NANO LETTERS
Attenuation of Protein Adsorption on Static and Oscillating Magnetostrictive Nanowires
2005 Vol. 5, No. 9 1852-1856
Kristy M. Ainslie,† Gaurav Sharma,‡ Maureen A. Dyer,† Craig A. Grimes,‡,§ and Michael V. Pishko*,†,‡,| Materials Research Institute, Department of Chemical Engineering, Department of Materials Science and Engineering, Department of Electrical Engineering, and Department of Chemistry, The PennsylVania State UniVersity, UniVersity Park, PennsylVania 16802 Received June 13, 2005; Revised Manuscript Received July 20, 2005
ABSTRACT The research described here investigates the hypothesis that nanoarchitecture contained in a nanowire array is capable of attenuating the adverse host response generated when medical devices are implanted in the body. This adverse host response, or biofouling, generates an avascular fibrous mass transfer barrier between the device and the analyte of interest, disabling the implant if it is a sensor. Numerous studies have indicated that surface chemistry and architecture modulate the host response. These findings led us to hypothesize that nanostructured surfaces will inhibit the formation of an avascular fibrous capsule significantly. We are investigating whether arrays of oscillating magnetostrictive nanowires can prevent protein adsorption. Magnetostrictive nanowires were fabricated by electroplating a ferromagnetic metal alloy into the pores of a nanoporous alumina template. The ferromagnetic nanowires are made to oscillate by oscillating the magnetic field surrounding the wires. Radiolabeled bovine serum albumin, enzyme-linked immunosorbent assay (ELISA), and other protein assays were used to study protein adhesion on the nanowire arrays. These results display a reduced protein adsorption per surface area of static nanowires. Comparing the surfaces, 14−30% of the protein that absorbed on the flat surface adsorbed on the nanowires. Our contact angle measurements indicate that the attenuation of protein on the nanowire surface might be due to the increased hydrophilicity of the nanostructured surface compared to a flat surface of the same material. We oscillated the magnetostrictive wires by placing them in a 38 G 10 Hz oscillating magnetic field. The oscillating nanowires show a further reduction in protein adhesion where only 7−67% of the protein on the static wires was measured on the oscillating nanowires. By varying the viscosity of the fluid the nanowires are oscillated in, we determined that protein detachment is shear-stress modulated. We created a high shearing fluid with dextran, which reduced protein adsorption on the oscillating nanowires by 70% over nanowires oscillating in baseline viscosity fluid. Our preliminary studies strongly suggest that the architecture in the static nanowire arrays and the shear created by oscillating the nanowire arrays would attenuate the biofouling response in vivo.
Implanted medical devices typically fail due to biofouling within one month of implantation. As in the case of subcutaneous glucose sensors, they fail due to an adverse host response that results in sensor drift, loss of sensitivity, and the need for frequent recalibration. The presence of a foreign body, such as an implanted sensor, initiates a response wherein the object would typically be degraded or extruded. If neither of these options is available, a chronic inflammation results until the object is encapsulated in fibrous tissue. Up to 21 days after implantation, proteins (e.g., albumin, fibrinogen, fibronectin, IgG, and collagen) and cells * Michael V. Pishko. 204 Fenske Laboratory, Pennsylvania State University, University Park, PA 16802-4400. Phone: (814) 863-4810 Fax: (814) 865-7846. E-mail:
[email protected]. † Department of Chemical Engineering. ‡ Department of Materials Science and Engineering. § Department of Electrical Engineering. | Department of Chemistry. 10.1021/nl051117u CCC: $30.25 Published on Web 08/26/2005
© 2005 American Chemical Society
(e.g., macrophages and red blood cells) adhere to the surface of the implant. The thickness of the protein layer is estimated from 0.5-9 µm of accumulation. Mass transfer to the glucose sensor is reduced by about 20%. After approximately 21 days post implantation, an avascular fibrous capsule will form around the implant. The capsule is characterized by a layer of giant macrophages and fibroblasts, up to 100 µm thick. The fibrous capsule separates the sensor from the vascular tissue that surrounded it at the time of implantation, leading to a significant reduction in glucose diffusion. The sensor becomes handicapped by the mass transfer resistance due to the fibrous capsule formation and increased distance from the vasculature.1 Protein adsorption on the implant surface initiates the biofouling response. Inhibition or attenuation of protein adsorption could be considered to rectify the mass transfer problems associated with biofouling. Biomolecule adsorption
Figure 1. Nanowire array as visualized with FE-SEM. The deposition conditions are 15 V AC 1000 Hz for 15 min. The alumina membrane has been etched partially in 0.2 M NaOH solution for 35 min to expose standing nanowire array. The freestanding nanowires are 4 µm in length and have 75 nm diameter.
