus) death over 18 h [
27]. This work examines the effect of surface morphology and structure size of hydrothermally synthesised TiO
2 surface textures on bactericidal properties and osteoblast metabolic activity. The mechanical properties, wettability and surface roughness, are statistically investigated. This study highlights the significance of nanoscale texture, topography and general surface properties in determining cellular and bacterial responses.
2. Results
2.1. Surface Morphology
Varying the NaOH concentration, hydrothermal reaction time and reaction temperature produced different TiO
2 nanotextured surfaces.
Figure 1 shows SEM topographical images of the fabricated samples. Each sample displays a unique structural distribution, spatial nanostructure arrangement and density. The array height and general morphology were heavily influenced by the process parameters, particularly NaOH concentration. Sample 1.0_3_240 (
Figure 1a) produced randomly spaced and sharp-tipped individual nanowires with an average length of 307 ± 29 nm (
Table 1). Sample 2.0_3_240 doubled this concentration to 2 M NaOH, producing a largely different structure (
Figure 1b) compared to the 1 M NaOH reaction (
Figure 1a). At this concentration, large fiber heights (1317 ± 259 nm) with an interconnected wire mesh was observed. In contrast, sample 0.1_3_240 (
Figure 1c), which used low NaOH concentrations (0.1 M), produced shorter length (180 ± 40 nm) and smaller diameter (17 ± 3 nm) wires.
Sample ¬1.0_3_120 was obtained at a reaction temperature of 120 °C. The topography of this surface (
Figure 1d) was similar to ¬sample 1.0_3_240 (
Figure 1a); however, the tips of these nanostructures appear to have fused together. The average height of these structures was larger (328 ± 55 nm), whereas their average diameter (18 ± 3 nm) was smaller than sample 1.0_3_240. Sample 1.0_1_240 (
Figure 1e) was fabricated in a short reaction time (1 h), producing a very similar shape and diameter to the 1.0_3_120 sample (
Figure 1d). The structures of the 1.0_1_240 sample grew to an average height of 244 ± 36 nm and diameter of 21 ± 4 nm.
Sample 2.0_10_240 (
Figure 1f) was fabricated under high concentration and temperature (2 M NaOH at 240 °C for 10 h). Similar to sample 2.0_3_240, at this NaOH concentration (2 M) the fibers grew and fused together, producing a mesh-like array. The long reaction time (10 h) was responsible for the significantly larger array dimensions compared to the same NaOH concentration and shorter reaction time (2 h,
Figure 1b). All surface structures grew in a random manner in terms of the orientation angle and spacing.
The average height and diameter of the surface structures for each sample is summarised in
Table 1. Three samples were fabricated for each reaction condition, with 10 height and diameter measurements taken from each sample.
An XRD was completed to confirm the surface material (TiO
2) (
Figure 2).
As can be seen in
Figure 2, the spectra agree well with the standard rutile (R) and anatase (A) diffraction peaks, confirming that the material is TiO
2 with a Ti substrate [
28]. While all samples showed a combination of these three phases, sample 2.0_10_240 showed a majority of the anatase phase with one minor rutile peak, indicating that the phase of the nanostructured sample was predominantly anatase. The Ti peak from the substrate was not observed in sample 2.0_10_240 due to the large thickness of the TiO
2 nanostructured surface, which obscured the substrate surface from detection.
2.2. Surface Roughness
Surface roughness was measured using a LEXT OLS4100 3D Laser Measuring Microscope (Olympus, Tokyo, Japan).
Table 2 summarizes the average arithmetic mean deviation (S
a) of each sample.
