pubs.acs.org/Langmuir © 2010 American Chemical Society
Anti-nonspecific Protein Adsorption Properties of Biomimetic Glycocalyx-like Glycopolymer Layers: Effects of Glycopolymer Chain Density and Protein Size Qian Yang, Christian Kaul, and Mathias Ulbricht* Lehrstuhl f€ ur Technische Chemie II, Universit€ at Duisburg-Essen, 45117 Essen, Germany Received October 14, 2009. Revised Manuscript Received December 23, 2009 In many cases, biomaterials surfaces are desired to be resistant to protein adsorption. A system fulfilling this task in nature is the so-called glycocalyx. The glycocalyx is an outer layer on the cell membrane with bound glycoproteins and glycolipids, exposing a pattern of carbohydrate groups. There is a growing interest to mimic this glycocalyx layer to have a tool to overcome the problems with uncontrolled protein adsorption on biomaterials. In this work a glycocalyx-like layer is artificially imitated by surface-initiated atom transfer radical polymerization (ATRP) of a glycomonomer, D-gluconamidoethyl methacrylate (GAMA), from a mixed self-assembled monolayer (SAM) of an ATRP initiatorimmobilized hydroxyl-terminated thiol and a methyl-terminated thiol as diluent. Fourier transform infrared spectroscopy (FT/IR-ATR), contact angle, and ellipsometry measurements were employed to confirm the grafting of the glycopolymer. The anti-nonspecific protein binding properties of this glycopolymer layer were then investigated with surface plasmon resonance (SPR). Three proteins with different size, lysozyme, bovine serum albumin (BSA), and fibrinogen were used as model solutes to investigate the influence of protein size on the protein resistance behavior. The glycopolymer chain density was controlled during surface-initiated ATRP by varying the ratio of the components in the mixed SAM, and the chain length was adjusted by ATRP time. The effect of chain density in combination with the protein size was also evaluated. The most important results are that poly(GAMA) layers of higher grafting density show resistance to adsorption of the model proteins used in this work and that the amount of adsorbed protein depends on the length and density of the glycopolymer chains and also on the size of the proteins.
Introduction In recent years, due to the increase in human life expectancy, there is an increasing demand for biomaterials as a whole or part of a system which treats, augments, or replaces any tissue and/or organ or function of the body.1 On the other hand, synthetic biomaterials have already been widely used in various fields for different purposes including implants, biosensors, and artificial organs. Synthetic polymers remain the most versatile class of biomaterials because of the ease in controlling their compositions, structures, and properties.2-4 However, polymers always suffer from the nonspecific adsorption of proteins from the surrounding body liquids, and within the first few seconds of contact proteins are deposited. This protein deposition controls further cell- and protein-surface interactions and often leads to fatal side effects such as thrombosis, infections, and other complications.5-7 In the past decade a large number of studies have been conducted to *To whom all correspondences should be addressed. E-mail: mathias.ulbricht@ uni-essen.de. (1) Klee, D.; H€ocker, H. Adv. Polym. Sci. 2000, 149, 1–56. (2) Griffith, L. G. Acta Mater. 2000, 48, 263–277. (3) Langer, R.; Tirrell, D. A. Nature 2004, 428, 487–492. (4) Metzke, M.; O’Connor, N.; Maiti, S.; Nelson, E.; Guan, Z. Angew. Chem., Int. Ed. 2005, 44, 6529–6533. (5) Goodman, S. L.; Tweden, K. S.; Albrecht, R. M. J. Biomed. Mater. Res. 1996, 32, 249–258. (6) Frank, R. D.; Dresbach, H.; Thelen, H.; Sieberth, H. G. J. Biomed. Mater. Res. 2000, 52, 374–381. (7) Douglas, L. J. Trends Microbiol. 2003, 11, 30–36. (8) Cho, W. K.; Kong, B. Y.; Choi, I. S. Langmuir 2007, 23, 5678–5682. (9) Sethuraman, A.; Han, M.; Kane, R. S.; Belfort, G. Langmuir 2004, 20, 7779– 7788. (10) Kim, H.; Doh, J.; Irvine, D. J.; Cohen, R. E.; Hammond, P. T. Biomacromolecules 2004, 5, 822–827. (11) Holzl, M.; Tinazli, A.; Leitner, C.; Hahn, C. D.; Lackner, B.; Tampe, R.; Gruber, H. J. Langmuir 2007, 23, 5571–5577.
