Lei
Yang
a,
Yawen
Li
c,
Brian W.
Sheldon
a and
Thomas J.
Webster
*ab
aSchool of Engineering, Brown University, Providence, RI 02912, USA. E-mail: yleibrown@gmail.com; Brian_Sheldon@brown.edu
bDepartment of Orthopaedics, Brown University, Providence, RI 02912, USA. E-mail: Thomas_Webster@brown.edu; Fax: +401-863-9107; Tel: +401-863-2318
cCollege of Engineering, Lawrence Technological University, Southfield, MI 48075, USA
First published on 12th October 2011
The use of wear-resistant nanocrystalline diamond (NCD) for orthopedic and other medical device applications has received increasing interest as of late. This has led to the exploration of NCD surface modification methods to both resist and promote tissue-forming cell functions, and to further mechanistically understand such biological responses. This study represents one in a series of studies aimed to elucidate a facile and efficient method to alter NCD surface chemistry and surface free energy, and to further understand the correlation between osteoblast functions and these surface properties. Here, direct plasma treatment was used to modify NCD films with H2, O2 and NH3 gases. Surface functionalization of NCD films after the various plasma treatments were determined by X-ray photoelectron spectroscopy (XPS) and the NCD surface free energy generated by such chemical terminations were calculated by contact angle measurements. Results indicated that the plasma treatments effectively altered NCD surface chemistry and surface free energy, and a monotonic correlation between surface free energy and the hydrophilicity of the NCD films was established. Most importantly, osteoblast (bone forming cell) responses (including short-term attachment and long-term differentiation from non-calcium depositing to calcium depositing cells) on NCD of various surface free energies were evaluated, and a strong correlation between all osteoblast responses and the surface free energy of NCD was discovered for the first time. In particular, NCD which was treated with NH3 plasma had the highest free energy and exhibited the best osteoblast responses, whereas the worst osteoblast responses was found on the lowest surface free energy NCD surface which was treated by H2 plasma. A competitive mechanism of cell aggregate cohesivity and cell-substrate adhesivity is discussed to explain the surface free energy-dependent osteoblast functions on NCD. The results of this study provide an important understanding of osteoblast functions as mediated by NCD surface free energy which could be controlled by various chemical terminations. Eventually, this understanding can guide the design and fabrication of NCD for numerous biomedical applications.
Pivotal for these studies is to understand the interactions between material properties (mostly, NCD surface properties) and cells or bio-molecules as well as the mechanisms behind such interactions. For example, the unique topography and crystallinity of NCD have been proposed to have a significant impact on osteoblast (bone forming cell) adhesion, proliferation and differentiation.4,9NCD surface terminations have also revealed large effects on the wettability of NCD and, consequently, initial protein adsorption and bioactivity enhancing select cell adhesion.7,10–12
Importantly, CVD-grown NCD is usually hydrogen terminated (due to H2 used or a breakdown of hydrocarbon gases in the process), resulting in a hydrophobic surface (water contact angles usually >80–90°). Modifying NCD surfaces with fluorine terminations seldom changes NCD film wettability, while oxygen terminations can dramatically decrease water contact angles, increasing NCD hydrophilicity. To date, many studies have reported that hydrophilic oxygen-terminated NCD may improve select protein adsorption and/or cell responses compared to hydrophobic terminations, although the mechanisms behind these wettability-mediated responses have not been studied. However, in contrast to these results, studies have shown that hydrogen terminated NCD promotes MSC adhesion, proliferation, and osteoblastic differentiation over oxygen and fluorine terminated NCD.7,10 The discrepancy among these results suggests a high necessity to further investigate how NCD surface terminations affect cellular responses.