The results of the optical profilometry show that surface textures with small height (sample 0.1_3_240, 180 ± 40 nm) had small average mean surface roughness (0.027 µm). Similarly, sample 2.0_3_240, which had the largest average array height (14796 ± 3053 nm) had the highest average mean surface roughness (2.296 µm). For other samples, surface roughness increased with array height, with the exception of sample 1.0_3_240 (307 ± 29 nm), which had an Sa of 0.140 µm, similar to that of sample 2.0_3_240 (array height 1317 ± 259 nm). Overall, the results showed a large variation in surface roughness properties for different samples and morphologies. The skewness and large kurtosis values (particularly for sample 1.0_1_240) show that surface roughness is not uniform across the scanned area, indicating that the fabricated surfaces were random in nature, due to the lack of control within the hydrothermal vessel.
2.3. Mechanical Properties
Mechanical properties were measured using a Hysitron TI 950 Nanoindenter (Bruker, Billerica, MA, USA). The average elastic modulus and hardness values are shown in
Table 3. Ti-6Al-4V (110 GPa elastic modulus and 0.36 Poisson ratio) was used to calculate the Young’s modulus of the sample material.
Surface textures with mesh-like morphologies produced structures with low Young’s moduli compared to surfaces with pillar-like structures. Sample 2.0_3_240, which had the largest array height and diameter, produced the lowest average Young’s modulus and hardness properties. Samples 1.0_1_240 and 1.0_3_120 showed the highest Young’s moduli of the sample group. Sample 0.1_3_240 produced high hardness values, potentially due to the small structure height, allowing the Ti substrate to be detected.
Figure 3 shows the load vs. displacement curve obtained by the nanoindentation tests for the 1.0_3_120 sample. As in typical load vs. displacement curves, the unloading gradient (dP/dh) is indicative of the hardness properties of the surface.
2.4. Contact Angle
The contact angle measured the hydrophilicity of each surface.
Table 4 shows the average contact angle, measured by the static sessile drop method.
Interestingly, the results showed that large mesh-like morphologies (samples 2.0_3_240 and 2.0_10_240) produced superhydrophilic surfaces (contact angle below 10°) due to the large porosity of the nanostructured surface, allowing liquid to penetrate. Structures with smaller heights (samples 1.0_3_120 and 1.0_1_240) repelled the spreading of the liquid, giving higher contact angles. In the case of sample 1.0_1_240, the contact angle was measured to be above 90°, indicating that the surface was on the borderline between hydrophilic and hydrophobic. This was the only surface out of the samples tested to produce a near hydrophobic surface.
2.5. Bactericidal Effects
The bactericidal properties of each surface against
S. aureus and
Pseudomonas aeruginosa (
P. aeruginosa) were measured using the standard plate count method [
29] at 0-, 3- and 18-h time points. A control well of cells alone (no substrate material) measured the natural death of the cells over the given incubation period, and a flat Ti surface was used as a comparison, representing the current implant surfaces.
Figure 4 shows the bacteria viability (CFU/mL) test results of
S. aureus and
P. aeruginosa as a percentage of the cells alone control for each surface texture. Results that were statistically significant are indicated.
The results show that sample 1.0_3_240 was the most efficient surface, reducing the CFU/mL of S. aureus by 46% over 18 h. This surface significantly (p < 0.0001) reduced the CFU/mL compared to both the control and flat Ti-6AL-4V samples, indicating that the textured surface played a significant role in killing S. aureus cells. Similarly, the results showed that there was a significant difference between the CFU/mL of the flat and nanostructured surface (sample 1.0_3_240) at 3 h. This showed that the bactericidal properties of the surface had this effect on the bacteria within the 3-h time frame.
Sample 0.1_3_240 showed similar results after 18 h of incubation, with the CFU/mL dropping significantly between the 3- and 18-h time points. The results showed that the CFU/mL of this sample and the control were comparable at 3 h; however, there was a large decrease in the CFU/mL for the textured surface between 3 and 18 h. This indicated that, while samples 1.0_3_240 and 0.1_3_240 showed similar results at 18 h, the 1.0_3_240 surface produced these effects much earlier.
Sample 1.0_1_240 was the only hydrophobic surface tested (contact angle 91.12°). However, this seems to have had no significant impact on the bactericidal properties of the material, which gave similar results to sample 2.0_10_240 (superhydrophilic contact angle < 10°). Similarly, sample 2.0_3_240 (superhydrophilic mesh-like array) did not exhibit any particularly extreme results, but rather a relatively average and moderate bactericidal efficiency compared to other samples.
