- Email: [email protected]
Effect of nitinol wire surface properties on albumin adsorption
Acta Biomaterialia 3 (2007) 103–111 www.actamat-journals.com
Effect of nitinol wire surface properties on albumin adsorption B. Clarke
a,*
, P. Kingshott
a
b,1
, X. Hou c, Y. Rochev a, A. Gorelov d, W. Carroll
a
National Centre for Biomedical Engineering Science, National University of Ireland, Galway, Ireland b Danish Polymer Centre, Risø National Laboratory, Roskilde, Denmark c Radiation Research Department, Risø National Laboratory, Roskilde, Denmark d Chemistry Department, National University of Ireland, Dublin, Ireland Received 26 April 2006; received in revised form 6 July 2006; accepted 31 July 2006
Abstract The superelastic, shape memory alloy nitinol (50% nickel and 50% titanium) is an important medical device material used for stent applications. However, the role specific surfaces properties have in protein adsorption remain controversial. In this study the effects of nitinol wire surface roughness, hydrophobicity and elemental composition upon albumin adsorption are investigated. In particular, we demonstrate that the technique of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry in the so-called surface mode can be used for the direct detection of albumin on the wire surfaces. In addition, albumin adsorbing to the wires was determined by using 125I-labelled albumin. Albumin was detected on all wire samples. Surface roughness and hydrophobicity appeared to have no effect on albumin adsorption. There was however a clear correlation between the surface nickel and oxygen concentration and the amount of albumin adsorbed. Samples with higher levels of nickel and less oxygen in the surface oxide layer of the wires showed increased albumin adsorption, which could lead to improved biocompatibility. However, nickel is a toxic substance and can cause many adverse effects on humans, and thus nitinol with a slightly enriched surface nickel concentration that does not exhibit nickel release may have potential as a medical device material. 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Protein adsorption; Albumin; Nitinol; Surface-MALDI-TOF; Radiolabelling
1. Introduction The adsorption of proteins onto biomaterial surfaces plays a key role in the way the surrounding body environment responds to the implanted material. Within seconds of implantation, protein adsorption onto the implant material will occur [1,2]. The proteins that spontaneously adsorb from the surrounding fluid medium onto the biomaterial surface after implantation are important in mediating the cellular response to the implant [3]. Albumin is important in biomedical applications as it has been identified on implants ex vivo [4]. The specific role *
Corresponding author. E-mail address: [email protected] (B. Clarke). 1 New address: The Interdisciplinary Nanoscience Centre (iNANO), University of Aarhus, Denmark.
that adsorbed albumin plays in responses to implants, however, remains controversial. While there is some evidence that shows that monocytes can and do adhere to albumin-coated surfaces in vitro, albumin is generally believed to ‘‘passivate’’ the surface and greatly reduce the acute inflammatory response [1,2]. A more recent study has shown that monocytes do not adhere to adsorbed albumin in vitro, but to macrophages via integrin receptors independent of the topography of the albumin. By topography we mean the two-dimensional structure of the adsorbed albumin layer, i.e. the formation of albumin clusters or a smooth monomolecular layer. The pro-inflammatory activity of macrophages was reduced on the albumin when compared to uncoated surfaces or surfaces coated with fibrinogen [5]. Albumin has been pre-adsorbed onto surfaces to suppress the non-specific adsorption of plasma proteins and concomitant cell–surface interactions [4]. A
1742-7061/$ - see front matter 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actbio.2006.07.006
104
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
thin layer of albumin has been found to minimise adhesion and aggregation of platelets, thus reducing the possibility of subsequent thrombus formation [6]. Surfaces that have been coated with albumin have also been associated with lower bacterial adhesion [6]. Albumin also has a low conformational stability and because of this can adsorb onto a variety of hydrophobic or hydrophilic surfaces [7]. Albumin is the predominant plasma protein in blood and accounts for 60–70% of plasma. It has been reported that supramolecular organisation of the adsorbed protein layer is controlled by both surface chemistry and topography of the implant surface [8]. Despite this, it still remains unclear if surface chemistry and topography have an effect on protein adsorption, and if so what effect they have. At small micron level roughness it has been reported that albumin adsorbs preferentially to smoother surfaces than rougher surfaces [9]. Other studies reported that the adsorption of albumin is greatly increased on porous surfaces when compared to smooth surfaces [4]. Conflicting results have also been reported on the effect of surface hydrophobicity and hydrophilicity on albumin adsorption. One group has reported that the hydrophobicity of titanium has no effect on the adsorption of albumin [4], while other groups have found that if the hydrophilicity of titanium is increased, the extent of albumin adsorption decreased [9,10]. Alloys of nickel and titanium, and in particular nitinol (50% Ni, 50% Ti), are of special interest in the medical device industry, due to their shape memory and superelastic properties. Superelasticity refers to the unusual ability of certain metals to undergo large elastic deformation [11]. The use of nitinol for medical purposes was first reported in the late 1960s [12], and today nitinol is commonly used for the manufacture of stents, which are primarily used in peripheral and coronary bypass graft interventions [13]. In the present study, the effect of nitinol wire surface roughness, hydrophobicity and elemental composition on the level of albumin adsorption is assessed. A variety of wires are used which contain varying amounts of oxygen and nickel in their oxide layers. It has been previously reported that albumin adsorption is higher on low oxygen containing surfaces [14]. A previous in vivo study has shown blood albumin levels to drop after 30 min at the site of nitinol implantation. This was believed to be due to the fact that albumin binds strongly to nickel [15]. However, since this was first proposed in 1987, no work has been carried out investigating the effect of nickel concentration on albumin adsorption. In order to obtain reliable and usable data, it is important to use samples whose surface properties mirror those used in actual stent production. As with the majority of analysis carried out on biomaterials, flat coupons are the main sample choice when assessing protein adsorption on biomaterials [4,16,17]. The aim of the present study was to estimate the degree of albumin adsorption on nitinol wire samples. The initial detection of albumin on the wires was carried out using matrix-assisted laser desorption/
ionization time-of-flight mass spectrometry in the so-called surface mode, which is more commonly known as surfaceMALDI-TOF. A detailed account of this method has previously been published [18]. To date, surface-MALDITOF has been used to analyse protein adsorption on polymeric materials, in particular contact lenses [18–21]. While this technique can be used to indicate whether proteins have adsorbed to surfaces or not, it provides little information on the amount of protein adsorbed. The most common method for quantitatively measuring protein adsorption onto surfaces is to use radioisotope labelled proteins [4,6,22]. This method is sensitive and reliable and it is the method that was used in this work to obtain quantitative data regarding albumin adsorption on nitinol wires. 2. Materials and methods 2.1. Materials All nitinol samples used in this study were prepared by Fort Wayne Metals Research Products Corporation, IN, USA, specifically for this study. Nitinol wire samples with a diameter of 0.762 mm were fabricated from a binary nickel–titanium alloy with a nominal composition of 50.8 at.% nickel and an austenite start temperature in the fully annealed condition of 31 C as measured by DSC per ASTM F 2004-00. Standard reduction and thermal processing was used to draw the wire to 1.02 mm. Additional processing to achieve 45% cold work at 0.762 mm followed by a heat straightening step to produce superelastic properties at room to body temperature were performed. The active austenite finish (Af) temperature of the final wire was measured using the bend and free recovery method per ASTM-F 2082-01 to confirm that a superelastic condition had been achieved. All specimens were cold drawn using either synthetic polycrystalline (Syn.Poly.) diamond dies or single crystal natural diamond (ND) dies. Heat straightening was performed at 500 C under various levels of an argon/oxygen atmosphere. Various levels of the argon/oxygen atmosphere were used to produce samples with varying oxide thickness. After heat straightening, two of the wires were subjected to additional chemical and mechanical treatments in order to achieve the desired surface states. Removal of the oxide by etching (E) using a proprietary Fort Wayne Metals acid solution was performed on one specimen with the intent of attacking only the oxide itself. Another specimen was exposed to a pickling process (P) after the initial etching, again using an acid solution of a proprietary nature to Fort Wayne metals. This second chemical exposure allowed attack of the base material. A second sample that was pickled was then mechanically polished (M) using a mechanical wire polishing machine fixed with abrasive pads. After treatment, samples were kept in normal atmosphere conditions. Details of the wire surface preparation procedures are outlined in Table 1. All samples
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111 Table 1 Wire surface conditions Sample no.
Surface condition
1 2 3 4 5
Syn. Poly./non-treated Syn. Poly./non-treated ND/non-treated ND/E/P/M ND/E
Wires drawn using either natural diamond wires (ND) or synthetic polycrystalline dies (Syn. Poly.). Non-treated samples were heat treated but not subjected to any additional treatments. Sample 4 was subjected to additional treatments of etching (E), pickling (P) and mechanical polishing (M). Sample 5 was subjected to the etching treatment.