Figure 2. Nanowire micrographs as imaged with an optical microscope are (A) in the absence of a magnetic field and (B) in the presence of a 50 G magnetic field generated from a traditional bar magnet above the microscope.
has been shown to be altered by the nanoarchitecture of the surface in vitro2-7 and in vivo.8-10 Suh et al. showed a decrease in protein adsorption on silica particles with decreasing pore size from 45 to 2.2 nm.2 Similarly, Luck et al. demonstrated that the adsorption of serum proteins was reduced on polymeric nanoparticles ranging in size from 141 to 61 nm, respectively.7 Lampin et al. concluded that increasing the surface roughness of poly(methyl methacrylate) (PMMA) to 0.20-3.4 µm resulted in increased protein adsorption.4 One reason for the altered response could be the change in the hydrophobicity of the surface. Roughening PMMA converted hydrophilic surfaces into hydrophobic ones.4 The wettability of a surface is governed by the chemical composition as well as geometrical microstructure of the surface.11-15 The influence of architecture on protein adsorption led us to hypothesize that nanofabricated surfaces would significantly inhibit the formation of an avascular fibrous capsule. To explore this hypothesis, we have created a nanowire array comprised of a Fe-Co-Ni ternary alloy imaged with SEM in Figure 1. The characterization of this array is outlined in a previous work.16 The diameter of the wires was 75 nm, which approaches the dimensions of several proteins (e.g., immunoglobin G, 45 × 23.5 nm; fibronectin, 45 × 0.6 nm; and albumin, 0.4 × 14 nm).17 The wire length is on the order of 10 µm, as are most mammalian cells. The microarchitecture element of the array will be explored in future publications. We hypothesize that nanoarchitecture in the form of nanowires within a nanoarray are capable of preventing the first steps of biofouling, protein adsorption, and this attenuation will be reduced further by placing the magnetostrictive material into an oscillating magnetic field (Figure 1). The initial steps in the testing of the above hypothesis are to adequately understand protein adsorption on these nanowire arrays. A uniform array of standing nanowires is imaged via SEM in Figure 1. To test whether the nanowires respond to a magnetic field, a magnetic bar was placed above an optical microscope (Zeiss Axiovert 200 M) with the nanowires
affixed to a glass slide focused on the stage. Figure 2 displays the nanowires first imaged in the absence of the magnetic field (A) and in the presence of a magnetic field (B). It can be clearly seen in the figure that the whorl structure imaged in A and highlighted with a white oval is resolved in the presence of the magnetic field. To test our hypothesis that the nanoarchitecture of the wires will attenuate protein adhesion, we studied protein adhesion on flat surfaces made of the same material as the nanowires (Fe-Co-Ni wafers)18 and the nanowire array.19 These results are presented in Table 1.20 For all of the radiolabeled BSA experiments, the amount of protein absorbed per square centimeter on the nanowires is significantly21 less than that on the Fe-Co-Ni wafers. For the modified ELISA22,23 (mELISA) results, the average concentration of protein on the wafers is higher and significant compared to the nanowires for all concentrations of BSA and IgG. These results were summed and normalized with respect to the amount of protein on the wafer and reflected in Table 1 as normalized. All of the normalized values are significant.21 One would expect the protein concentration surrounding the implanted nanowires to be higher than 5 mg/ mL in vivo; however, the proteins surrounding the implant would be a cocktail of several extracellular matrix proteins and not a single protein. For the illustrative use of this study, a single protein mixture is suitable for proof of concept. Bradford assays were performed to estimate