The sample with the largest CFU/mL of S. aureus at 18 h was the cells-alone control. This was an interesting finding, as it indicates that all TiO2 structured surfaces accelerated the death of S. aureus over 18 h.
The figure also shows that the CFU/mL of P. aeruginosa cells increased significantly over 18 h for all samples. This figure displays the CFU/mL results as a percentage of the cells-alone control sample. All P. aeruginosa sample wells experienced a statistically significant increase in CFU/mL from 0 to 3 h and from 3 to 18 h. Again, the sample which produced the least P. aeruginosa growth was 1.0_3_240, indicating that this surface texture is effective against both gram-negative and gram-positive bacteria types. This surface had significantly (p < 0.0001) reduced the growth of the bacteria compared to both the flat and control samples.
Unlike the observed
S. aureus behaviour,
P. aeruginosa cells thrived on the samples, including the cells-alone control and the flat Ti-6Al-4V. The results showed that the cells-alone control had the second lowest CFU/mL at 18 h (after sample 1.0_3_240). Sample 2.0_10_240 produced the largest CFU/mL at 18 h due to the size of the structures, which had largely increased the surface area to which bacteria could attach [
5,
30]. In addition, the growth rate of
P. aeruginosa was faster than
S. aureus, explaining the large increase in
P. aeruginosa cells during the first 3–15 h of incubation [
31,
32]. The hydrophobic surface (sample 0.1_3_240) and the two superhydrophilic surfaces (samples 2.0_3_240 and 2.0_10_240) showed relatively similar behaviour, suggesting that contact angle did not significantly affect the bactericidal properties of the material.
2.6. Human Osteoblast Cellular Metabolic Activity
The AlamarBlue
TM assay determined the cellular metabolic activity of human osteoblast cells on the TiO
2 surfaces.
Figure 5 shows the AlamarBlue
TM results after 24 h of incubation (see additional information in
Supplementary file). The results have been normalised to the cells-alone control.
Figure 5 shows that flat Ti-6Al-4V and the 0.1_3_240 surface produced the highest cellular metabolic activity among the tested surfaces. Interestingly, these two surfaces have similar contact angles (50 and 36°, respectively) (
Table 4), with sample 0.1_3_240 having the smallest surface structures. This indicates that the osteoblast cells had a higher cellular metabolic activity on surfaces within a specific contact angle range or a small structure height (below 200 nm).
Figure 6 shows the SEM images of osteoblast cells attached to the flat and nanostructured surfaces after 4 and 24 h of incubation. The figure shows the effective spreading and growth of the cells on flat surfaces, which is reflected in the metabolic activity results (
Figure 5). Images for samples 1.0_1_240 and 2.0_10_240 show that structures pierce the osteoblast cells, explaining the low activity rate for these samples. Alternatively, sample 0.1_3_240 showed less piercing (due to the small structure size and surface roughness of the sample), giving the highest metabolic activity of the textured surfaces.
Figure 7 shows a direct comparison between each surface and their contact with
S. aureus,
P. aeruginosa and osteoblast cells. It is interesting to note that the surface 1.0_3_240 produced the best results against both bacteria types and a modest behaviour in promoting osteoblast activity.
Figure 7a shows the two bacteria cells types being pierced by the nanostructures, whereas the osteoblast cell did not experience this. Sample 2.0_3_240 (
Figure 7b), which has a high CFU/mL of both
S. aureus and
P. aeruginosa and a low cellular activity, showed both bacteria and osteoblast cells being pierced by the structures.
Figure 7f shows that the large surface area of the structures provides more surfaces to which the bacteria can adhere, showing that bacteria adhere on all locations of the structures and stretch and pierce the osteoblast cells. Shorter structure heights cause less damage to the
S. aureus cells (
Figure 7c–e) but effectively puncture
P. aeruginosa and osteoblast cells. From this figure and the results obtained from the bacteria viability and AlamarBlue
TM tests, it can be said that sample 1.0_3_240 was the most effective surface in killing gram-negative and gram-positive bacteria and promoting osteoblast activity.