were cleaned using laboratory detergent, sonicated for 15 min and then rinsed in ethanol and water. Samples were left to air dry. By air dry we mean that samples were left to dry on a bench in normal atmosphere. For the protein studies, these samples were cut to the desired lengths and further sonicated for 5 min in acetone, methanol, hexane twice and Millipore water. Samples were then left to air dry. 2.2. Surface roughness analysis A Vecco Digital Instruments Dimension 3100 atomic force microscope (AFM) was used to measure the surface roughness of the various wire samples. A triangular siliconnitride tip mounted on a cantilever (stiffness constant 0.57 N/m) was operated in contact mode. The values quoted (Rz nm) are the arithmetic averages of the absolute values of the surface height deviations measured from the mean plane within the box cursor of a total scan area of 20 · 20 lm. The width of the box was approximately 2 lm and the length 15 lm and it was aligned on the topmost part of the wire surface and parallel to the wire direction. This is illustrated in Fig. 1 which shows a 20 lm scan obtained from sample 5. The 2 · 15 lm box from where the
105
results were obtained can be seen in the middle of the scan area. The samples were analysed at a rate of 0.5 Hz; this was chosen as it is the slowest rate that samples could be analysed at. While it takes some time to analyse samples using this rate, the resolution is better and results are more accurate. Each wire sample was randomly scanned in six different locations. Rz is obtained from the z range values displayed on the AFM scan. 2.3. Water contact angle analysis Water contact angles were used to access the hydrophobicity or hydrophilicity of the nitinol surfaces. The water contact angle of the wires was measured using the sessile drop method. This method is based on axisymmetric drop surface analysis (ASDA). The angle between the baseline of the drop and the tangent at the drop boundary is measured. Measurements of contact angle were carried out in triplicate. For samples showing varying results, up to five measurements were carried out. The process used for measuring the contact angles was as follows: The sample was secured in a holder and placed on a stage with a camera and illuminator source present. A glass capillary with a small diameter was used to administer the water droplet on to wire surface. The glass capillaries were constructed from thicker capillaries (Samco) which were heated over a Bunsen burner and pulled apart to the point just before they break apart. The capillaries were then left to cool for 10 min. To make them hydrophobic, 300 lL of dimethyldichlorosilane (Fluka) was added to 1 mL of cyclohexane (Aldrich) and the solution was then pipetted down the sides of the glass capillaries. The capillaries were then left in an oven to dry for 1 h. The capillaries were then broken down to the desired length. The same capillary was used for all measurements to ensure no error occurred between measurements. The glass capillary was attached to a syringe pump so as to ensure the same size water droplet was delivered to the wire surface each time analysis was carried out. With a camera attached to a computer, the wire image in the holder could be observed on the computer screen, as could the image of the water droplet once it was administered to the wire surface. It is from this image of the water droplet on the computer screen, that the contact angle was determined. The angle the water droplet makes to the surface was measured. Analysis was carried out using Ramehart DROPimage Advanced Surface Tension Tool software. 2.4. Surface oxide analysis
Fig. 1. AFM scan for sample 5.
Auger electron Spectroscopy (AES) was used to obtain the elemental composition of the surface oxide. The samples were analysed using a Physical Electronics PHI 600 scanning Auger spectrometer. A section of each wire was ultrasonically cleaned in methylene chloride for 5 min before introduction into the Auger spectrometer. Auger electron spectroscopy was performed for each wire using
106
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
a 5 keV primary electron beam rastered over a 50 by 50 lm area. The electron beam was rastered to minimize desorption of oxygen during the analysis. Low-energy resolution survey scans were obtained from each sample to determine the elemental composition on the surface. The detection limit for elements in the oxide was 0.2 at.%. AES combined with sputter etching was used to obtain depth profiles. Auger sputter depth profiles were acquired for each sample (except sample No. 1) using a 1.0 lA, 3 keV Ar+ ion beam rastered over a 3 by 3 mm area. The sputtering rate using these parameters is estimated to be 2.5 nm min 1 based on SiO2 standard. The oxide thickness was calculated using the full-width at half-maximum (FWHM) of the oxygen signal. The sputtering time where the FWHM of the oxygen concentration occurred was used as a marker for the oxide/metal interface. At the FWHM oxygen concentration the sputtering time was recorded. This time was then used to calculate the thickness of the oxide layer using the sputter rate. The oxide layer for sample 1 was too thick for oxide thickness measurements by Auger sputter depth profiling. Thus, a section of sample 1 was mounted on a castable epoxy mounting material to provide a transverse section through the oxide layer. The mounted wire was metallographically ground and polished. The polished section was examined by scanning electron microscopy (SEM) and micrographs of the typical oxide thickness were acquired. The oxide layer thickness was measured from the digital micrographs using image analysis software. 2.5. Detection of albumin adsorption on nitinol wires Nine pieces of each wire type, cut to a length of 13 mm, were placed in the bottom of 20 ml glass vials. As a result, a total surface area of 2.8 cm2 was available for protein adsorption. Samples were then incubated with 1 ml of 1 mg/mL human serum albumin (99.9% from Sigma) in phosphate-buffered saline (PBS, pH 7.4). Samples were placed in a water bath at 37 C for 2 h. Samples were shaken at regular intervals. After 2 h the samples were removed from the water bath and the protein solution removed. Samples were then rinsed three times in PBS and twice in Millipore water. Samples were then left to air dry. Once dry the samples were then mounted on small glass slides using double-sided adhesive tape, which were then mounted onto a metal holder. 2 lL of 75% acetonitrile in 0.1% trifluoroacetic acid (TFA) was then applied to the surface of the wire samples which were then left to air dry. The acetonitrile and TFA were used as solvents for the matrix which is added to the surface. The TFA acts as an acid to protonate the proteins and minimise the effects of salt on crystal formation. The acetonitrile/TFA mix was applied prior to adding the matrix solution in order to help desorb the proteins from the surface. 3 lL of matrix solution was then added to the samples which were again left to air dry. The matrix solution was a saturated solution of sinapinic acid in 50:50 (v/v) solution of acetonitrile
and 0.1% TFA. For these experiments 20 mg of sinapinic acid was added to 0.25 ml of acetonitrile and 0.25 ml of 0.1% TFA. The matrix has a high UV absorption coefficient meaning that a large amount of energy is absorbed, allowing for conversion from solid to gas phase. During the expanding gaseous plume above the surface, photochemical reactions between the matrix and protein will also take place which leads to ionisation of the protein molecules. These are extracted and analysed by the mass spectrometer. In addition, the matrix, because of its ability to absorb the laser light, protects the proteins from direct laser damage, and thus minimal fragmentation takes place. After addition of the matrix, samples were then ready for analysis using MALDI-TOF. In addition, protein adsorption onto nitinol wire after an incubation period of 24 h was investigated using sample 4. The use of this sample was a random choice. The same procedure was used as when the incubation time was 2 h. An albumin standard was first analysed and all other samples compared to this. All samples were analysed in triplicate. A Bruker Reflex IV MALDI-TOF mass spectrometer was used in these experiments. The instrument utilises a pulsed UV laser (N2, k = 337 nm, 3 ns pulse-width) with a TOF analyser and linear detection. The data was analysed using the Xmass software supplied with the instrument. 2.6. Quantification of adsorbed albumin on nitinol wires Quantification of the amount of adsorbed albumin on the wire surfaces was performed using 125I-labelled protein, labelled using the Chloramine-T method. Albumin solution (99.9% from Sigma made up in PBS) and Na125I solution was mixed in a glass tube and the tube was then sealed with a rubber cap. Chloramine-T solution (H&S Chemical Co., Inc. Covington, USA made up in PBS) was added to the tube using a syringe and mixed with the protein solution. The labelling reaction was carried out at room temperature for 5 min, and then stopped by addition of NaHSO3 solution (prepared from Na2S2O5) (BASF Corporation, New Jersey, USA) in PBS (Sigma). 125I-labelled albumin was separated from the free iodine and other chemical reagents by gel chromatography. The reaction solution was transferred to a PD-10 desalting column (Amersham Bioscience, Buckinghamshire, UK packed with SephadexTM G-25 medium), which had been washed and conditioned with 30 ml PBS. After the loaded solution flowed out, the column was eluted with 30 ml PBS. The effluent and elute were collected in 0.5 ml glass vials. 125I activity in each vial was measured by a NaI gamma detector. The elution curve of 125 I in the gel chromatography column was then constructed. A similar experiment was carried out using nonlabelled albumin. This was detected using a UV-detector. The elution curve of albumin on the DP-10 column was also constructed. The first peak in the elution curve of 125 I activity corresponds to the labelled albumin, while the second peak corresponds to free 125I. The fractions of the first peak were combined and used as 125I labelled
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
albumin, which was transferred into 0.5 ml glass vials and stored at 20 C. The labelling efficiency was calculated to be 85%. An albumin solution was then prepared by mixing a non-labelled albumin solution with 125I labelled albumin. This was thawed at room temperature before the experiment, in PBS, which gave a specific concentration of albumin and a specific 125I activity of 5000–8000 cpm/lg albumin (measured by NaI gamma detector). Nine pieces of each wire sample, cut to a length of 13 mm, were added to each vial containing 1.0 ml of the 1.0 mg/ml albumin solution. Each sample was analysed in triplicate. Each vial was then placed into a shaker water bath for incubation at 37 C and shaken for 2 h. After 2 h the protein was removed from each vial and the samples washed four times in PBS. Samples were transferred to clean vials and once again rinsed four times in PBS. Samples were then put into new plastic vials for analysis. A NaI gamma detector (well type, Risø National Laboratory, Denmark) with a Canberra Series 20 multichannel analyser was used for the measurement of 125I activity. The amount of albumin adsorbed on the material surface (Alb in ng/cm2) was calculated from the radioactivity (A in cpm, corrected for background), the specific activity of the albumin in the adsorption solution (C, cpm/ng), and the surface area of the test material (B, cm2) by the equation Alb = A/C/B.
107
Fig. 2. Image of water droplet on sample 3. Contact angle measurement was obtained from this.
and hence were relatively hydrophilic. The surface water contact angle measurements ranged from 50.6 for sample 4 to 65.2 for sample 5. An image of a water droplet (from which contact angles were measured) on sample 3 is shown in Fig. 2. While the two roughest samples had the higher contact angles, there does not appear to be a direct correlation between surface roughness and contact angle measurements.
3. Results 3.3. Surface oxide composition and thickness 3.1. Surface roughness analysis The surface roughness values for all wire samples are presented in Table 2. Of all the samples under investigation, the non-treated sample 1 exhibited the highest average roughness value of 450 nm. The other two non-treated samples 2 and 3 had similar average roughness values of 124 and 133 nm. Sample 4 which was mechanically polished had an average roughness of 186 nm, while sample 5 which was etched had an average roughness of 278 nm. 3.2. Water contact angle analysis The results of the measurements of contact angles of water on the different wire surfaces are also presented in Table 2. All of the surfaces had contact angles below 90 Table 2 Average ± standard deviation (SD) Rz roughness values (nm) and average ± SD contact angle values () for all wire samples Sample no.