the protein concentration on the surface of nanowires and Fe-Co-Ni wafers. These results indicate that the average protein concentration per area on nanowires is less than that on wafers for surfaces incubated in single protein solutions with concentrations of 1 mg/mL BSA, 1 mg/mL IgG, and 0.1 mg/ mL fibronectin. The radiolabled, mELISA, and Bradford studies indicate that the nanoarchitecture of the wires attenuates protein adsorption and the attenuation is a function of protein concentration. Captive-bubble contact angle (CA) measurements yielded advancing angles of 78° ( 3° for Fe-Co-Ni wafers and 6° ( 2° for nanowires. Both surfaces are considered hydrophilic (CA < 90°). Because the Fe-Co-Ni wafers and nanowires
Nano Lett., Vol. 5, No. 9, 2005
1853
Table 1. Protein Adsorption on Nanowires and Fe-Co-Ni Wafersa protein concentration (ng/cm2) [125I]-BSA 400 ng BSA/mL 1 mg BSA/mL 5 mg BSA/mL 1 mg IgG/mL
40 ( 8.1 30 285 ( 7921 40 086 ( 14 164.8
400 ng BSA/mL 1 mg BSA/mL 5 mg BSA/mL 1 mg IgG/mL
5 ( 1.7 9195 ( 5149 11 911 ( 10 052
ELISA Fe-Co-Ni Wafers n)9 n)3 8 ( 2.8 n)3 28 ( 10.4 20 ( 5.7 nanowires n ) 12 n)9 n)5
normalized 100% ( 20% 100% ( 26% 100% ( 35% 100% ( 28%
n ) 16 n ) 12 n)3
3.7 ( 1.5 13 ( 5 5.7 ( 3.4
25% ( 8%* 30% ( 17%* 30% ( 25%* 14% ( 8%*
n ) 27 n ) 28 n ) 31
a The percent reduction represents the amount of protein adsorbed on the nanowire surface as compared to the flat nanowire surface.20,21 A * signifies statistical significance compared to Fe-Co-Ni wafers at the same protein concentration.
Table 2. Protein Adsorption on Static and Oscillating Nanowiresa protein concentration (ng/cm2) [125I]-BSA
nanowires Pro
Pro-Osc
Pro & Osc
ELISA
400 ng BSA/mL 1 mg BSA/mL 5 mg BSA/mL 1 mg IgG/mL
5 ( 1.7 9195 ( 5149 11 911 ( 10 052
n ) 12 n)9 n)5
400 ng BSA/mL 1 mg BSA/mL 5 mg BSA/mL 1 mg IgG/mL
3 ( 0.7 655 ( 98 2943 ( 387
n)6 n)3 n)3
400 ng BSA/mL 1 mg BSA/mL 5 mg BSA/mL 1 mg IgG/mL
4 ( 1.0 964 ( 138 4400 ( 590
n)4 n)3 n)3
normalized
3.7 ( 1.5 13 ( 5 5.7 ( 3
n ) 27 n ) 28 n ) 31
100% ( 32% 100% ( 56% 100% ( 84% 100% ( 98%
1.3 ( 0.5 5.7 ( 2 5.1 ( 3
n ) 10 n ) 12 n ) 28
61% ( 13% 7% ( 1%* 25% ( 3% 55% ( 32%
1.4 ( 0.8 4.4 ( 2 5.9 ( 4
n ) 17 n ) 16 n ) 28
67% ( 18% 10% ( 2%* 37% ( 5% 65% ( 39%
a See ref 28. Oscillating nanowires were placed in a 38 g magnetic field oscillating at 10 Hz. The Pro cases are static nanowires in the presence of the indicated concentration of protein in PBS. The Pro-Osc case is that in which the nanowires were incubated in the protein solution and then taken out of the solution and oscillated in PBS. The Pro & Osc case is that in which the protein and oscillation were applied simultaneously.20, 21
have the same composition, the nanoarchitecture of the system could account for the observed increase in hydrophilicity of the nanowires over the wafers. However, literature reports that architecture usually leads to increased hydrophobicity on a surface, rather than hydrophilicity. For example, the discrete nano- and microarchitecture found in lotus leaves results in ultrahydrophobic surfaces.24 Additionally, Feng et al. observed superhydrophobic (contact angle greater than 150°) nanostructured polyacrylonitrile (PAN) nanofibers.15 Furthermore, O ¨ ner and McCarthy report that a maximum length scale of roughness on the order of 32 µm imparts superhydrophobicity.14 Theoretical derivations as well as experimental observations indicate that roughening a hydrophilic surface magnifies its hydrophilic character. The use of a modified half-infinite Ising model and manipulation of Young’s equation indicate that roughening a hydrophilic surface increases hydrophilicity.25,26 Bico et al. investigated the effect of roughness on wetting of a micropatterned hydrophilic surface. They found that increasing roughness of the hydrophilic surface did increase its wettability.27 Because the Fe-Co-Ni wafers were hydrophilic, it was expected that increasing roughness would increase hydrophilicity in accordance with Bico et 1854