Moreover, in terms of techniques, certain surface modifications can be facile and effective to improve the biological performance of NCD films or coatings. Therefore, seeking appropriate surface terminations on NCD and understanding the relevant cellular responses are important for a wide range of biomedical applications of NCD. Specifically, for orthopaedic prosthetic applications, wear-resistant NCD can still be improved in terms of osseointegration properties, which are highly desirable for forming strong biological bonding between bone tissue and implants, ultimately, decreasing the chance of implant failure. For example, an in vitro study of as-received NCD revealed that only ∼10% of the originally seeded osteoblasts adhered to the diamond surface after 4 h; in comparison, ∼57% and ∼41% of the originally seeded osteoblasts adhered to nanostructured titanium and CoCrMo after 3 h, respectively.13
Although enhanced osteoblast or MSC osteogenic responses have been reported on oxygen terminated NCD compared to untreated NCD, even better cell responses as well as simple surface modification methods are desired. For example, a preliminary study on ammonia-hydrogen plasma treated NCD demonstrated increased 4-hour osteoblast adhesion compared to oxygen and hydrogen terminated NCD, indicating a strong potential for ammonia plasma treatment processes to promote even better osteoblast responses than other modification methods;14 however, this potential has not been systematically studied to date.
Thus, in the present study, surface chemical modification of NCD using a variety of plasmas (specifically, hydrogen, oxygen and ammonia plasmas, and resultant films were termed as NCDH, NCDO and NCDN, respectively) was developed and their impact on osteoblasts (i.e., attachment, morphology and differentiation from non-calcium depositing to calcium depositing cells) were studied in vitro. To create a focused study on NCD chemical modification alone and, thus, eliminate NCD topographical effects, NCD films with the same nanotopography (grain size 30–80 nm, root mean square roughness values ∼27nm) were carefully prepared and modified by the plasmas. The resultant wettability properties and surface energetics of the NCD films were characterized. Most importantly, the ability to alter the surface free energy of NCD by the various plasma treatments facilitated an effort to correlate osteoblast responses to the altered NCD surface free energy. As a result, a strong correlation between NCD surface free energy (resulting only from changes in surface functionalisation and not changes in roughness) and pertinent osteoblast responses was established here for the first time.
Fig. 1 SEM and AFM images of NCDH (a and d), NCDO (b and e) and NCDN (c and f). |
Substrate | RMS roughness (nm) | Contact angle measured by different liquids (°) | |||
---|---|---|---|---|---|
θ (H2O) | θ (Glycerol) | θ (Formamide) | θ (Tetradecane) | ||
a Obtained from manufacturer. b Borosilicate glass (etched by 1 M NaOH). | |||||
NCDH | 26.7 | 81 | 60 | 54 | ∼0 |
NCDO | 27.9 | 66 | 52 | 31 | ∼0 |
NCDN | 27.7 | 31 | 20 | 17 | ∼0 |
Si | ∼0.5a | 54 | 64 | 59 | 17 |
BG b | 5.5 | 61 | 68 | 36 | ∼0 |
The surface chemistry of the plasma-treated NCD films was analyzed by XPS and the spectra are shown in Fig. 2. Clear O 1s and N 1s peaks were found for NCDO and NCDN at binding energies of 531.5 eV and 397.5 eV, respectively. O 1s peaks were also observed in all the films, even the NCDH treated with pure H2 plasma; the same phenomenon has been reported in other studies which was attributed to the physiosorption of oxygen-containing molecules or surface oxidation.12,16,17 For the NCDO, most likely, the O 1s peak was attributed to C–O or C–OH groups on the diamond surface, according to numerous analyses from other studies on O2 plasma treated NCD.8,18 For the NCDN, according to Szunerits et al.,19 it is more likely to have –NH2groups bonded on NH3/H2 plasma treated diamond surfaces.
Fig. 2 XPS spectra of NCDH (a), NCDO (b) and NCDN (c). |
The controls (Si and borosilicate glass (BG)) had similar water contact angles which were higher than that of NCDN. Contact angles of tetradecane on all the samples were close to zero except a contact angle of 17° was observed on Si. Since surface tension of tetradecane has only a dispersive component (polar component ∼ 0, see Experimental section Table 2), one can expect Si to have a much lower dispersive component in its surface free energy than that of tetradecane (i.e., 26.6 mN m−1). This expectation was confirmed by surface free energy calculation as shown in Fig. 3a, in which the dispersive component in surface free energy of Si was only 6.9 mN m−1.