Average roughness Rz (nm) ± SD
Average water contact angle () ± SD
1 2 3 4 5
450 ± 103 124 ± 32 133 ± 63 186 ± 41 278 ± 96
63.9 ± 0.9 52.6 ± 0.8 60.5 ± 0.5 50.6 ± 1.0 65.2 ± 1.2
Table 3 summarizes the elements detected by AES surface analysis for each wire. The highest Ni concentration in the surface layer is exhibited by sample 1 (15.2%), which is significantly higher than all the other samples. Sample 2 had 3.6% Ni and sample 3 1.8% Ni. The two treated samples (4 and 5) also have nickel present, but in somewhat smaller amounts than the non-treated samples. For the treated samples, the signal from the Ni 2p region was dominated by one extremely small peak at 850 eV. The three non-treated samples also contain this peak at 850 eV (but much larger than for the treated sample) and they also contain two other smaller peaks at 730 and 780 eV. The peak at 850 eV corresponds to elemental nickel while the two peaks at 730 and 780 eV correspond to the nickel oxides NiO and Ni2O3, respectively. For all non-treated samples, the peak at 850 eV is much bigger than the other two peaks at 730 and 780 eV implying that nickel is mainly present in the elemental form. Carbon levels are relative to the O, Ni and Ti concentrations. The two samples with the higher nickel concentration have the lowest oxygen concentration. Analysis of the spectra in the Ti 2p region indicates all the titanium to be present in the form of TiO2, having binding energies at 380 and 420 eV, for all samples. Carbon is mainly present in the form of C–C bonds at 280 eV, suggesting a residual contamination, probably from the atmosphere. Other elements are present in some of the samples
108
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
Table 3 Outer surface chemical composition (at.%) and oxide thickness (nm) as determined by Auger spectroscopy for wire samples (n/d = not detected) Sample no.
Thickness (nm)
Outer surface chemical composition (at.%) Ni
Ti
O
C
Ca
Cl
K
Na
S
1 2 3 4 5
340 120 170 14.5 15
15.2 3.6 1.8 1.7 1.3
7.3 12.0 17.3 11.3 15.7
33.9 38.8 52.1 60.1 51.6
33.9 41.6 25.2 22.9 24.7
n/d n/d 0.7 1.2 3.3
4.5 0.7 0.2 0.1 0.2
n/d 0.6 0.5 n/d 0.8
2.9 1.6 1.4 2.4 1.9
2.3 1.0 0.8 0.4 0.4
but in relatively small quantities. The AES depth profile results shown in Table 3 indicate that there is a substantial difference in oxide thickness between the treated and nontreated wire samples. More than a 10-fold difference is evident when the oxide thickness of samples 1, 2, and 3 are compared with wires 4 and 5. For the non-treated wires the oxide thickness values are typical for those of air formed oxides.
respectively. From the presence of peaks mentioned above in these spectra it can be concluded that albumin has adsorbed onto the wires. These peaks were present in the
3.4. Albumin detection on nitinol wires with surface-MALDI The spectra obtained from the wire samples were compared to an albumin standard obtained by conventional MALDI, as shown in Fig. 3a. The mass spectrum of the standard albumin sample shows a main peak at m/z 66,400, with other peaks at 33,200, 13,800 and 133,500. The main peak at 66,400 corresponds to the protonated molecule [M+H]+. The peak at 33,200 correspond to the doubly protonated molecule [M+2H]2+, while the peak at 133,500 corresponds to protonated dimer [2M+H]+. The small peak at 22,200 is a triply protonated molecule [M+3H]3+. Interestingly, there is also a peak detected at m/z 13,800, which is not assignable to any charged ions of albumin. This is most likely a protein contamination of the albumin when received from the supplier. Despite being 99.9% pure as claimed by the supplier, the preparation is still likely to contain some contaminants that could adsorb to the surface. The presence of the molecular ion [M+H]+ of albumin in the mass spectrum, would be the main peak expected if albumin had adsorbed on the samples. The presences of the other charged species is dependent on the concentration in the matrix, and often are not seen. The mass spectra recorded for samples 1 and 5 after 2 h albumin adsorption are shown in Fig. 3b and 3c,
Fig. 3a. MALDI-TOF mass spectrum of an albumin standard.
Fig. 3b. Surface-MALDI-TOF mass spectrum for albumin adsorbed to sample 1.
Fig. 3c. Surface-MALDI-TOF mass spectrum for albumin adsorbed to sample 5.
Fig. 3d. Surface-MALDI-TOF mass spectrum for sample 4 after 24 h albumin adsorption.
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
Fig. 4. Amount of adsorbed albumin (ng/cm2) adsorbed to the wire samples.
spectra for the other samples (spectra not shown), also indicating that albumin had adsorbed on all the nitinol wires. For wire sample 4 which was incubated with albumin for 24 h, no adsorbed protein was detected on its surface. The mass spectrum obtained for this sample is shown in Fig. 3d. 3.5. Albumin quantification on nitinol wires The amounts of albumin adsorbed on each of the wires, as estimated using 125I labelled albumin, are shown in Fig. 4. The highest amount of albumin adsorbed on sample 1, giving a concentration of 381 ng/cm2. The next highest amount adsorbed was on sample 2 at a concentration of 329 ng/cm2. Sample 3 and sample 4 had similar amounts of adsorbed albumin at levels of 206 and 199 ng/cm2, respectively. The least amount of protein adsorbed was on sample 5. 4. Discussion The surface-MALDI-TOF mass spectra show that albumin adsorbed on all of the wire samples after 2 h incubation. The estimation of protein adsorption on metal wire samples is a new application of MALDI-TOF mass spectrometry. The advantage of the technique is the ability to detect multiply-adsorbed proteins in one simple experiment. In this study, we aimed to just show that it was possible to detect albumin adsorption on metal alloys using the method, since the success of detecting adsorbed proteins by these experiments is highly dependent on the elution of the proteins by the matrix solution. If the proteins are too tightly bound to the wire samples, then detection is reduced or not possible. In the case of the nitinol samples the proteins seemed to desorb reasonably well after 2 h incubation. Further studies using more complex solutions, such as human serum, are under way, and will provide complementary information on complex protein adsorption to stent materials. It can be seen from the higher intensity of the peaks in the mass spectra in Figs. 3b and 3c that more albumin seems to be present on sample 1 than on sample 5. Sample 2 (spectrum not shown) also appeared to have more albumin present than the treated samples. MALDI has been
109
shown to be quantitative, however, only under well controlled sample preparation conditions. In our experiments, the difference in the signal intensity is not necessarily indicative of the amount of protein adsorbed on the sample surface. Slight variations in the matrix preparation procedure can lead to non-uniform co-crystallisation of the protein with the matrix solution, which can subsequently lead to the production of irregular crystals on the surface, which can then lead to variable signal intensity in the spectra depending on the area analysed [19]. The surface-MALDI is reproducible from the point that the protein is detectable. However, due to solvent spreading (and subsequent dilution of proteins inside the resultant crystals) the technique is not quantitative. The solvent spreading is affected by the orientation of the wire samples on the holder and the packing density. Quite often it was not possible to pack them tightly together and thus some the crystals were on the sides of the wires which were still accessible to the laser. For this reason MALDI-TOF mass spectrometry alone is not sufficient for the quantification of large proteins on the surface of some biomaterials. As already mentioned, no albumin could be detected on sample 4, which was incubated with albumin overnight. The most likely explanation for this is time dependent protein denaturation resulting in the albumin molecules becoming too strongly bound to the surface for the matrix solution to facilitate elution. Many studies to date using 125 I labelled iodine to determine the extent of protein adsorption on biomaterials have involved the use of films and disc-type samples [4,6]. Here it has been shown that it is possible to successfully use the method for wire type samples and the small error between replicates demonstrates the reproducibility of the method. On examination of the surface roughness results and surface hydrophobicity results, Table 2, there appears to be no correlation between these surface parameters and the protein adsorption results, Fig. 4. From this study it appears that surface roughness in the range of 124 to 450 nm is not significant enough to have an effect on the amount of albumin adsorbed. The difference in surface water contact angles of 52 to 65 also appears to have no effect on adsorption of albumin to the surfaces of the nitinol wires, although a broader range of contact angles is necessary to truly test that hypothesis. However, it can be concluded that the albumin adsorption did not appear to be influenced by either surface hydrophobicity or roughness of nitinol wires that are produced by conventional manufacturing processes. From our results, it also appears that surface roughness had no effect on contact angle measurements. This is contrary to what others have found in the literature. Roughness and in some cases nanometre roughness, has been found to have an effect on contact angle measurements [23–25]. In our study, it is possible that it is not roughness that is the controlling factor on contact angle measurements but the surface preparation procedure and subsequently surface chemistry.
110
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
From the Auger spectroscopy results in Table 3, it can be seen that the non-treated samples have much thicker oxides than the treated samples. Samples 1 and 2, both of which were subjected to a similar preparation procedure, have very different oxide thickness. Different levels of argon/oxygen atmosphere were used for the heat treatment of these samples which resulted in oxides of varying thickness. It appears that the additional surface treatments used on samples 4 and 5 removed the thick oxide that remained after heat treatment and replaced it with a thin oxide. On examination of both the Auger spectroscopy results in Table 3 and the protein adsorption results in Fig. 4, there appears to be some correlation between the elemental composition and the albumin adsorption. Auger spectroscopy results show that the untreated wire samples, in particular sample 1 and sample 2 have much lower surface oxygen concentrations than the other samples. They also have much higher nickel concentrations in their oxide layers when compared to the other samples. It also appeared that much of this nickel is present in its elemental form. These two samples with higher nickel concentrations and lower oxygen concentrations in their oxide layers exhibit higher albumin adsorption. Sample 1 with the highest surface nickel concentration in the oxide layer and the lowest oxygen concentration, also had the highest amount of adsorbed albumin on its surface. Sample 2 with the next highest surface nickel concentration and the next lowest oxygen concentration, also has the next highest amount of albumin adsorption. Samples 3, 4, and 5 have much higher oxygen levels and lower nickel levels. When analysed statistically (multiple analysis of variance (ANOVA) with Bonferroni multiple comparison test and level of significance set at p < 0.05) these samples show no significant difference in albumin adsorption. From this study it appears that while roughness and hydrophobicity have little or no effect, surface elemental composition has a clear effect on albumin adsorption to nitinol substrates. The effect is due to either lower oxygen levels or higher nickel levels. More albumin adsorption has been found on surfaces with reduced oxygen levels when compared to samples with higher levels [14]. Human albumin contains the N-terminal sequence X-YHis that constitutes a strong physiological bonding site for nickel (II) [26]. Therefore, it is possible that surfaces with high nickel levels have a specific binding affinity for albumin, which results in higher rates of adsorption and higher adsorbed amounts of protein. One significant factor to be considered is the activity of the adsorbed albumin on the various surfaces, since its conformations, and thus accessibility to receptor cites on the molecule, may be different. Such experiments are beyond the scope of the current study. In nitinol the levels of oxygen will be related to the levels of nickel. Heat treatment of nitinol often results in surfaces with high nickel levels and lower oxygen levels. Additional treatment can remove the nickel and increase the oxygen concentration.