al.’s results. The lower contact angle observed for the nanowires compared to the Fe-Co-Ni wafers agrees with this prediction. The increase in surface hydrophilicity of the nanowires would aid in decreasing protein adsorption on the surface of the wires compared to the Fe-Co-Ni wafers. We then tested the results of placing the nanowires in an oscillating magnetic field.28,29 It is our hypothesis that the wires, placed in an oscillating magnetic field, will oscillate causing a further reduction in protein adsorption.19 Table 2 compares the results of static nanowires and oscillating nanowires.28 Two oscillation states28 were explored to test whether the reduction in protein adsorption was due to shearing of the protein off of the surface as a result of nanowire displacement (Pro-Osc) or the movement of the wires inhibited protein adsorption (Pro & Osc). In both cases, the average attenuation in protein adsorption is less on the wires placed in an oscillating magnetic field, over the static nanowires. The reduction in protein is significant21 for the 1 mg/mL BSA case, comparing the static and oscillating with radiolabeled BSA. The trends displayed in Table 3 align with the results from Bradford assays preformed with single protein solutions with the concentrations of 1 mg/mL BSA, 1 mg/mL IgG, and 0.1 mg/mL fibronectin. Nano Lett., Vol. 5, No. 9, 2005
Figure 3. Normalized protein concentration for Pro & Osc oscillation cases at different viscosities. The control viscosity is PBS, the medium viscosity is 13 g/mL of dextran (150 000 MW), and the high viscosity is 38 g/mL of dextran. The protein concentration is normalized with respect to the control case.21 A * indicates significance compared to control viscosity.
When nonmagnetic silicon was placed in an oscillating magnetic field, the difference in protein adsorption (from 1 mg/mL BSA solution) on the static and oscillating magnetic field cases were not significantly21 different (P > 0.75). In addition, Fe-Co-Ni wafers were placed in an oscillating magnetic field and 1 mg/mL BSA solution resulting in a difference in protein adsorption that was not significant21 to the static wafers incubated with protein (P > 0.25). These controls indicate that the wire structure of the array in the oscillating field is responsible for protein adsorption attenuation. The normalized percents for all of the cases in which protein was applied, removed, and the wires placed in PBS in an oscillating magnetic field (Pro-Osc)28 were lower than cases in which the protein and oscillation were applied together (Pro & Osc).28 This indicates that the proteins desorb off of the surface as a result of nanowire movement, most likely do to shear stress. To test the theory of shear related detachment, Pro & Osc studies were performed in PBS with different viscosities. The change in protein attenuation with oscillation as a function viscosity is presented in Figure 3. Bovine serum albumin was added to the wires in the presence of PBS only. The wires were then removed from the protein solution and oscillated in the presence of control, medium (13 gm/mL of 150 000 MW dextran), or high (38 gm/mL of 150 000 MW dextran) viscosity PBS. Dextran (MW, 150 000) was added to the PBS to raise the viscosity of the solution. Dextran has been used to alter the viscosity of media in other studies.30 When comparing the control Pro & Osc response to the high viscosity Pro & Osc, the results are significant (P < 0.005). Figure 3 indicates that the reason for protein desorption or detachment from the oscillating surface is a function of the shear on the surface because surface shear is directly proportional to fluid viscosity.31 These results also might indicate that in vivo the attenuation of protein adsorption as a result of the nanowire movement might be higher than that in vitro because the interstitial fluid would be of a higher viscosity then PBS because of higher protein concentrations. Nano Lett., Vol. 5, No. 9, 2005