Fig. 3 Calculated surface free energy (a) and Zisman plots to determine critical surface free energy (γc) (b) of NCDH, NCDO, NCDN and Si and borosilicate glass (BG) controls. The γc of each substrate was read where the trend line meets cosθ = 1. |
Surface free energy results of NCD films calculated from Eqn (1) are shown in Fig. 3a. Among all the NCD samples, NCDN exhibited the highest surface free energy (62.9 mN m−1), NCDO revealed the second highest (51.8 mN m−1), and NCDH had the lowest surface free energy (40.2 mN m−1). As expected, this trend matched the trend of hydrophilicity (water contact angle) measurements. Because the topographical effects on the hydrophilicity of NCD films were suppressed by creating samples with the same nanotopography, a monotonic relationship between surface free energy and hydrophilicity of NCD films can be established by the results from this study. This monotonic relationship was not valid for the Si and BG, which had similar surface free energies (47.3 and 47.5 mN m−1, respectively) but significantly different contact angles, probably because of the large differences in their surface roughness (topography) and bulk chemistry. It was also observed that different plasmas exerted various influences on the dispersive and polar components of the surface free energy on NCD. Compared to hydrogen plasma, oxygen plasma increased both the dispersive and polar components of surface free energy on NCDO, which has also been reported by others.12 In contrast, –NH2 terminations on NCDN significantly decreased the dispersive component but dramatically increased the polar component of surface free energy compared to –H terminations.
Lastly, critical surface energy (γC) of each sample was determined by the Zisman plot shown in Fig. 3b. A linear regression of fitting the data obtained from the various test liquids gave a reading of γC when cosθ = 1, as also shown in Fig. 3b. As expected, γC of NCD films revealed the same trend as the surface free energy or hydrophilicity measurements: NCDN > NCDO > NCDH. Interestingly, BG had a higher γC compared to Si, but the surface free energy measurements showed an opposite result in which surface free energy of Si was higher. This discrepancy again suggests that surface roughness and bulk chemistry have complicated effects on material hydrophilicity, surface free energy and γC.
Fig. 4 Osteoblast attachment on NCDH, NCDO, NCDN and controls of Si and borosilicate glass (BG) after12 h. Osteoblast seeding density was 104cells cm−2. Data are expressed as the mean ± standard deviation of the mean (N = 3). *p < 0.05, **p < 0.1. |
Osteoblast density and morphology on the NCD films observed by fluorescence microscopy (Fig. 5) confirmed the above results. Specifically, osteoblast spreading morphologies on the various substrates were different and osteoblasts on NCDN and NCDO spread more than those on NCDH. Spreading area per cell was quantified on each substrate and the results are shown in Fig. 6. Each osteoblast on average spread 112% and 100% more on NCDN and NCDO than on NCDH, respectively, and osteoblasts on NCDN and NCDO also spread more than those on the controls. Osteoblasts also exhibited more and thicker contractile filaments on NCDN and NCDO than on any other sample (Fig. 5), indicating possible stronger cell-substrate interactions or active cell movements on these two substrates. In addition, osteoblasts on these two substrates had more cell-cell interactions than those on the other substrates.
Fig. 5 Fluorescence microscopy images showing osteoblast morphology on NCDH (a), NCDO (b), NCDN (c), Si (d) and borosilicate glass (e) after culturing for 12 h (bar = 100 μm). Osteoblasts were stained by rhodamine phalloidin and the nuclei were stained by DAPI. |
Fig. 6 Average spreading area of osteoblasts on NCDH, NCDO, NCDN and Si and borosilicate glass (BG) controls after culturing for 12 h. The area per cell was calculated by dividing the rhodamine phalloidin-stained cytoskeleton area by the number of DAPI-stained nuclei. Over 75 cells were measured for each substrate. Data are expressed as the mean ± standard deviation of the mean. *p < 0.01. |
Fig. 7 Alkaline phosphatase (ALP) activity of osteoblasts on NCDH, NCDO, NCDN and Si and borosilicate glass (BG) controls after 14 days. The ALP activity was normalized by total intracellular protein synthesis on the substrates. Osteoblast seeding density was 105cells cm−2. Data are expressed as the mean ± standard deviation of the mean (N = 3). *p < 0.05. |
Results of osteoblast collagen synthesis exhibited a similar trend, in which NCDN promoted osteoblast collagen synthesis the most among all the NCD films (Fig. 8). NCDO also revealed a significant promotion in collagen synthesis compared to NCDH. In this test, osteoblasts on Si revealed very high collagen synthesis, even a little higher than that on NCDN. The reason behind this promoted collagen synthesis on Si needs further investigation.