2.7. Statistical Correlations
Statistical analysis completed in the IBM SPSS Statistics program tested correlations between CFU/mL (
S. aureus and
P. aeruginosa) and cellular metabolic activity, and the measured the surface properties. As these relationships were not linear, nonparametric correlation tests were used.
Table 5 shows the correlation coefficients between the cellular metabolic activity and CFU/mL and each surface property, where 0.1 < |r| < 0.3 indicates a small or weak correlation, 0.3 < |r| < 0.5 shows a medium or moderate correlation, and 0.5 < |r| is a strong or large correlation.
The results of this correlation test showed that both the contact angle and Young’s modulus had a statistically significant (p < 0.01) moderate correlation to osteoblast cellular metabolic activity. This is a significant finding as it shows that the cellular metabolic activity of osteoblast cells is statistically influenced by surface properties. The correlation data shows that the metabolic activity is more strongly correlated to Young’s modulus compared to the surface contact angle. Interestingly, these results show that the mean surface roughness is not correlated to the metabolic activity of the osteoblast cells over 24 h.
The correlation tests also showed that both the CFU/mL of the gram-negative and gram-positive bacteria tested was statistically (and significantly) correlated to the height of the surface structures. The correlation was statistically weak for S. aureus and moderate for P. aeruginosa; however, both correlations are significant (p < 0.05). There are no other tested surface properties that are significantly correlated to the CFU/mL of these bacteria. This is an important finding, as it has the potential to influence future surface texture design.
These findings are significant as it statistically confirms the phenomenon of surface morphology and wettability playing key roles in implant osseointegration [
7,
11,
17,
18]. It has been previously believed that surface roughness influences the death of bacteria on nanostructured surfaces; however, these results show that the structure height influences bacteria death.
3. Discussion
The purpose of this study was to test the effect of NaOH concentration, reaction time and reaction temperature on the surface fabrication of hydrothermally synthesised TiO
2 structured surfaces. In addition, the bactericidal properties of these surfaces to gram-negative and gram-positive bacteria and their human osteoblast metabolic activity were tested. Changing the hydrothermal process parameters produced various surface textures. The tests showed that the structure height grew with hydrothermal reaction time [
31,
32]. The reaction temperature had a modest effect on the structure height, with higher temperatures producing smaller structures and with larger diameters than surfaces formed at low temperatures (120 °C).
From the data collected, it can be said that NaOH concentration had the largest effect on the array height within the range of concentrations tested (0.1–2 M NaOH). At low concentrations (0.1 M), structures were less than 200 nm in height and 18 nm in diameter. These small structures also produced the lowest surface roughness (0.027 µm), with a contact angle of 41.4° and the highest osteoblast cellular metabolic activity after 24 h of incubation.
When the NaOH concentration was increased to 2 M NaOH, a highly dense, closely packed mesh-like array formed [
33], due to the increased number of nucleation sites. The most significant increase in array height occurred when a high NaOH concentration (2 M) was combined with a long reaction time (10 h) and a high reaction temperature (240 °C). Surfaces with mesh-like morphologies (samples 2.0_3_240 and 2.0_10_240) produced large dimensions (height and diameter) and air pockets, resulting in high surface roughness characteristics and superhydrophilic contact angles. The large surface structures of these samples allowed liquid to penetrate the roughness grooves, reducing the wettability angles [
20]. Morphological and structural parameters have a significant effect on the wetting angles, where adjusting the scale or roughness of the topography can largely impact the wetting ability [
34]. This is reflected in the results, where various surface roughness values and textures produced surfaces with varying contact angles and wetting characteristics.