We have presented here results that demonstrate nanostructured surfaces, in the form of a nanowire array, attenuate protein adsorption compared to flat Fe-Co-Ni wafers comprised of the same material. The attenuation of protein on the surface of the wires might be enhanced because of the increased hydrophilicity of the nanowire surface over the wafers. Furthermore, this attenuation is increased when the magnetostrictive array is placed in an oscillating magnetic field. The protein attenuation in the presence of the oscillating magnetic field has been shown to be the result of shear modulated protein desorption. Because protein adsorption is the initial step in the biofouling process, additional studies will need to be carried out to conclude the impact the nanowire array has on cell adhesion and its behavior in vivo. Nonetheless, these preliminary studies strongly suggest that the oscillating nanowires would attenuate the biofouling response in vivo. Acknowledgment. We would like to thank Craig Baumrucker for the use of his lab and equipment for the radiolabeled BSA studies. We gratefully acknowledge funding from NIH (no. 5R01EB000684). References (1) Padera, R. F.; Colton, C. K. Biomaterials 1996, 17, 277-284. (2) Suh, C. W.; Kim, M. Y.; Choo, J. B.; Kim, J. K.; Kim, H. K.; Lee, E. K. J. Biotechnol. 2004, 112, 267-277. (3) Pallandre, A.; De Meersman, B.; Blondeau, F.; Nysten, B.; Jonas, A. M. J. Am. Chem. Soc. 2005, 127, 4320-4325. (4) Lampin, M.; Warocquier, C.; Legris, C.; Degrange, M.; Sigot-Luizard, M. F. J. Biomed. Mater. Res. 1997, 36, 99-108 (5) McFarland, C. D.; Thomas, C. H.; DeFilippis, C.; Steele, J. G.; Healy, K. E. J. Biomed. Mater. Res. 2000, 49, 200-210. (6) Denis, F. A.; Hanarp, P.; Sutherland, D. S.; Gold, J.; Mustin, C.; Rouxhet, P. G.; Dufrene, Y. F. Langmuir 2002, 18, 819-828. (7) Luck, M.; Paulke, B. R.; Schroder, W.; Blunk, T.; Muller, R. H. J. Biomed. Mater. Res. 1998, 39, 478-485. (8) Brauker, J. H.; Carrbrendel, V. E.; Martinson, L. A.; Crudele, J.; Johnston, W. D.; Johnson, R. C. J. Biomed. Mater. Res. 1995, 29, 1517-1524. (9) Rosengren, A.; Bjursten, L. M.; Danielsen, N.; Persson, H.; Kober, M. J. Mater. Sci.: Mater. Med. 1999, 10, 75-82. (10) Rosengren, A.; Bjursten, L. M. J. Biomed. Mater. Res. 2003, 67A, 918-926. (11) Gau, H.; Herminghaus, S.; Lenz, P.; Lipowsky, R. Science 1999, 283, 46-9. (12) Abbott, N. L.; Folkers, J. P.; Whitesides, G. M. Science 1992, 257, 1380. (13) Lenz, P. AdV. Mater 1999, 11, 1531. (14) Oner, D.; McCarthy, T. Langmuir 2000, 16, 7777-7782. (15) Feng, L.; Li, S.; Li, H.; Zhai, J.; Song, Y.; Jiang, L.; Zhu, D. Agnew. Chem. Int. Ed. 2002, 41, 1221-1223. (16) Sharma, G.; Grimes, C. J. Mater. Res. 2004, 19, 3695. (17) Zhang, M.; Desai, T.; Ferrari, M. Biomaterials 1998, 19, 953-60. (18) The Fe-Co-Ni wafers are flat surfaces comprised of the same material as the nanowires but lacking the architecture of the wires. To adhere to the alloy, a thin film of copper was electrodeposited on aluminum from an electrolyte solution of CuSO4‚5H2O (221.5 gm/L) and H2SO4 (52.5 gm/L) at a constant current density of 60 mA/cm2 for 15 min. The electrolyte solution for Fe-Co-Ni film deposition was the same as that used for nanowire deposition (14.055 gm/L of CoSO4‚7H2O, 52.5718 gm/L of NiSO4‚6H2O, 5.56 gm/L of FeSO4‚7H2O, and 24.7328 gm/L of H3BO3). The deposition of the alloy was carried out at a current density of 10 mA/cm2 for 30 min. (19) The proteins at concentrations of 1 mg/mL BSA, 1 mg/mL IgG, and 0.1 mg/mL fibronectin were let to absorb on the surface for 15 min. (20) BET analysis was performed to determine the surface area of the nanowire array relative to the 2D measurements of the array. The results of this analysis revealed that the surface area ratio of the 1855