5746 DOI: 10.1021/la903895q
identify surfaces with low protein adsorption tendency.8-15 Several theories have been established to explain the protein resistance, but unfortunately, the fundamental mechanism at the molecular level is still unknown.16-19 On the other hand, it has been proven that modification of a biomaterial surface with a chemical or a biological substance can reduce/prevent surface adhesion phenomena, and many efforts have been made by using both physical and chemical methods. Among them, surface graft copolymerization is one of the most commonly used and effective strategy; initiation methods include radical, high-energy radiation, plasma or UV initiated graft copolymerization,20 and the newly developed surface-initiated ATRP.21 In nature, organisms solve the problem of nonspecific adsorption and adhesion very well. On the outmost surface of cells are glycosylated species (polysaccharides such as glycoproteins, glycolipids, and other glyco-conjugates), called glycocalyx.22 The (12) Ostuni, E.; Chapman, R. G.; Holmlin, R. E.; Takayama, S.; Whitesides, G. M. Langmuir 2001, 17, 5605–5620. (13) Prime, K. L.; Whitesides, G. M. J. Am. Chem. Soc. 1993, 115, 10714–10721. (14) Khademhosseini, A.; Jon, S.; Suh, K. Y.; Tran, T. N. T.; Eng, G.; Yeh, J.; Seong, J.; Langer, R. Adv. Mater. 2003, 15, 1995–2000. (15) Ma, H.; Wells, M.; Beebe, T. P., Jr.; Chilkoti, A. Adv. Funct. Mater. 2006, 16, 640–648. (16) Benhabbour, S. R.; Sheardown, H.; Adronov, A. Macromolecules 2008, 41, 4817–4823. (17) Halperin, A. Langmuir 1999, 15, 2525–2533. (18) (a) Zolk, M.; Eisert, F.; Pipper, J.; Herrwerth, S.; Eck, W.; Buck, M.; Grunze, M. Langmuir 2000, 16, 5849–5852. (b) Pertsin, A. J.; Grunze, M. Langmuir 2000, 16, 8829–8841. (19) Kane, R. S.; Deschatelets, P.; Whitesides, G. M. Langmuir 2003, 19, 2388– 2391. (20) Susanto, H.; Ulbricht, M. Langmuir 2007, 23, 7818–7830. (21) Zhang, F.; Xu, F. J.; Kang, E. T.; Neoh, K. G. Ind. Eng. Chem. Res. 2006, 45, 3067–3073. (22) Holland, N. B.; Qiu, Y.; Ruegsegger, M.; Marchant, R. E. Nature 1998, 392, 799–801.
Published on Web 01/28/2010
Langmuir 2010, 26(8), 5746–5752
Yang et al.
Article
Scheme 1. Schematic Flow Diagram Illustrating the Processes of Constructing Grafted Poly(GAMA) Layer on SPR Sensor Surface (a) and Control of Chain Density by Varying Functional Groups Density in SAM (b)
glycocalix forms a dense layer on the surface of the cell membrane and dominates the intercellular specific interaction and, at the same time, contributes to the steric repulsion that prevents nonspecific adsorption/adhesion of other molecules/particles. For example, it is known that the glycocalyx layer on endothelial cells plays an important role in maintaining the nonadhesive/ nonthrombogenic property of the native intravascular wall.23 Therefore, the construction of a carbohydrate-rich structure on a biomaterial offers a possibility for suppressing nonspecific protein adsorption, and there is a growing interest to mimic this glycocalyx layer to overcome the protein adsorption problem on biomaterials.12,22,24-26 Luk et al.25 reported that a SAM containing mannitol end groups has high resistance to several different proteins, and the anti-nonspecific property is indistinguishable from a monolayer with tri(ethylene glycol) end groups. Gupta et al.23 synthesized a so-called surfactant polymer including a flexible poly(vinylamine) backbone and oligosaccharide side chains. This polymer formed a stable adsorbed layer and suppressed plasma protein adsorption by at least 90% compared with the bare graphite surface. Here we report on the evaluation of the anti-nonspecific adsorption property of grafted glycopolymer layers by SPR sensor measurements. Scheme 1 shows the processes