Fig. 8 Osteoblast collagen synthesis on NCDH, NCDO, NCDN and Si and borosilicate glass (BG) controls after 14 days. The collagen content was normalized by substrate area. Osteoblast seeding density was 105cells cm−2. Data are expressed as the mean ± standard deviation of the mean (N = 3). *p < 0.05. |
The most important biomarker of osteoblast differentiation is extracellular calcium deposition. The results demonstrated that NCDN enhanced osteoblast calcium deposition more than any other sample after 14 days (Fig. 9). In this test, Si and BG controls were less effective at inducing osteoblast calcium deposition than NCD films.
Fig. 9 Osteoblast extracellular calcium deposition on NCDH, NCDO, NCDN and Si and borosilicate glass (BG) controls after 14 days. The calcium deposition was normalized by total intracellular protein synthesis on the substrates. Osteoblast seeding density was 105cells cm−2. Data were expressed as the mean ± standard deviation of the mean (N = 3). *p < 0.01 and **p < 0.1. |
All the osteoblast differentiation results revealed the same trend among the three NCD films: NCDN promoted the best osteoblast functions and NCDH was the worst. Considering previous trends in osteoblast attachment (Fig. 4) and cell spreading area (Fig. 6), consistent results of osteoblast responses to plasma treated NCD were observed here. More importantly, a strong correlation between surface free energy or γC of plasma treated NCD and osteoblast responses was established. In other words, the osteoblast short and long-term functions (attachment, morphology and differentiation) on NCD had a strong dependence on NCD surface free energy or γC that can be easily altered by different surface terminations.
Stability of the surface terminations on the NCD films produced by H2, O2 or NH3 plasmas remains a question. A previous study and our experience suggests that –H terminations on NCD are stable for several months but are partially lost when cleaned by a series of solutions of acetone, ethanol and deionised water.5 A preliminary study showed that –OH and –NH2 terminations on NCD were only stable for a few weeks and the water contact angles of NCDO and NCDN increased to similar values as NCDH after storage in ambient conditions for 4 weeks.25 It was speculated that the lower stability of –OH and –NH2 terminations on NCD were due to their high chemical activities. Studies on this speculation are out of the scope of the present study, but are worth further investigation.
The most significant finding from this study was the establishment of a strong correlation between the surface energetics of NCD and osteoblast responses/functions. For years, material hydrophilicity has been speculated to influence cell-adhesive protein (fibronectin, vitronectin, etc.) adsorption and bioactivity as well as subsequent cell responses; however, no ubiquitous results have been reported.26 In the present study, variable NCD hydrophilicity and underlying surface free energy were obtained by introducing various surface terminations, and moreover, the influence of surface roughness or topography on hydrophilicity and surface free energy was eliminated through carefully designed experiments. A consistent relationship between contact angles and surface free energy of NCD revealed monotonically increased values of surface free energy and γC when NCD hydrophilicity increased. This relationship can be thermodynamically understood since the surface with higher surface free energy tends to lower its total free energy by wetting of liquid. More importantly, both short- and long-term osteoblast responses demonstrated a clear correlation to surface free energy or γC of NCD: significantly increased osteoblast attachment, spreading and differentiation were found on NCD with higher surface free energy or γC.
To explain this strong correlation, the possible differential adsorption of proteins from serum (i.e., fetal bovine serum in the culture medium) on plasma treated NCD needs to be considered. Fibronectin adsorption was quantified by an enzyme-linked immunosorbent assay (ELISA). However, no difference in fibronectin adsorption on NCDH, NCDO and NCDN was observed.25 Further investigation into the reasons behind this similar fibronectin adsorption and studies which compare the adsorption (and conformation) of other proteins need to be completed.