nanowires to the flat surface was 10.88. The protein adsorption per surface area reported in all of the figures is that of the entire nanowire surface and not the 2D area. Analysis was performed off-site at Micromeritics Lab (Norcross, GA). (21) All of the values are reported as averages plus or minus the 95% confidence interval. Reported P-values were calculated with a twotail student T-test. A P-value of less than 0.05 was considered significant. (22) Protein adsorption on the nanowire was measured through an ELISA colorimetric technique and quantified by using a calibration standard ran in a 96-well immuno plate. The sandwich ELISA was performed in a manner similar to the manufacturer’s directions (Bethyl Labs). Surface samples were estimated by first adsorbing protein, blocking with inert protein, and attaching HRP-conjugated antibody. To quantify the surface adsorption samples, 100 µL of surface sample was taken from each sample and control group and placed in triplicate into the 96-well plate containing the standard solutions. (23) The modified ELISA (mELISA) was compared to two common techniques for studying protein adsorption on silicon, ellipsometry, and quartz crystal microbalance (QCM). The results of the ellipsometry and QCM experiments were in good agreement with the mELISA data gathered on silicon wafers. The concentration of adsorbed protein calculated via the Sauerbrey equation (Sauerbrey, G. Z. Phys. 1959, 155, 206-222) with the QCM is in very good agreement with the values determined on silicon wafers. The variable concentration values determined on silicon by mELISA are 69.7 ( 5.1 (average ( 95% CI), 102.4 ( 18.8, and 265.3 ( 10.9 ng/cm2 for the respective concentrations of 400 ng/mL, 1 and 5 mg/mL bovine serum albumin compared to the values reported for the QCM of 40.4 ( 11.3, 82.9 ( 20.1, and 147.7 ( 65.4 ng/cm2 for 400 ng/ mL, 1 and 5 mg/mL bovine serum albumin, respectively. The probability that the two samples are likely to have come from the same population is 0.27%, 25.6%, and 6.4% for the 400 ng/mL, 1 mg/mL, and 5 mg/mL concentrations, respectively. The values
1856
(24) (25) (26) (27) (28)
(29)
(30) (31)
calculated via mELISA and QCM for the 1 and 5 mg/mL results are not statistically significant from each other. Barthlott, W.; Neinhuis, C. Planta 1997, 202, 1-8. Bico, J.; Marzolin, C.; Quere, D. Europhys. Lett. 1999, 47, 220226. Borgs, C.; De Coninck, J.; Kotecky, R.; Zinque, M. Phys. ReV. Lett. 1995, 74, 2292-2294. Bico, J.; Tordeux, C.; Quere, D. Europhys. Lett. 2001, 55, 214220. To explore the effect of placing the nanowires in an oscillating magnetic field, the introduction of the protein and application of magnetic field to the nanowires was varied. The combinations were as follows: protein incubation followed by oscillation (Pro-Osc); simultaneous oscillation and protein incubation (Osc & Pro); oscillation without protein (Osc); protein incubation without vibration (Pro); and no oscillation or protein incubation (control). All of the control and Osc values were significantly less than any of the test values. To oscillate the nanowires, we placed the substrates inside of an oscillating magnetic field. The magnetic field was generated via a KEPCO (Seoul, Korea; Bipolaroperational; BOP 36-12D; 0 ( 36 V and 0 ( 12 A) in series with a Hewlett-Packard 33120A 15 MHz Function Arbitrary Waverform Generator and copper wire coil (12.5 in. diameter, 15 gauge, and 285 coils dense). The power supply was set at a fixed current of 4.3 amps, which resulted in a fluctuating voltage around 12.5 V. The waveform generator was set at 10 Hz. These settings were consistent throughout all of the experiments and resulted in a maximum magnetic field of 38 G. Ainslie, K. M.; Garanich, J. S.; Dull, R. O.; Tarbell, J. M. J. Appl. Physiol. 2005, 98, 242-9. Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena, 1st ed.; John Wiley & Sons: New York, 1960; Vol. 1, p 780.
NL051117U
Nano Lett., Vol. 5, No. 9, 2005