of constructing poly(GAMA) layers on the SPR sensor surface and of controlling the chain density by varying functional group density in the SAM. These functionalizations are based on using a controlled, “pseudo-living” polymerization, ATRP, for a surfaceinitiated synthesis of grafted functional polymer layers (see, e.g., refs 27 and 28). First, the gold sensor surface was covered by a SAM layer with hydroxyl end groups and subsequently immobilized (23) Gupta, A. S.; Wang, S.; Linka, E.; Andersone, E. H.; Hofmann, C.; Lewandowski, J.; Kottke-Marchant, K.; Marchant, R. E. Biomaterials 2006, 27, 3084–3095. (24) (a) Chapman, R. G.; Ostuni, E.; Takayama, S.; Holmlin, R. E.; Yan, L.; Whitesides, G. M. J. Am. Chem. Soc. 2000, 122, 8303–8304. (b) Ostuni, E.; Chapman, R. G.; Liang, M. N.; Meluleni, G.; Pier, G.; Ingber, D. E.; Whitesides, G. M. Langmuir 2001, 17, 6336–6343. (c) Chapman, R. G.; Ostuni, E.; Yan, L.; Whitesides, G. M. Langmuir 2000, 16, 6927–6936. (25) Luk, Y. Y.; Kato, M.; Mrksich, M. Langmuir 2000, 16, 9604–9608. (26) Frazier, R. A.; Matthijs, G.; Davies, M. C.; Roberts, C. J.; Schacht, E.; Tendler, S. J. B. Biomaterials 2000, 21, 957–966. (27) Kim, J., B.; Bruening, M. L.; Baker, G. L. J. Am. Chem. Soc. 2000, 122, 7616–7617. (28) Braunecker, W. A.; Matyjaszewski, K. Prog. Polym. Sci. 2007, 32, 93–146.
Langmuir 2010, 26(8), 5746–5752
with the ATRP initiator. Then, a glycopolymer with glucitol side groups was grafted to the sensor surface by surface-initiated ATRP. The chain length and chain density were well controlled by varying ATRP time and initiator density, respectively. After that, BSA was used as model protein to evaluate the protein resistance property of this glycocalyx-like layer. The effect of protein size was investigated by comparing the adsorption of BSA with adsorption of lysozyme and fibrinogen. Moreover, the influence of glycopolymer chain density on the anti-nonspecific adsorption property was also studied.
Experimental Section Materials. SPR sensors with a diameter of 25 mm and a goldcoated area of 18 mm were purchased from Xantec Bioanalytics GmbH (Germany). D-Gluconamidoethyl methacrylate (GAMA) was synthesized in our lab as described previously.29 Acetonitrile was purified by refluxing with boric anhydride and distillation before use. Copper(I) bromide (Aldrich, 99.999%) and copper(II) bromide (Acros, 99þ%) were commercial products and used without further purification. 11-Mercapto-1-undecanol (MUD), 1-undecanethiol, 2-bromo-2-methylpropionyl bromide (BMPB), triethylamine (TEA), 2,20 -bipyridine (Bpy), 4-(N,N-dimethylamino)pyridine (DMAP), N,N,N0 ,N00 ,N00 -pentamethyldiethylenetriamine (PMDETA), and methanol were used as received. The water used in all synthesis and measurements was from a Milli-Q system. Characterization. Water contact angle measurement was carried out on an OCA20 contact angle system (Dataphysics GmbH, Germany) at room temperature. Static contact angle was measured by the sessile drop method as follows. First, a water drop (∼5 μL) was lowered onto the sensor surface from a needle tip. Then, the images of the droplet were recorded. Contact angles were determined from these images with the calculation software from Dataphysics. Advancing angles were measured by adding 5 μL water to the stationary droplet. Receding angles were then measured by removing 5 μL of water from the droplet. All results were an average of at least five measurements. FT-IR/ATR measurement was carried out on a FT-IR instrument (Varian 3100, USA) equipped with an ATR cell (KRS-5 crystal, 45°). Sixty-four scans were taken for each spectrum at a resolution of 4 cm-1. (29) Yang, Q.; Xu, Z. K.; Dai, Z. W.; Wang, J. L.; Ulbricht, M. Chem. Mater. 2005, 17, 3050–3058.