Instead, here we can offer a surface tensiometric approach comparing cell-cell cohesivity and cell-substrate adhesivity to understand the NCD surface free energy-dependent responses of osteoblasts. Also for cells like osteoblasts, because proliferation and differentiation (characterized by ALP activity, collagen and calcium synthesis) is closely related to their initial cell shape and spreading,4,27 the focus here was to highlight the effects of NCD surface free energy on osteoblast spreading. According to Ryan et al., competition between cell-cell cohesivity and cell-substrate adhesivity determines tissue spreading on implantable substrates and two criteria can be used to evaluate cell adhesivity to substrates: the rate (density) of attachment and the degree of cell spreading over time.28
Clearly, results from this study revealed that NCDN had the highest density of cell attachment and degree of cell spreading, suggesting the highest cell adhesivity to the NCD with the highest surface free energy (or γC). Similarly, one can obtain osteoblast adhesivity on the various NCD films following the relationship NCDN > NCDO > NCDH, which agrees with the relationship of surface free energy and γC. Therefore, NCD with higher surface free energy or γC possesses better adhesiveness to overcome the cohesivity of osteoblast aggregates, allowing osteoblasts to remain spread and consequently, to promote cell proliferation and differentiation. This relationship can be further understood by comparing cell-cell cohesivity (i.e., surface tension of cell aggregates) and the surface free energy of the substrates (i.e., surface tension between substrates and air). The aggregate cohesivity of anchorage-dependent cells is usually between ∼2 mN m−1 and ∼9 mN m−1,28,29 which is much less than the surface free energy or γC of NCD films (surface free energy between 40.2 mN m−1 and 62.9 mN m−1 and γC between 46.8 mN m−1 and 53.8 mN m−1 for NCD films in this study). Lower surface tension promotes cell aggregates to “wet” the substrate surfaces with higher surface free energy rather than clustering together, resulting in strong cell-substrate adhesivity revealed in the present study. Moreover, thermodynamically, substrates with higher surface free energy have more potential to cause “wetting” of cell aggregates on their surfaces. Thus, this increased potential explains the improved osteoblast spreading and subsequent cell functions on NCD with higher surface free energy or γC. This thermodynamic mechanism has been experimentally verified by comparing the spreading of osteoblast aggregates (aggregate sizes 100–150 μm, containing 50–100 cells) on NCDH, NCDO and NCDN.25 The results revealed that osteoblast aggregates on NCDN and NCDO spread much faster than that on NCDH, indicating strong cell-substrate adhesivity on NCD with higher surface free energy.
Based on this mechanism, the observations of less osteoblast responses to controls (BG and Si) than NCDN and NCDO may also be explained, because Si and BG had less surface free energy than NCDN and NCDO. However, this correlation did not stand when comparing to NCDH or between Si and BG, probably due to large differences in their surface roughness and bulk chemistry.
Liquid | Specifications | γL (mN m−1) | γLd (mN m−1) | γLp (mN m−1) |
---|---|---|---|---|
Deionized water | Milipore systems | 72.8 | 21.8 | 51 |
Glycerol | Sigma-Aldrich, >99% | 64 | 34 | 30 |
Formamide | Fisher, ≥99.5% | 58 | 39 | 19 |
Tetradecane | Fisher, ≥99.5% | 26.56 | 26.56 | ∼0 |
The surface free energy of the samples were calculated by the Owens-Wendt model following the method reported by Lamour et al.31 Briefly, water and formamide contact angles on the substrates of interest to this study were used to calculate the surface free energy by the following equation:
γSL = (1 + cosθ) γL = 2 (γSd γLd)1/2 + 2 (γSp γLp)1/2 | (1) |
The critical surface energy (γC, or critical surface tension) was calculated using a first order approximation of the Good-Girifalco equation.31 A linear regression analysis, also known as the Zisman plots (cosθ = f(γL)), was conducted for each sample by fitting the γL and cosθ data obtained from the different test liquids. The γC value of each sample was read where the line of the best fit approximation intersected cosθ = 1 (i.e., θ = 0).
20× images were analyzed by Image J (free resource from http://rsbweb.nih.gov/ij/) for the quantification of each osteoblast spreading area (area per cell). For each image, the total osteoblast spreading area was quantified by threshold processing on the rhodamine phalloidin stained image and cell density was counted from the DAPI stained images, and then spreading area per cell was calculated. At least three images per sample were analyzed and over 75 cells per sample were averaged.
This journal is © The Royal Society of Chemistry 2012 |