Previous studies have reported that high nickel concentrations in the oxide layer resulting from heat treatment can lead to increased nickel release in the physiological environment [27,28]. As much as 20–30% of the population show hypersensitivity to nickel [29] with nickel exposure being linked to genotoxic, immunotoxic, reproductive, neurotoxic and carcinogenic effects [30]. For these reasons the possibility of nickel release into implant adjacent tissues is not advisable and should be avoided. A previous study has also found that surface nickel could be a determining factor in the corrosion resistance of nitinol [31]. Corrosion can lead to degradation of the implant material, which can greatly reduce the implant functionality. It appears a balance needs to be found where nickel is present at levels sufficient to decrease oxygen concentration, promote albumin binding, but at concentrations that its release, and influence on corrosion resistance, does not cause adverse effects to the host. Additional treatment of nitinol after heat treating (sample 4 and sample 5 in this case) has been reported to increase corrosion resistance, and prevent the release of nickel ions present in the oxide layer [27,28,31]. Therefore, additional treatment of nitinol by either chemical or mechanical means may provide the critical balance in relation to the positive and negative effects of nickel content. To further explore the factors that control albumin adsorbance, a wider range of samples needs to be investigated. To find the single controlling factor, samples with varying nickel concentration and constant oxygen concentration and samples with varying oxygen concentration and constant nickel concentration need to be used. 5. Conclusion Surface-MALDI mass spectrometry and radiolabelling methods have been successfully used to detect and quantify albumin adsorption on nitinol wire samples. In this study the surface roughness and hydrophobicity seem to have little or no effect on albumin adsorption. However, some correlation was found between the level of nickel and oxygen in the surface oxide and the degree of albumin adsorption. As the nickel concentration in the oxide layers increased and the oxygen decreased, the level of albumin adsorption increased. As albumin adsorption onto a biomaterial surface is desirable after implantation of devices into the human body, control of the outer nickel and oxygen content may have an added advantage. However, it is important to note that the level of nickel in the oxide layer is not present to a degree where it can cause adverse effects to surrounding cells and tissues. Future work is needed to investigate whether it is the nickel or the oxygen that is the controlling factor in the adsorption of albumin to nitinol surfaces. Acknowledgements Financial support for this study was provided by the Research Council for Science Engineering and Technology (IRSCET): funded by the national development plan.
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
The authors also thank Dr. Maria Holmberg of at the Danish Polymer Centre, Risø National Laboratory, Roskilde, Denmark, for all her help during the visit of Brenda Clarke to Risø. References [1] Tang L, Eaton JW. Inflammatory responses to biomaterials. Am J Clin Pathol 1995;103:466–71. [2] Tang L, Eaton JW. Natural responses to unnatural materials: a molecular mechanism for foreign body reactions. Mol Med 1999;5:351–8. [3] Werthe´n M, Sellborn A, Ka¨lltorp M, Elwing H, Thomsen P. In vitro study of monocyte viability during the initial adhesion to albumin- and fibrinogen-coated surfaces. Biomaterials 2001;22: 827–32. [4] Jansson E, Tengvall P. Adsorption of albumin and IgG to porous and smooth titanium. Colloids Surf B 2004;35:45–51. [5] Brevig T, Holst B, Ademovic Z, Rozlosnik N, Røhrmann JH, Larsen NB, et al. The recognition of adsorbed and denatured proteins of different topographies by beta2 integrins and effects on leukocyte adhesion and activation. Biomaterials 2005;26:3039–53. [6] Martins MC, Wang D, Ji J, Feng L, Barbosa MA. Albumin and fibrinogen adsorption on PU-PHEMA surfaces. Biomaterials 2003;24:2067–76. [7] Giacomelli CE, Marcello JA, De Pauli CP. Adsorption of bovine serum albumin onto TiO2 particles. J Colloid Interf Sci 1997;188: 387–95. [8] Denis FA, Hanarp P, Sutherland DS, Gold J, Mustin C, Rouxhet PG, et al. Protein adsorption on model surfaces with controlled nanotopography and chemistry. Langmuir 2002;18:819–28. [9] Deligianni DD, Katsala N, Ladas S, Sotiropoulou D, Amedee J, Missirlis YF. Effect of surface roughness of the titanium alloy Ti– 6Al–4V on human bone marrow cell response and on protein adsorption. Biomaterials 2001;22:1241–51. [10] McClellan SJ, Franses EI. Adsorption of bovine serum albumin at solid/aqueous interfaces. Colloids Surf A 2005;260:265–75. [11] Duerig TW, Pelton AR, Sto¨ckel D. Superelastic nitinol for medical devices. Med Plast Biomater 1997:30. [12] Castleman LS, Motzkin SM, Alicandri FP, Bonawit VL. Biocompatibility of nitinol alloy as an implant material. J Biomed Mater Res 1976;10:695–731. [13] Walsh EG, Holton AD, Brott BC, Venugopalan R, Anagiotos AS. Magnetic resonance phase velocity mapping through NiTi stents in a flow phantom model. J Magn Reson Imaging 2004;21: 59–65. [14] Browne MM, Lubarsky GV, Davidson MR, Bradley RH. Protein adsorption onto polystyrene surfaces studied by XPS and AFM. Surf Sci 2004;553:155–67. [15] Shabalovskaya S. On the nature of the biocompatibility and on medical applications of NiTi shape memory and superelastic alloys. Bio-Med Mater Eng 2002;12:69–109.
111
[16] Zhang F, Kang ET, Neoh KG, Wang P, Tan KL. Surface modification of stainless steel by grafting of poly(ethylene glycol) for reduction in protein adsorption. Biomaterials 2001;22:1541–8. [17] Sakiyama T, Tomura J, Imamura K, Kakanishi K. Adsorption characteristics of bovine serum albumin and its peptide fragments on a stainless steel surface. Colloids Surf B Biointerf 2004;33:77–84. [18] Griesser HJ, Kingshott P, Mc Arthur SL, McLean KM, Kinsel GR, Timmons RB. Surface-MALDI mass spectrometry in biomaterials research. Biomaterials 2004;25:4861–75. [19] Oleschuk RD, McComb ME, Chow A, Ens W, Standing KG, Perreault H, et al. Characterization of plasma proteins adsorbed onto biomaterials by MALDI-TOFMS. Biomaterials 2000;16: 1701–10. [20] Kingshott P, St John HA, Griesser HJ. Direct detection of proteins adsorbed on synthetic materials by matrix-assisted laser desorption ionisation-mass spectrometry. Anal Biochem 1999;273:156–62. [21] Kingshott P, St John HA, Chatelier RC, Griesser HJ. SurfaceMALDI mass spectrometry detection of proteins adsorbed in vivo onto contact lenses. J Biomed Mater Res 2000;49:36–42. [22] Yu J, Lamba NM, Courtney JM, Whateley TL, Gaylor JD, Lowe GD, et al. Polymeric biomaterials: influence of phosphorylcholine polar groups on protein adsorption and complement activation. Int J Artif Organs 1994;17:499–504. [23] Rupp F, Scheideler L, Rehbein D, Axmann D, Geis-Gerstorfer J. Roughness induced dynamic changes of wettability of acid etched titanium implant modifications. Biomaterials 2004;25:1429–38. [24] Grundmeier G, Thiemann P, Carpentier J, Shirtcliffe N, Stratmann M. Tailoring of the morphology and chemical composition of thin organosilane microwave plasma polymer layers on metal substrates. Thin Solid Films 2004;446:61–71. [25] Miller J, Veeramasuneni S, Drelich J, Yalamanchili M, Yamauchi G. Effect of roughness as determined by atomic force microscopy on the wetting properties of PTFE thin films. Polym Eng Sci 1996;36:1849–55. [26] Bal W, Christodoulou J, Sadler PJ, Tucker A. Multi-metal binding site of serum albumin. J Inorg Biochem 1998;70:33–9. [27] Brien O’, Carroll WM, Kelly MJ. Passivation of nitinol wire for vascular implants – a demonstration of the benefits. Biomaterials 2002;23:1739–48. [28] Cisse O, Savadogo O, Wu M, Yahia L. Effect of surface treatment of NiTi alloy on its corrosion behavior in Hanks’ solution. J Biomed Mater Res 2002;61:339–45. [29] Jia W, Beatty MW, Reinhardt RA, Petro TM, Cohen DM, Maze CR, et al. Nickel release from orthodontic arch wires and cellular immune response to various nickel concentrations. J Biomed Mater Res 1999;48:488–95. [30] Wataha JC, Lockwood PE, Marek M, Ghazi M. Ability of Nicontaining biomedical alloys to activate monocytes and endothelial cells in vitro. J Biomed Mater Res 1999;45:251–7. [31] Clarke B, Carroll W, Rotchev Y, Hynes M, Bradley D, Plumley D. Influence of nitinol wire surface treatment on oxide thickness and composition and its subsequent effect on corrosion resistance and nickel ion release. J Biomed Mater Res A 2006;79:61–70.