DOI: 10.1021/la903895q
5747
Article
Yang et al.
The thickness of the polymer films grafted on the SPR sensors was determined by ellipsometry. The measurements were carried out on a variable-angle spectroscopic ellipsometer (MM-16 system, HORIBA Jobin-Yvon, France) at an incident angle of 70° in the wavelength range of 500-1000 nm. The refractive index of the dried PGMA films at all wavelengths was assumed to be 1.5, and for data analysis, a one-layer modified Cauchy model was used to calculate the thickness. All measurements were conducted in dry air at room temperature. For each sample, measurements were made on at least three different surface locations, and average values are reported.
Preparation of ATRP Initiator-Functionalized SPR Sensors. For grafting glycopolymer on the gold sensor by ATRP, a two-step process was conducted to prepare an ATRP initiator functionalized surface. The gold sensors were cleaned by immersing in a solution of potassium dichromate in concentrated sulfuric acid (70 g/L) for 5 min. Then, the sensors were thoroughly rinsed with water and dried under nitrogen flow. 1 mM MUD was dissolved in ethanol, and sensors were immersed in 10 mL of that MUD solution and incubated for 24 h at room temperature to form a SAM. The sensors were then washed with ethanol and water and dried in a stream of nitrogen. For variation of the initiator density on the surface, mixed thiol solutions containing MUD and 1-undecanethiol in a predetermined ratio were used. Immobilization of the ATRP initiator onto the gold sensors was achieved by the reaction between BMPB and hydroxyl groups on the SAM layer. The sensors with the SAM layer on their surface were put in 10 mL of freshly dried acetonitrile containing DMAP (5 mM) and TEA (10 mM), and then 100 μL of BMPB was added. After reaction for 2 h at room temperature, sensors were taken out and washed with acetonitrile and ethanol and then dried in a stream of nitrogen.
Figure 1. FT/IR-ATR spectra of unmodified and poly(GAMA) grafted (100% density, 4 h ATRP) gold surface.
Surface-Initiated ATRP of GAMA on SPR Sensors. Initiator immobilized sensors were put in a Schlenck flask, and the flask was sealed. Then, the flask was evacuated and backfilled with argon three times. GAMA (3.68 g, 12 mmol) and 37.5 mg of Bpy (0.24 mmol) were dissolved in 30 mL of 1:1 (v/v) methanol/ water mixture solution and purged with nitrogen for 30 min. Then, 17.2 mg of copper(I) bromide (0.12 mmol) was added to the solution under drastic stirring and argon stream. Subsequently, the monomer solution was injected into the flask (6 mL for each sample), and the reaction mixture was incubated at room temperature for a predetermined time. After the ATRP reaction, a quenching solution was used to stop the polymerization. The sensors were quickly removed from the Schlenck flask and immersed in 50 mL 1:1 (v/v) methanol/water solution of 250 mg of copper(II) bromide and 625 μL of PMDETA. A water-methanol-ethanol washing sequence was then applied to clean the sensors. SPR Measurement. The SPR biosensor system, IBIS I with a 670 nm wavelength laser, obtained from Xantec Bioanalytics GmbH, was used.30 A phosphate buffered saline solution (PBS, 10 mM, pH = 7.4) was used to prepare protein solutions of BSA, fibrinogen, and lysozyme with a concentration of 1 g/L. Typical SPR measurements were performed at 25 °C involving the following steps. After the injection of the buffer solution onto the sensor surface and the stabilization of the baseline, the protein solution was injected quickly, and the sensor resonance angle was followed for 300 s. Then, the protein solution was withdrawn and buffer solution was injected quickly, and the resonance was followed until a stable value had been reached again. Data were evaluated in terms of the response, that is, the change in the resonance angle (throughout given in 0.001°) due to protein adsorption, and they were based on an analysis of at least two independently prepared sensors, including the results (30) Lazos, D.; Franzka, S.; Ulbricht, M. Langmuir 2005, 21, 8774–8784.