All surfaces contained anatase and rutile phases, indicating TiO
2. The largest sample (sample 2.0_10_240) gave an XRD spectrum of predominately anatase. While it can be thought of as the less stable TiO
2 phase, anatase is quite often produced during TiO
2 formation due to its constrained structure and enhanced formation kinetics [
35].
The nanostructured TiO
2 surface caused the cells to respond to the physical morphology and interaction forces of the structures [
29], leading to an effective bactericidal activity and an increased osteoblast activity compared to flat Ti, as seen in the experimental results. The deformation of the peptidoglycan cell walls by the TiO
2 nanotextured surfaces resulted in cell death [
36]. From the SEM images obtained of the bacteria cells in contact with the nanostructure surfaces, piercing of the cell walls was clearly observed. The piercing led to the disfiguration and collapse of the bacteria cells, resulting in bacteria death and highlighting this as the possible mechanism of bacteria killing. The reason for this behaviour can be explained by the stiffness of the bacteria cells compared to the TiO
2 nanostructures.
S. aureus and
P. aeruginosa have a bacteria cell stiffness between 10–100 MPa [
37], whereas the stiffness of the 1.0_3_240 structures was 14.4 GPa. This large difference in stiffness caused the cells to be easily pierced by the TiO
2 nanostructures, causing cell death. The high killing efficiency of the nanostructured surface against
S. aureus appears practical in inhibiting initial bacteria adhesion towards gram-positive bacteria, preventing further bacterial infections on implant surfaces. This is an important finding as the killing of gram-positive bacteria cells has proven difficult [
38,
39]. At this stage, it is unclear whether the physical size of the bacteria cell has an impact on the bactericidal efficiency of the nanostructured surface. Studies currently postulate that the difference in the bactericidal efficiency of nanostructured surfaces against gram-negative and gram-positive microbes is due to the higher number of peptidoglycan layers in the cell wall of gram-positive bacteria [
36]. This results in a higher wall stiffness of the gram-positive cell, making it more difficult to kill than its gram-negative counterparts. To improve the accuracy of the plate-count method, sonication of the sample after incubation could be added to reduce the error in CFU/mL by eliminating non-adhered bacteria [
40,
41]. In addition,
P. aeruginosa cells reached a stable growth rate at approximately 14 h of incubation [
31], during which the rod-shaped cells divided, rapidly increasing the bacteria concentration in the suspension. Round
S. aureus cells took longer to approach this stable growth (approximately 15–16 h) [
32], resulting in the comparatively high growth of
P. aeruginosa colonies observed in these tests.
Sample 1.0_3_240 was the most effective surface against both
S. aureus and
P. aeruginosa. It appears that this structure height (307 nm) and morphology was effective at piercing and stretching bacteria cells, resulting in cell-wall rupture and death [
36]. This phenomenon was not observed with osteoblast cells, which remained intact. It has previously been reported that surface roughness has a large impact on the bacterial attachment of
S. aureus and
P. aeruginosa on titanium thin films [
39]. This study found that structure height was also significantly correlated to the bacteria death of these two pathogens.
Sample 1.0_3_240 produced higher cellular metabolic activity rates than the flat surfaces at 4 and 24 h. As the surface of medical implants is currently flat Ti-6Al-4V, the results indicate that osteoblast cell activity may increase with the presence of nanostructured surfaces. This enhancement of osteoblast cell growth by the nanostructured surfaces may be due to the increased surface area which improves cell migration and attachment as compared to flat surfaces [
42].
The osteoblast cellular metabolic activity was not significantly affected by the array height or surface roughness. Statistical analysis showed that surface wettability played a significant role in determining the cellular metabolic activity after 24 h. The highest metabolic activity was observed on surfaces hydrophilic in nature with a contact angle between 35–50°, supported by previous studies [
20,
43]. This finding is significant as most clinical implants are currently hydrophobic in nature [
7,
44] and can assist future implant design in improving osteoblast cell activity. This could lead to reductions in orthopaedic surgery recovery time, improving osseointegration, bone remodeling and implant fixation [
45].