Effect of nitinol wire surface properties on albumin adsorption B. Clarke
a,*
, P. Kingshott
a
b,1
, X. Hou c, Y. Rochev a, A. Gorelov d, W. Carroll
a
National Centre for Biomedical Engineering Science, National University of Ireland, Galway, Ireland b Danish Polymer Centre, Risø National Laboratory, Roskilde, Denmark c Radiation Research Department, Risø National Laboratory, Roskilde, Denmark d Chemistry Department, National University of Ireland, Dublin, Ireland Received 26 April 2006; received in revised form 6 July 2006; accepted 31 July 2006
Abstract The superelastic, shape memory alloy nitinol (50% nickel and 50% titanium) is an important medical device material used for stent applications. However, the role specific surfaces properties have in protein adsorption remain controversial. In this study the effects of nitinol wire surface roughness, hydrophobicity and elemental composition upon albumin adsorption are investigated. In particular, we demonstrate that the technique of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry in the so-called surface mode can be used for the direct detection of albumin on the wire surfaces. In addition, albumin adsorbing to the wires was determined by using 125I-labelled albumin. Albumin was detected on all wire samples. Surface roughness and hydrophobicity appeared to have no effect on albumin adsorption. There was however a clear correlation between the surface nickel and oxygen concentration and the amount of albumin adsorbed. Samples with higher levels of nickel and less oxygen in the surface oxide layer of the wires showed increased albumin adsorption, which could lead to improved biocompatibility. However, nickel is a toxic substance and can cause many adverse effects on humans, and thus nitinol with a slightly enriched surface nickel concentration that does not exhibit nickel release may have potential as a medical device material. 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Protein adsorption; Albumin; Nitinol; Surface-MALDI-TOF; Radiolabelling
1. Introduction The adsorption of proteins onto biomaterial surfaces plays a key role in the way the surrounding body environment responds to the implanted material. Within seconds of implantation, protein adsorption onto the implant material will occur [1,2]. The proteins that spontaneously adsorb from the surrounding fluid medium onto the biomaterial surface after implantation are important in mediating the cellular response to the implant [3]. Albumin is important in biomedical applications as it has been identified on implants ex vivo [4]. The specific role *
Corresponding author. E-mail address: [email protected] (B. Clarke). 1 New address: The Interdisciplinary Nanoscience Centre (iNANO), University of Aarhus, Denmark.
that adsorbed albumin plays in responses to implants, however, remains controversial. While there is some evidence that shows that monocytes can and do adhere to albumin-coated surfaces in vitro, albumin is generally believed to ‘‘passivate’’ the surface and greatly reduce the acute inflammatory response [1,2]. A more recent study has shown that monocytes do not adhere to adsorbed albumin in vitro, but to macrophages via integrin receptors independent of the topography of the albumin. By topography we mean the two-dimensional structure of the adsorbed albumin layer, i.e. the formation of albumin clusters or a smooth monomolecular layer. The pro-inflammatory activity of macrophages was reduced on the albumin when compared to uncoated surfaces or surfaces coated with fibrinogen [5]. Albumin has been pre-adsorbed onto surfaces to suppress the non-specific adsorption of plasma proteins and concomitant cell–surface interactions [4]. A
1742-7061/$ - see front matter 2006 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.actbio.2006.07.006
104
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
thin layer of albumin has been found to minimise adhesion and aggregation of platelets, thus reducing the possibility of subsequent thrombus formation [6]. Surfaces that have been coated with albumin have also been associated with lower bacterial adhesion [6]. Albumin also has a low conformational stability and because of this can adsorb onto a variety of hydrophobic or hydrophilic surfaces [7]. Albumin is the predominant plasma protein in blood and accounts for 60–70% of plasma. It has been reported that supramolecular organisation of the adsorbed protein layer is controlled by both surface chemistry and topography of the implant surface [8]. Despite this, it still remains unclear if surface chemistry and topography have an effect on protein adsorption, and if so what effect they have. At small micron level roughness it has been reported that albumin adsorbs preferentially to smoother surfaces than rougher surfaces [9]. Other studies reported that the adsorption of albumin is greatly increased on porous surfaces when compared to smooth surfaces [4]. Conflicting results have also been reported on the effect of surface hydrophobicity and hydrophilicity on albumin adsorption. One group has reported that the hydrophobicity of titanium has no effect on the adsorption of albumin [4], while other groups have found that if the hydrophilicity of titanium is increased, the extent of albumin adsorption decreased [9,10]. Alloys of nickel and titanium, and in particular nitinol (50% Ni, 50% Ti), are of special interest in the medical device industry, due to their shape memory and superelastic properties. Superelasticity refers to the unusual ability of certain metals to undergo large elastic deformation [11]. The use of nitinol for medical purposes was first reported in the late 1960s [12], and today nitinol is commonly used for the manufacture of stents, which are primarily used in peripheral and coronary bypass graft interventions [13]. In the present study, the effect of nitinol wire surface roughness, hydrophobicity and elemental composition on the level of albumin adsorption is assessed. A variety of wires are used which contain varying amounts of oxygen and nickel in their oxide layers. It has been previously reported that albumin adsorption is higher on low oxygen containing surfaces [14]. A previous in vivo study has shown blood albumin levels to drop after 30 min at the site of nitinol implantation. This was believed to be due to the fact that albumin binds strongly to nickel [15]. However, since this was first proposed in 1987, no work has been carried out investigating the effect of nickel concentration on albumin adsorption. In order to obtain reliable and usable data, it is important to use samples whose surface properties mirror those used in actual stent production. As with the majority of analysis carried out on biomaterials, flat coupons are the main sample choice when assessing protein adsorption on biomaterials [4,16,17]. The aim of the present study was to estimate the degree of albumin adsorption on nitinol wire samples. The initial detection of albumin on the wires was carried out using matrix-assisted laser desorption/
ionization time-of-flight mass spectrometry in the so-called surface mode, which is more commonly known as surfaceMALDI-TOF. A detailed account of this method has previously been published [18]. To date, surface-MALDITOF has been used to analyse protein adsorption on polymeric materials, in particular contact lenses [18–21]. While this technique can be used to indicate whether proteins have adsorbed to surfaces or not, it provides little information on the amount of protein adsorbed. The most common method for quantitatively measuring protein adsorption onto surfaces is to use radioisotope labelled proteins [4,6,22]. This method is sensitive and reliable and it is the method that was used in this work to obtain quantitative data regarding albumin adsorption on nitinol wires. 2. Materials and methods 2.1. Materials All nitinol samples used in this study were prepared by Fort Wayne Metals Research Products Corporation, IN, USA, specifically for this study. Nitinol wire samples with a diameter of 0.762 mm were fabricated from a binary nickel–titanium alloy with a nominal composition of 50.8 at.% nickel and an austenite start temperature in the fully annealed condition of 31 C as measured by DSC per ASTM F 2004-00. Standard reduction and thermal processing was used to draw the wire to 1.02 mm. Additional processing to achieve 45% cold work at 0.762 mm followed by a heat straightening step to produce superelastic properties at room to body temperature were performed. The active austenite finish (Af) temperature of the final wire was measured using the bend and free recovery method per ASTM-F 2082-01 to confirm that a superelastic condition had been achieved. All specimens were cold drawn using either synthetic polycrystalline (Syn.Poly.) diamond dies or single crystal natural diamond (ND) dies. Heat straightening was performed at 500 C under various levels of an argon/oxygen atmosphere. Various levels of the argon/oxygen atmosphere were used to produce samples with varying oxide thickness. After heat straightening, two of the wires were subjected to additional chemical and mechanical treatments in order to achieve the desired surface states. Removal of the oxide by etching (E) using a proprietary Fort Wayne Metals acid solution was performed on one specimen with the intent of attacking only the oxide itself. Another specimen was exposed to a pickling process (P) after the initial etching, again using an acid solution of a proprietary nature to Fort Wayne metals. This second chemical exposure allowed attack of the base material. A second sample that was pickled was then mechanically polished (M) using a mechanical wire polishing machine fixed with abrasive pads. After treatment, samples were kept in normal atmosphere conditions. Details of the wire surface preparation procedures are outlined in Table 1. All samples
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111 Table 1 Wire surface conditions Sample no.
Surface condition
1 2 3 4 5
Syn. Poly./non-treated Syn. Poly./non-treated ND/non-treated ND/E/P/M ND/E
Wires drawn using either natural diamond wires (ND) or synthetic polycrystalline dies (Syn. Poly.). Non-treated samples were heat treated but not subjected to any additional treatments. Sample 4 was subjected to additional treatments of etching (E), pickling (P) and mechanical polishing (M). Sample 5 was subjected to the etching treatment.