5748 DOI: 10.1021/la903895q
Figure 2. Contact angles (static and dynamic) of mixed SAM layers with different MUD content. of two or three measurements on different locations of one sensor.
Results and Discussion Characterization and Surface Properties. Figure 1 shows FT-IR/ATR spectra of the unmodified and the poly(GAMA) grafted sensors. The modified sensor shows several new peaks compared to bare gold or the SAM on gold (spectra for the bare gold and the SAM-coated gold sensor in ATR mode were identical). The peak at 1710 cm-1 is the absorption attributed to the carbonyl groups in the ester bond. Absorptions at 1650 and 1540 cm-1 belong to the amide I and amide II, respectively. Additional broad absorption bands between 3000 and 3600 cm-1 due to N-H and OH stretching vibration can also be seen. The FT-IR spectra indicate clearly the diagnostic absorptions of the monomer GAMA and suggest the grafting of poly(GAMA). As can be seen from Figure 2, the contact angle measurements show a decrease in the values with increasing MUD content in the SAM. This decrease is expected because 1-undecanethiol with a hydrophobic methyl end group is replaced by MUD with a hydrophilic hydroxyl end group. The advancing and the receding contact angle measurements also yield the hysteresis as a measure of the heterogeneity of the SAM surface. For SAM with 0% or 100% MUD contents, the contact angle hysteresis is very small and indicates that almost homogeneous hydrophobic or hydrophilic surfaces were obtained. With the addition of 1-undecanethiol to the thiol mixture, a chemically heterogeneous SAM was achieved, and an obvious increase of hysteresis can be observed from the larger differences between advancing and Langmuir 2010, 26(8), 5746–5752
Yang et al.
Article
Figure 4. Ellipsometric thickness of the poly(GAMA) layers grown from 100% MUD SAMs. Figure 3. Contact angles of a series of mixed SAMs before and after immobilization of ATRP initiator and after ATRP grafting of glycopolymer for 1 and 4 h.
receding contact angles. It has been well studied and generally accepted that alkanethiols, despite their different end groups, produce well-mixed SAMs.31-34 Particularly for those mixed thiols with similar chain length, phase segregation does not occur. Hence, a SAM with randomly mixed hydroxyl and methyl end groups was successfully prepared on the gold sensor surface, and the density of hydroxyl groups can be controlled by the fraction of 1-undecanethiol in the thiol mixture. The water contact angle of the sensor surface after each functionalization step was also measured, and the results are shown in Figure 3. Compared to the SAM covered sensor, contact angle was higher after the ATRP initiator immobilization step. This can be ascribed to the introduction of the initiator which exposes a relatively hydrophobic end group on the SAM surface. On the other hand, the difference between contact angles before and after initiator immobilization increases with the increase of MUD content. The difference between the values is negligible for the lowest MUD content because there was already a hydrophobic background by the large 1-undecanethiol fraction on the surface. This hinders the observation of a contact angle change upon introduction of the ATRP initiator. The following polymerization by ATRP leads to a decrease in contact angle for all MUD densities. After 1 min of ATRP from the 100% MUD sample the contact angle is relatively high (57°) because polymer chains are not able to cover the surface completely. After 60 min of polymerization the contact angle became considerably lower, and from 60 to 240 min there was again a decrease, indicating further growth of the grafted chains and completion of coverage of the sensor surface. It should be noted that all contact angle measurements had been done with dried samples. Figure 4 shows the results of ellipsometry on polyGAMA grafted sensors obtained on SAMs with 100% MUD, also measured for dry samples. After 60 min ATRP, about 30 nm glycopolymer layer thickness had been obtained on the sensor surface. (31) (a) Li, L.; Chen, S.; Jiang, S. Langmuir 2003, 19, 666–671. (b) Li, L.; Chen, S.; Jiang, S. Langmuir 2003, 19, 3266–3271. (32) Shevade, A. V.; Zhou, J.; Zin, M. T.; Jiang, S. Langmuir 2001, 17, 7566– 7572. (33) Dabirian, R.; Zdravkova, A. N.; Liljeroth, P.; van Walree, C. A.; Jenneskens, L. W. Langmuir 2005, 21, 10497–10503. (34) Shon, Y. S.; Lee, S.; Perry, S. C.; Lee, T. R. J. Am. Chem. Soc. 2000, 122, 1278–1281.