were cleaned using laboratory detergent, sonicated for 15 min and then rinsed in ethanol and water. Samples were left to air dry. By air dry we mean that samples were left to dry on a bench in normal atmosphere. For the protein studies, these samples were cut to the desired lengths and further sonicated for 5 min in acetone, methanol, hexane twice and Millipore water. Samples were then left to air dry. 2.2. Surface roughness analysis A Vecco Digital Instruments Dimension 3100 atomic force microscope (AFM) was used to measure the surface roughness of the various wire samples. A triangular siliconnitride tip mounted on a cantilever (stiffness constant 0.57 N/m) was operated in contact mode. The values quoted (Rz nm) are the arithmetic averages of the absolute values of the surface height deviations measured from the mean plane within the box cursor of a total scan area of 20 · 20 lm. The width of the box was approximately 2 lm and the length 15 lm and it was aligned on the topmost part of the wire surface and parallel to the wire direction. This is illustrated in Fig. 1 which shows a 20 lm scan obtained from sample 5. The 2 · 15 lm box from where the
105
results were obtained can be seen in the middle of the scan area. The samples were analysed at a rate of 0.5 Hz; this was chosen as it is the slowest rate that samples could be analysed at. While it takes some time to analyse samples using this rate, the resolution is better and results are more accurate. Each wire sample was randomly scanned in six different locations. Rz is obtained from the z range values displayed on the AFM scan. 2.3. Water contact angle analysis Water contact angles were used to access the hydrophobicity or hydrophilicity of the nitinol surfaces. The water contact angle of the wires was measured using the sessile drop method. This method is based on axisymmetric drop surface analysis (ASDA). The angle between the baseline of the drop and the tangent at the drop boundary is measured. Measurements of contact angle were carried out in triplicate. For samples showing varying results, up to five measurements were carried out. The process used for measuring the contact angles was as follows: The sample was secured in a holder and placed on a stage with a camera and illuminator source present. A glass capillary with a small diameter was used to administer the water droplet on to wire surface. The glass capillaries were constructed from thicker capillaries (Samco) which were heated over a Bunsen burner and pulled apart to the point just before they break apart. The capillaries were then left to cool for 10 min. To make them hydrophobic, 300 lL of dimethyldichlorosilane (Fluka) was added to 1 mL of cyclohexane (Aldrich) and the solution was then pipetted down the sides of the glass capillaries. The capillaries were then left in an oven to dry for 1 h. The capillaries were then broken down to the desired length. The same capillary was used for all measurements to ensure no error occurred between measurements. The glass capillary was attached to a syringe pump so as to ensure the same size water droplet was delivered to the wire surface each time analysis was carried out. With a camera attached to a computer, the wire image in the holder could be observed on the computer screen, as could the image of the water droplet once it was administered to the wire surface. It is from this image of the water droplet on the computer screen, that the contact angle was determined. The angle the water droplet makes to the surface was measured. Analysis was carried out using Ramehart DROPimage Advanced Surface Tension Tool software. 2.4. Surface oxide analysis
Fig. 1. AFM scan for sample 5.
Auger electron Spectroscopy (AES) was used to obtain the elemental composition of the surface oxide. The samples were analysed using a Physical Electronics PHI 600 scanning Auger spectrometer. A section of each wire was ultrasonically cleaned in methylene chloride for 5 min before introduction into the Auger spectrometer. Auger electron spectroscopy was performed for each wire using
106
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
a 5 keV primary electron beam rastered over a 50 by 50 lm area. The electron beam was rastered to minimize desorption of oxygen during the analysis. Low-energy resolution survey scans were obtained from each sample to determine the elemental composition on the surface. The detection limit for elements in the oxide was 0.2 at.%. AES combined with sputter etching was used to obtain depth profiles. Auger sputter depth profiles were acquired for each sample (except sample No. 1) using a 1.0 lA, 3 keV Ar+ ion beam rastered over a 3 by 3 mm area. The sputtering rate using these parameters is estimated to be 2.5 nm min 1 based on SiO2 standard. The oxide thickness was calculated using the full-width at half-maximum (FWHM) of the oxygen signal. The sputtering time where the FWHM of the oxygen concentration occurred was used as a marker for the oxide/metal interface. At the FWHM oxygen concentration the sputtering time was recorded. This time was then used to calculate the thickness of the oxide layer using the sputter rate. The oxide layer for sample 1 was too thick for oxide thickness measurements by Auger sputter depth profiling. Thus, a section of sample 1 was mounted on a castable epoxy mounting material to provide a transverse section through the oxide layer. The mounted wire was metallographically ground and polished. The polished section was examined by scanning electron microscopy (SEM) and micrographs of the typical oxide thickness were acquired. The oxide layer thickness was measured from the digital micrographs using image analysis software. 2.5. Detection of albumin adsorption on nitinol wires Nine pieces of each wire type, cut to a length of 13 mm, were placed in the bottom of 20 ml glass vials. As a result, a total surface area of 2.8 cm2 was available for protein adsorption. Samples were then incubated with 1 ml of 1 mg/mL human serum albumin (99.9% from Sigma) in phosphate-buffered saline (PBS, pH 7.4). Samples were placed in a water bath at 37 C for 2 h. Samples were shaken at regular intervals. After 2 h the samples were removed from the water bath and the protein solution removed. Samples were then rinsed three times in PBS and twice in Millipore water. Samples were then left to air dry. Once dry the samples were then mounted on small glass slides using double-sided adhesive tape, which were then mounted onto a metal holder. 2 lL of 75% acetonitrile in 0.1% trifluoroacetic acid (TFA) was then applied to the surface of the wire samples which were then left to air dry. The acetonitrile and TFA were used as solvents for the matrix which is added to the surface. The TFA acts as an acid to protonate the proteins and minimise the effects of salt on crystal formation. The acetonitrile/TFA mix was applied prior to adding the matrix solution in order to help desorb the proteins from the surface. 3 lL of matrix solution was then added to the samples which were again left to air dry. The matrix solution was a saturated solution of sinapinic acid in 50:50 (v/v) solution of acetonitrile
and 0.1% TFA. For these experiments 20 mg of sinapinic acid was added to 0.25 ml of acetonitrile and 0.25 ml of 0.1% TFA. The matrix has a high UV absorption coefficient meaning that a large amount of energy is absorbed, allowing for conversion from solid to gas phase. During the expanding gaseous plume above the surface, photochemical reactions between the matrix and protein will also take place which leads to ionisation of the protein molecules. These are extracted and analysed by the mass spectrometer. In addition, the matrix, because of its ability to absorb the laser light, protects the proteins from direct laser damage, and thus minimal fragmentation takes place. After addition of the matrix, samples were then ready for analysis using MALDI-TOF. In addition, protein adsorption onto nitinol wire after an incubation period of 24 h was investigated using sample 4. The use of this sample was a random choice. The same procedure was used as when the incubation time was 2 h. An albumin standard was first analysed and all other samples compared to this. All samples were analysed in triplicate. A Bruker Reflex IV MALDI-TOF mass spectrometer was used in these experiments. The instrument utilises a pulsed UV laser (N2, k = 337 nm, 3 ns pulse-width) with a TOF analyser and linear detection. The data was analysed using the Xmass software supplied with the instrument. 2.6. Quantification of adsorbed albumin on nitinol wires Quantification of the amount of adsorbed albumin on the wire surfaces was performed using 125I-labelled protein, labelled using the Chloramine-T method. Albumin solution (99.9% from Sigma made up in PBS) and Na125I solution was mixed in a glass tube and the tube was then sealed with a rubber cap. Chloramine-T solution (H&S Chemical Co., Inc. Covington, USA made up in PBS) was added to the tube using a syringe and mixed with the protein solution. The labelling reaction was carried out at room temperature for 5 min, and then stopped by addition of NaHSO3 solution (prepared from Na2S2O5) (BASF Corporation, New Jersey, USA) in PBS (Sigma). 125I-labelled albumin was separated from the free iodine and other chemical reagents by gel chromatography. The reaction solution was transferred to a PD-10 desalting column (Amersham Bioscience, Buckinghamshire, UK packed with SephadexTM G-25 medium), which had been washed and conditioned with 30 ml PBS. After the loaded solution flowed out, the column was eluted with 30 ml PBS. The effluent and elute were collected in 0.5 ml glass vials. 125I activity in each vial was measured by a NaI gamma detector. The elution curve of 125 I in the gel chromatography column was then constructed. A similar experiment was carried out using nonlabelled albumin. This was detected using a UV-detector. The elution curve of albumin on the DP-10 column was also constructed. The first peak in the elution curve of 125 I activity corresponds to the labelled albumin, while the second peak corresponds to free 125I. The fractions of the first peak were combined and used as 125I labelled
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
albumin, which was transferred into 0.5 ml glass vials and stored at 20 C. The labelling efficiency was calculated to be 85%. An albumin solution was then prepared by mixing a non-labelled albumin solution with 125I labelled albumin. This was thawed at room temperature before the experiment, in PBS, which gave a specific concentration of albumin and a specific 125I activity of 5000–8000 cpm/lg albumin (measured by NaI gamma detector). Nine pieces of each wire sample, cut to a length of 13 mm, were added to each vial containing 1.0 ml of the 1.0 mg/ml albumin solution. Each sample was analysed in triplicate. Each vial was then placed into a shaker water bath for incubation at 37 C and shaken for 2 h. After 2 h the protein was removed from each vial and the samples washed four times in PBS. Samples were transferred to clean vials and once again rinsed four times in PBS. Samples were then put into new plastic vials for analysis. A NaI gamma detector (well type, Risø National Laboratory, Denmark) with a Canberra Series 20 multichannel analyser was used for the measurement of 125I activity. The amount of albumin adsorbed on the material surface (Alb in ng/cm2) was calculated from the radioactivity (A in cpm, corrected for background), the specific activity of the albumin in the adsorption solution (C, cpm/ng), and the surface area of the test material (B, cm2) by the equation Alb = A/C/B.
107
Fig. 2. Image of water droplet on sample 3. Contact angle measurement was obtained from this.
and hence were relatively hydrophilic. The surface water contact angle measurements ranged from 50.6 for sample 4 to 65.2 for sample 5. An image of a water droplet (from which contact angles were measured) on sample 3 is shown in Fig. 2. While the two roughest samples had the higher contact angles, there does not appear to be a direct correlation between surface roughness and contact angle measurements.
3. Results 3.3. Surface oxide composition and thickness 3.1. Surface roughness analysis The surface roughness values for all wire samples are presented in Table 2. Of all the samples under investigation, the non-treated sample 1 exhibited the highest average roughness value of 450 nm. The other two non-treated samples 2 and 3 had similar average roughness values of 124 and 133 nm. Sample 4 which was mechanically polished had an average roughness of 186 nm, while sample 5 which was etched had an average roughness of 278 nm. 3.2. Water contact angle analysis The results of the measurements of contact angles of water on the different wire surfaces are also presented in Table 2. All of the surfaces had contact angles below 90 Table 2 Average ± standard deviation (SD) Rz roughness values (nm) and average ± SD contact angle values () for all wire samples Sample no.
Average roughness Rz (nm) ± SD
Average water contact angle () ± SD
1 2 3 4 5
450 ± 103 124 ± 32 133 ± 63 186 ± 41 278 ± 96
63.9 ± 0.9 52.6 ± 0.8 60.5 ± 0.5 50.6 ± 1.0 65.2 ± 1.2
Table 3 summarizes the elements detected by AES surface analysis for each wire. The highest Ni concentration in the surface layer is exhibited by sample 1 (15.2%), which is significantly higher than all the other samples. Sample 2 had 3.6% Ni and sample 3 1.8% Ni. The two treated samples (4 and 5) also have nickel present, but in somewhat smaller amounts than the non-treated samples. For the treated samples, the signal from the Ni 2p region was dominated by one extremely small peak at 850 eV. The three non-treated samples also contain this peak at 850 eV (but much larger than for the treated sample) and they also contain two other smaller peaks at 730 and 780 eV. The peak at 850 eV corresponds to elemental nickel while the two peaks at 730 and 780 eV correspond to the nickel oxides NiO and Ni2O3, respectively. For all non-treated samples, the peak at 850 eV is much bigger than the other two peaks at 730 and 780 eV implying that nickel is mainly present in the elemental form. Carbon levels are relative to the O, Ni and Ti concentrations. The two samples with the higher nickel concentration have the lowest oxygen concentration. Analysis of the spectra in the Ti 2p region indicates all the titanium to be present in the form of TiO2, having binding energies at 380 and 420 eV, for all samples. Carbon is mainly present in the form of C–C bonds at 280 eV, suggesting a residual contamination, probably from the atmosphere. Other elements are present in some of the samples
108
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
Table 3 Outer surface chemical composition (at.%) and oxide thickness (nm) as determined by Auger spectroscopy for wire samples (n/d = not detected) Sample no.
Thickness (nm)
Outer surface chemical composition (at.%) Ni
Ti
O
C
Ca
Cl
K
Na
S
1 2 3 4 5
340 120 170 14.5 15
15.2 3.6 1.8 1.7 1.3
7.3 12.0 17.3 11.3 15.7
33.9 38.8 52.1 60.1 51.6
33.9 41.6 25.2 22.9 24.7
n/d n/d 0.7 1.2 3.3
4.5 0.7 0.2 0.1 0.2
n/d 0.6 0.5 n/d 0.8
2.9 1.6 1.4 2.4 1.9
2.3 1.0 0.8 0.4 0.4
but in relatively small quantities. The AES depth profile results shown in Table 3 indicate that there is a substantial difference in oxide thickness between the treated and nontreated wire samples. More than a 10-fold difference is evident when the oxide thickness of samples 1, 2, and 3 are compared with wires 4 and 5. For the non-treated wires the oxide thickness values are typical for those of air formed oxides.
respectively. From the presence of peaks mentioned above in these spectra it can be concluded that albumin has adsorbed onto the wires. These peaks were present in the
3.4. Albumin detection on nitinol wires with surface-MALDI The spectra obtained from the wire samples were compared to an albumin standard obtained by conventional MALDI, as shown in Fig. 3a. The mass spectrum of the standard albumin sample shows a main peak at m/z 66,400, with other peaks at 33,200, 13,800 and 133,500. The main peak at 66,400 corresponds to the protonated molecule [M+H]+. The peak at 33,200 correspond to the doubly protonated molecule [M+2H]2+, while the peak at 133,500 corresponds to protonated dimer [2M+H]+. The small peak at 22,200 is a triply protonated molecule [M+3H]3+. Interestingly, there is also a peak detected at m/z 13,800, which is not assignable to any charged ions of albumin. This is most likely a protein contamination of the albumin when received from the supplier. Despite being 99.9% pure as claimed by the supplier, the preparation is still likely to contain some contaminants that could adsorb to the surface. The presence of the molecular ion [M+H]+ of albumin in the mass spectrum, would be the main peak expected if albumin had adsorbed on the samples. The presences of the other charged species is dependent on the concentration in the matrix, and often are not seen. The mass spectra recorded for samples 1 and 5 after 2 h albumin adsorption are shown in Fig. 3b and 3c,
Fig. 3a. MALDI-TOF mass spectrum of an albumin standard.
Fig. 3b. Surface-MALDI-TOF mass spectrum for albumin adsorbed to sample 1.
Fig. 3c. Surface-MALDI-TOF mass spectrum for albumin adsorbed to sample 5.
Fig. 3d. Surface-MALDI-TOF mass spectrum for sample 4 after 24 h albumin adsorption.
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
Fig. 4. Amount of adsorbed albumin (ng/cm2) adsorbed to the wire samples.
spectra for the other samples (spectra not shown), also indicating that albumin had adsorbed on all the nitinol wires. For wire sample 4 which was incubated with albumin for 24 h, no adsorbed protein was detected on its surface. The mass spectrum obtained for this sample is shown in Fig. 3d. 3.5. Albumin quantification on nitinol wires The amounts of albumin adsorbed on each of the wires, as estimated using 125I labelled albumin, are shown in Fig. 4. The highest amount of albumin adsorbed on sample 1, giving a concentration of 381 ng/cm2. The next highest amount adsorbed was on sample 2 at a concentration of 329 ng/cm2. Sample 3 and sample 4 had similar amounts of adsorbed albumin at levels of 206 and 199 ng/cm2, respectively. The least amount of protein adsorbed was on sample 5. 4. Discussion The surface-MALDI-TOF mass spectra show that albumin adsorbed on all of the wire samples after 2 h incubation. The estimation of protein adsorption on metal wire samples is a new application of MALDI-TOF mass spectrometry. The advantage of the technique is the ability to detect multiply-adsorbed proteins in one simple experiment. In this study, we aimed to just show that it was possible to detect albumin adsorption on metal alloys using the method, since the success of detecting adsorbed proteins by these experiments is highly dependent on the elution of the proteins by the matrix solution. If the proteins are too tightly bound to the wire samples, then detection is reduced or not possible. In the case of the nitinol samples the proteins seemed to desorb reasonably well after 2 h incubation. Further studies using more complex solutions, such as human serum, are under way, and will provide complementary information on complex protein adsorption to stent materials. It can be seen from the higher intensity of the peaks in the mass spectra in Figs. 3b and 3c that more albumin seems to be present on sample 1 than on sample 5. Sample 2 (spectrum not shown) also appeared to have more albumin present than the treated samples. MALDI has been
109
shown to be quantitative, however, only under well controlled sample preparation conditions. In our experiments, the difference in the signal intensity is not necessarily indicative of the amount of protein adsorbed on the sample surface. Slight variations in the matrix preparation procedure can lead to non-uniform co-crystallisation of the protein with the matrix solution, which can subsequently lead to the production of irregular crystals on the surface, which can then lead to variable signal intensity in the spectra depending on the area analysed [19]. The surface-MALDI is reproducible from the point that the protein is detectable. However, due to solvent spreading (and subsequent dilution of proteins inside the resultant crystals) the technique is not quantitative. The solvent spreading is affected by the orientation of the wire samples on the holder and the packing density. Quite often it was not possible to pack them tightly together and thus some the crystals were on the sides of the wires which were still accessible to the laser. For this reason MALDI-TOF mass spectrometry alone is not sufficient for the quantification of large proteins on the surface of some biomaterials. As already mentioned, no albumin could be detected on sample 4, which was incubated with albumin overnight. The most likely explanation for this is time dependent protein denaturation resulting in the albumin molecules becoming too strongly bound to the surface for the matrix solution to facilitate elution. Many studies to date using 125 I labelled iodine to determine the extent of protein adsorption on biomaterials have involved the use of films and disc-type samples [4,6]. Here it has been shown that it is possible to successfully use the method for wire type samples and the small error between replicates demonstrates the reproducibility of the method. On examination of the surface roughness results and surface hydrophobicity results, Table 2, there appears to be no correlation between these surface parameters and the protein adsorption results, Fig. 4. From this study it appears that surface roughness in the range of 124 to 450 nm is not significant enough to have an effect on the amount of albumin adsorbed. The difference in surface water contact angles of 52 to 65 also appears to have no effect on adsorption of albumin to the surfaces of the nitinol wires, although a broader range of contact angles is necessary to truly test that hypothesis. However, it can be concluded that the albumin adsorption did not appear to be influenced by either surface hydrophobicity or roughness of nitinol wires that are produced by conventional manufacturing processes. From our results, it also appears that surface roughness had no effect on contact angle measurements. This is contrary to what others have found in the literature. Roughness and in some cases nanometre roughness, has been found to have an effect on contact angle measurements [23–25]. In our study, it is possible that it is not roughness that is the controlling factor on contact angle measurements but the surface preparation procedure and subsequently surface chemistry.
110
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
From the Auger spectroscopy results in Table 3, it can be seen that the non-treated samples have much thicker oxides than the treated samples. Samples 1 and 2, both of which were subjected to a similar preparation procedure, have very different oxide thickness. Different levels of argon/oxygen atmosphere were used for the heat treatment of these samples which resulted in oxides of varying thickness. It appears that the additional surface treatments used on samples 4 and 5 removed the thick oxide that remained after heat treatment and replaced it with a thin oxide. On examination of both the Auger spectroscopy results in Table 3 and the protein adsorption results in Fig. 4, there appears to be some correlation between the elemental composition and the albumin adsorption. Auger spectroscopy results show that the untreated wire samples, in particular sample 1 and sample 2 have much lower surface oxygen concentrations than the other samples. They also have much higher nickel concentrations in their oxide layers when compared to the other samples. It also appeared that much of this nickel is present in its elemental form. These two samples with higher nickel concentrations and lower oxygen concentrations in their oxide layers exhibit higher albumin adsorption. Sample 1 with the highest surface nickel concentration in the oxide layer and the lowest oxygen concentration, also had the highest amount of adsorbed albumin on its surface. Sample 2 with the next highest surface nickel concentration and the next lowest oxygen concentration, also has the next highest amount of albumin adsorption. Samples 3, 4, and 5 have much higher oxygen levels and lower nickel levels. When analysed statistically (multiple analysis of variance (ANOVA) with Bonferroni multiple comparison test and level of significance set at p < 0.05) these samples show no significant difference in albumin adsorption. From this study it appears that while roughness and hydrophobicity have little or no effect, surface elemental composition has a clear effect on albumin adsorption to nitinol substrates. The effect is due to either lower oxygen levels or higher nickel levels. More albumin adsorption has been found on surfaces with reduced oxygen levels when compared to samples with higher levels [14]. Human albumin contains the N-terminal sequence X-YHis that constitutes a strong physiological bonding site for nickel (II) [26]. Therefore, it is possible that surfaces with high nickel levels have a specific binding affinity for albumin, which results in higher rates of adsorption and higher adsorbed amounts of protein. One significant factor to be considered is the activity of the adsorbed albumin on the various surfaces, since its conformations, and thus accessibility to receptor cites on the molecule, may be different. Such experiments are beyond the scope of the current study. In nitinol the levels of oxygen will be related to the levels of nickel. Heat treatment of nitinol often results in surfaces with high nickel levels and lower oxygen levels. Additional treatment can remove the nickel and increase the oxygen concentration.