Langmuir 2010, 26(8), 5746–5752
Figure 5. Adsorption of BSA (1 g/L in 10 mM PBS buffer, pH = 7.4) onto poly(GAMA) layer with different chain densities measured by SPR.
With the further increase of ATRP time, polymer layer thickness increased continuously. However, the growth rate of layer thickness slowed down after 120 min. This may be ascribed to some loss of living chain ends which may have been caused by burying of propagating chains within the polymer brushes. Zhang et al.21 investigated surface-initiated ATRP of a methacrylate with a large hydrophilic side group, poly(ethylene glycol) methacrylate (PEGMA, n ∼ 6), and they obtained 35 nm grafted layer in 3 h. Apparently, the growth rate in our work for a hydrophilic monomer with similar size was significantly higher which may be due to the different reaction conditions. For the further investigations we assume, in agreement with previous studies of other groups,35,36 that the stepwise reduced (35) Jones, D. M.; Brown, A. A.; Huck, W. T. S. Langmuir 2002, 18, 1265–1269. (36) Nagase, K.; Kobayashi, J.; Kikuchi, A.; Akiyama, Y.; Kanazawa, H.; Okano, T. Langmuir 2008, 24, 511–517.
DOI: 10.1021/la903895q
5749
Article
Yang et al.
Table 1. SPR Responses and Relative BSA Adsorption on the Poly(GAMA) Grafted Sensors with Different Chain Densities and Layer Thicknesses SPR response (0.001°)
bare gold 20% MUD ATRP 4 h 50% MUD ATRP 1 h 50% MUD ATRP 4 h 80% MUD ATRP 1 h 80% MUD ATRP 4 h 100% MUD ATRP 4 h
plateau value after BSA injection
plateau value after buffer injection
total BSA adsorption with respect to bare gold before washing (%)
weakly adsorbed BSA washed off with buffer (%)
irreversibly adsorbed BSA with respect to bare gold (%)
444.8 ( 1.4 265.9 ( 7.2 158.5 ( 4.8 133.3 ( 0.3 40.2 ( 0.3 14.5 ( 0.3 0.6 ( 0.3
342.7 ( 2.0 154.3 ( 1.5 137.8 ( 1.6 91.3 ( 1.1 16.5 ( 1.6 1.4 ( 0.2 0.1 ( 0.07
100 59.8 35.6 30.0 9.0 3.3 0.13
23.0 42.0 13.1 31.5 58.0 90.3 83.3
100 45.0 40.2 26.6 4.8 0.4 0.03
MUD density along with the high reactivity of the acyl bromide with the surface hydroxyl groups will lead to a stepwise reduced grafting density. Absolute values cannot be given because it is known that usage of surface-immobilized ATRP initiator is typically not complete,35,37 and direct analysis of the grafted polymer on the planar substrates is impossible due to too small amounts. However, the systematic increase of contact angle for grafted layers obtained after the same ATRP time from SAMs with decreasing MUD content (cf. Figure 3) is taken as evidence that at the same grafted chain length the grafting density is indeed reduced. Adsorption of BSA onto Glycopolymer Grafted SPR Sensors. BSA was used as a model protein to evaluate the antinonspecific adsorption property of the grafted poly(GAMA). A protocol already used in previous work was applied,30 and exemplaric raw SPR data are shown in Figure 5. It can be seen that BSA showed a very fast adsorption to the bare gold surface after the protein solution was added to the sensor. After washing with buffer, about 23% of the adsorbed BSA was removed from the sensor surface. Data for different layers with varied grafted chain densities on the sensor surface are shown in Table 1. Relative adsorbed BSA amounts are also expressed relative to the amount of BSA adsorbed on the bare sensor surface and in terms of the percentage of weakly bound BSA which was washed off from the surface simply with the buffer solution, We may assume that the weak adsorption of proteins to a surface only occurs on the top of the glycopolymer layer and the weakly bound protein rapidly dissociates from the surface when the solution of protein is replaced with buffer.24 On gold, BSA can adsorb both covalently (via thiol groups) and noncovalently (hydrophobic and van der Waals interactions). With the grafted glycopolymer layer on the surface, the adsorption of BSA was obviously suppressed, especially in the high chain density cases. On the 20% MUD/initiator coverage sample, about 60% of the total adsorbed protein and 45% of the irreversibly adsorbed amount can be detected compared to the amounts on bare gold. For 80% and 