Previous studies have reported that high nickel concentrations in the oxide layer resulting from heat treatment can lead to increased nickel release in the physiological environment [27,28]. As much as 20–30% of the population show hypersensitivity to nickel [29] with nickel exposure being linked to genotoxic, immunotoxic, reproductive, neurotoxic and carcinogenic effects [30]. For these reasons the possibility of nickel release into implant adjacent tissues is not advisable and should be avoided. A previous study has also found that surface nickel could be a determining factor in the corrosion resistance of nitinol [31]. Corrosion can lead to degradation of the implant material, which can greatly reduce the implant functionality. It appears a balance needs to be found where nickel is present at levels sufficient to decrease oxygen concentration, promote albumin binding, but at concentrations that its release, and influence on corrosion resistance, does not cause adverse effects to the host. Additional treatment of nitinol after heat treating (sample 4 and sample 5 in this case) has been reported to increase corrosion resistance, and prevent the release of nickel ions present in the oxide layer [27,28,31]. Therefore, additional treatment of nitinol by either chemical or mechanical means may provide the critical balance in relation to the positive and negative effects of nickel content. To further explore the factors that control albumin adsorbance, a wider range of samples needs to be investigated. To find the single controlling factor, samples with varying nickel concentration and constant oxygen concentration and samples with varying oxygen concentration and constant nickel concentration need to be used. 5. Conclusion Surface-MALDI mass spectrometry and radiolabelling methods have been successfully used to detect and quantify albumin adsorption on nitinol wire samples. In this study the surface roughness and hydrophobicity seem to have little or no effect on albumin adsorption. However, some correlation was found between the level of nickel and oxygen in the surface oxide and the degree of albumin adsorption. As the nickel concentration in the oxide layers increased and the oxygen decreased, the level of albumin adsorption increased. As albumin adsorption onto a biomaterial surface is desirable after implantation of devices into the human body, control of the outer nickel and oxygen content may have an added advantage. However, it is important to note that the level of nickel in the oxide layer is not present to a degree where it can cause adverse effects to surrounding cells and tissues. Future work is needed to investigate whether it is the nickel or the oxygen that is the controlling factor in the adsorption of albumin to nitinol surfaces. Acknowledgements Financial support for this study was provided by the Research Council for Science Engineering and Technology (IRSCET): funded by the national development plan.
B. Clarke et al. / Acta Biomaterialia 3 (2007) 103–111
The authors also thank Dr. Maria Holmberg of at the Danish Polymer Centre, Risø National Laboratory, Roskilde, Denmark, for all her help during the visit of Brenda Clarke to Risø. References [1] Tang L, Eaton JW. Inflammatory responses to biomaterials. Am J Clin Pathol 1995;103:466–71. [2] Tang L, Eaton JW. Natural responses to unnatural materials: a molecular mechanism for foreign body reactions. Mol Med 1999;5:351–8. [3] Werthe´n M, Sellborn A, Ka¨lltorp M, Elwing H, Thomsen P. In vitro study of monocyte viability during the initial adhesion to albumin- and fibrinogen-coated surfaces. Biomaterials 2001;22: 827–32. [4] Jansson E, Tengvall P. Adsorption of albumin and IgG to porous and smooth titanium. Colloids Surf B 2004;35:45–51. [5] Brevig T, Holst B, Ademovic Z, Rozlosnik N, Røhrmann JH, Larsen NB, et al. The recognition of adsorbed and denatured proteins of different topographies by beta2 integrins and effects on leukocyte adhesion and activation. Biomaterials 2005;26:3039–53. [6] Martins MC, Wang D, Ji J, Feng L, Barbosa MA. Albumin and fibrinogen adsorption on PU-PHEMA surfaces. Biomaterials 2003;24:2067–76. [7] Giacomelli CE, Marcello JA, De Pauli CP. Adsorption of bovine serum albumin onto TiO2 particles. J Colloid Interf Sci 1997;188: 387–95. [8] Denis FA, Hanarp P, Sutherland DS, Gold J, Mustin C, Rouxhet PG, et al. Protein adsorption on model surfaces with controlled nanotopography and chemistry. Langmuir 2002;18:819–28. [9] Deligianni DD, Katsala N, Ladas S, Sotiropoulou D, Amedee J, Missirlis YF. Effect of surface roughness of the titanium alloy Ti– 6Al–4V on human bone marrow cell response and on protein adsorption. Biomaterials 2001;22:1241–51. [10] McClellan SJ, Franses EI. Adsorption of bovine serum albumin at solid/aqueous interfaces. Colloids Surf A 2005;260:265–75. [11] Duerig TW, Pelton AR, Sto¨ckel D. Superelastic nitinol for medical devices. Med Plast Biomater 1997:30. [12] Castleman LS, Motzkin SM, Alicandri FP, Bonawit VL. Biocompatibility of nitinol alloy as an implant material. J Biomed Mater Res 1976;10:695–731. [13] Walsh EG, Holton AD, Brott BC, Venugopalan R, Anagiotos AS. Magnetic resonance phase velocity mapping through NiTi stents in a flow phantom model. J Magn Reson Imaging 2004;21: 59–65. [14] Browne MM, Lubarsky GV, Davidson MR, Bradley RH. Protein adsorption onto polystyrene surfaces studied by XPS and AFM. Surf Sci 2004;553:155–67. [15] Shabalovskaya S. On the nature of the biocompatibility and on medical applications of NiTi shape memory and superelastic alloys. Bio-Med Mater Eng 2002;12:69–109.
111
[16] Zhang F, Kang ET, Neoh KG, Wang P, Tan KL. Surface modification of stainless steel by grafting of poly(ethylene glycol) for reduction in protein adsorption. Biomaterials 2001;22:1541–8. [17] Sakiyama T, Tomura J, Imamura K, Kakanishi K. Adsorption characteristics of bovine serum albumin and its peptide fragments on a stainless steel surface. Colloids Surf B Biointerf 2004;33:77–84. [18] Griesser HJ, Kingshott P, Mc Arthur SL, McLean KM, Kinsel GR, Timmons RB. Surface-MALDI mass spectrometry in biomaterials research. Biomaterials 2004;25:4861–75. [19] Oleschuk RD, McComb ME, Chow A, Ens W, Standing KG, Perreault H, et al. Characterization of plasma proteins adsorbed onto biomaterials by MALDI-TOFMS. Biomaterials 2000;16: 1701–10. [20] Kingshott P, St John HA, Griesser HJ. Direct detection of proteins adsorbed on synthetic materials by matrix-assisted laser desorption ionisation-mass spectrometry. Anal Biochem 1999;273:156–62. [21] Kingshott P, St John HA, Chatelier RC, Griesser HJ. SurfaceMALDI mass spectrometry detection of proteins adsorbed in vivo onto contact lenses. J Biomed Mater Res 2000;49:36–42. [22] Yu J, Lamba NM, Courtney JM, Whateley TL, Gaylor JD, Lowe GD, et al. Polymeric biomaterials: influence of phosphorylcholine polar groups on protein adsorption and complement activation. Int J Artif Organs 1994;17:499–504. [23] Rupp F, Scheideler L, Rehbein D, Axmann D, Geis-Gerstorfer J. Roughness induced dynamic changes of wettability of acid etched titanium implant modifications. Biomaterials 2004;25:1429–38. [24] Grundmeier G, Thiemann P, Carpentier J, Shirtcliffe N, Stratmann M. Tailoring of the morphology and chemical composition of thin organosilane microwave plasma polymer layers on metal substrates. Thin Solid Films 2004;446:61–71. [25] Miller J, Veeramasuneni S, Drelich J, Yalamanchili M, Yamauchi G. Effect of roughness as determined by atomic force microscopy on the wetting properties of PTFE thin films. Polym Eng Sci 1996;36:1849–55. [26] Bal W, Christodoulou J, Sadler PJ, Tucker A. Multi-metal binding site of serum albumin. J Inorg Biochem 1998;70:33–9. [27] Brien O’, Carroll WM, Kelly MJ. Passivation of nitinol wire for vascular implants – a demonstration of the benefits. Biomaterials 2002;23:1739–48. [28] Cisse O, Savadogo O, Wu M, Yahia L. Effect of surface treatment of NiTi alloy on its corrosion behavior in Hanks’ solution. J Biomed Mater Res 2002;61:339–45. [29] Jia W, Beatty MW, Reinhardt RA, Petro TM, Cohen DM, Maze CR, et al. Nickel release from orthodontic arch wires and cellular immune response to various nickel concentrations. J Biomed Mater Res 1999;48:488–95. [30] Wataha JC, Lockwood PE, Marek M, Ghazi M. Ability of Nicontaining biomedical alloys to activate monocytes and endothelial cells in vitro. J Biomed Mater Res 1999;45:251–7. [31] Clarke B, Carroll W, Rotchev Y, Hynes M, Bradley D, Plumley D. Influence of nitinol wire surface treatment on oxide thickness and composition and its subsequent effect on corrosion resistance and nickel ion release. J Biomed Mater Res A 2006;79:61–70.