100% coverage samples after 4 h ATRP, only very little protein was adsorbed at all (3.3 and 0.13% for total adsorption and 0.4 and 0.03% for irreversible adsorption, all relative to bare gold; cf. Table 1). These results are surprisingly low compared to the most wellknown anti-nonspecific adsorption properties of grafted poly(ethylene glycol) (PEG) or oligo(ethylene glycol) (OEG). The ability of PEG layers to resist protein adsorption had been related to its highly hydrated nature, the formation of a steric barrier by the PEG chains, and the high mobility of the PEG chains in water.38-40 (37) Edmondson, S.; Osborne, V. L.; Huck, W. T. S. Chem. Soc. Rev. 2004, 33, 14–22. (38) Unsworth, L. D.; Sheardown, H.; Brash, J. L. Langmuir 2005, 21, 1036– 1041. (39) Li, L. Y.; Chen, S. F.; Zheng, J.; Ratner, B. D.; Jiang, S. Y. J. Phys. Chem. B 2005, 109, 2934–2941. (40) McGurk, S. L.; Green, R. J.; Sanders, G. H. W.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B.; Williams, P. M. Langmuir 1999, 15, 5136–5140.
5750 DOI: 10.1021/la903895q
Whitesides and co-workers19,24 attribute the protein resistance property of surfaces to high hydrophilicity and the presence of hydrogen-bond acceptors, but the absence of hydrogen-bond donors and a neutral electrical charge, all preconditions for kosmotropic properties. However, there are also other studies showing that highly protein-resistant surfaces can be achieved by surface layers displaying a large amount of hydrogen-bond donors, such as hydroxyl groups and carboxylic acid groups.25,26,41 Mrksich and co-workers25 compared the adsorption of several different proteins on the mannitol-terminated SAM. This work suggested that the SAM prevented the adsorption of several proteins and that this anti-nonspecific adsorption property was indistinguishable from that of a SAM displaying tri(ethylene glycol) groups. In our case, the abundance of hydroxyl groups is much higher than in the above-mentioned saccharide endcapped SAMs. Moreover, the glycopolymer chains are more mobile in water and are much better hydrated to induce a steric repulsion, and both effects can contribute to the protein resistance property. Therefore, this result indicates that the grafted synthetic glycopolymer is more promising for mimicking the sugar-rich hydrated environment of the glycocalyx layer on cell membranes than SAMs with end-grafted sugar groups. An increase of chain length leads to a reduction of protein adsorption, and nevertheless, this effect is not as significant as the result caused by increasing chain density. However, what can also be seen when comparing the values of weakly adsorbed BSA washed off with buffer from 50% MUD content and from 80% MUD content samples is an increase of the relative amount of weakly adsorbed protein with increasing chain length (18% difference between the percentage values for 1 and 4 h ATRP at 50% MUD content and 32% difference for the same ATRP times at 80% MUD content) because the grafted layer becomes more dense and shielding of the underlying SAM more efficient. However, when comparing all values for the weakly adsorbed protein, no clear trend can be deduced. A cyclic BSA adsorption experiment was carried out on a sensor sample at medium grafting density, and the results are shown in Figure 6. The BSA adsorption reached the maximal value quickly, and a part of this adsorbed amount could be washed off by buffer. When the same BSA solution was injected again, the same amount of BSA as previously desorbed by the washing buffer readsorbed to the grafted layer. Moreover, it was found that this reversible fraction of adsorbed amount kept almost identical in all three cycles, which indicates that no further irreversible adsorption of BSA occurred after the first cycle. Therefore, we can also conclude that this surface grafted poly(GAMA) layer possesses excellent stability with respect to accumulation of nonspecifically bound protein during repeated exposure to protein solutions. (41) Ajikumar, P. K.; Ng, J. K.; Tang, Y. C.; Lee, J. Y.; Stephanopoulos, G.; Too, H. P. Langmuir 2007, 23, 5670–5677.
Langmuir 2010, 26(8), 5746–5752
Yang et al.
Article
Figure 6. Cyclic BSA adsorption/desorption experiment on 50% MUD sample followed by SPR. Figure 8. Adsorption of fibrinogen (1 g/L in 10 mM PBS buffer, pH = 7.4) onto poly(GAMA) layer with different chain densities measured by SPR. Table 2. Molecular Weight, pI Value, and Shape of the Proteins Used in This Work31 protein
MW (kDa)
lysozyme BSA fibrinogen
14.7 66.4 340.0
pI
shape
10.5 ellipsoid: 5 3 3 nm3 4.7 ellipsoid: 4 4 11.5 nm3 5.5 rabdoid: diameter 6 nm; length 45 nm
Figure 7. Adsorption of lysozyme (1 g/L in 10 mM PBS buffer, pH = 7.4) onto poly(GAMA) layer with different chain densities measured by SPR.
Adsorption of Lysozyme and Fibrinogen onto Glycopolymer Grafted SPR Sensors. To evaluate the effect of protein size on adsorption, lysozyme and fibrinogen were also used for adsorption experiments (Figures 7 and 8). As showed in Table 2, lysozyme is a very small protein with an almost spherical shape and about half of the size of BSA. Fibrinogen is a cylindrical protein and much larger than BSA and lysozyme. Fibrinogen caused a large change in SPR response when it was allowed to contact bare gold. According to the literature,42 a change of SPR peak angle of 0.1° corresponds to about 0.1 μg/cm2 adsorbed protein on the surface. Using this approximation, a value of 0.6 μg/cm2 was obtained which is almost 2.5 times the value for (42) Stenberg, E.; Persson, B.; Roos, H.; Urbaniczky, C. J. Colloid Interface Sci. 1991, 143, 513–526.
Langmuir 2010, 26(8), 5746–5752
Figure 9. Schematic illustration of the adsorption behavior of proteins with different size to the glycopolymer grafted surface with different chain density.
lysozyme and 0.26 μg/cm2 more than that of BSA adsorbed on the same surface. When the cylindrical fibrinogen adsorbs on materials surface, it can bind in two conformations: vertical and lyingdown. In these two conformations, the amounts of adsorbed fibrinogen monolayer could be 1.57 and 0.21 μg/cm2, respectively.43 The about 0.6 μg/cm2 fibrinogen adsorbed on bare gold in this study indicates that fibrinogen adsorbed randomly on the bare gold, in both vertical and lying-down conformations. Compared to BSA and lysozyme, fibrinogen exhibits much higher SPR (43) Hu, J.; Yang, D.; Kang, Q.; Shen, D. Sens. Actuators, B 2003, 96, 390–398.
DOI: 10.1021/la903895q
5751
Article
Yang et al.
Table 3. SPR Responses and Relative Protein Amounts on the Poly(GAMA) Grafted Sensors with 80% MUD Density Prepared with 1 h ATRP from Adsorption of BSA, Lysozyme, and Fibrinogen SPR response (0.001°) plateau value after plateau value after total protein adsorption with protein injection buffer injection respect to bare gold (%) BSA lysozyme fibrinogen
40.2 ( 0.3 20.2 ( 0.4 3.1 ( 0.6
16.5 ( 1.6 1.2 ( 0.5 1.35 ( 0.2
9.0 5.8 0.4
angle change on bare gold and consequently more mass of protein adsorbed on the surface which could be ascribed to the larger molecular weight and size of this protein. However, after grafting the sensor surface with glycopolymer layers even in low density (ATRP on SAM with 20% MUD), the adsorption of fibrinogen was obviously suppressed: Only 20% of the total protein adsorbed on bare gold was bound to this sensor surface, whereas for lysozyme the value was almost 50% and for BSA it was about 60%. Moreover, a different adsorption behavior can be seen when the shapes of the SPR curves of BSA, lysozyme, and fibrinogen are compared. For lysozyme and BSA the SPR angles changed sharply after adding the protein solutions, and the curves quickly reached a plateau value. However, for fibrinogen a relatively slow increase of SPR angle can be observed (cf. Figure 8); i.e., it needs more time to achieve the maximum of adsorbed protein amount. This difference can again be ascribed to the larger size of fibrinogen; compared to the smaller proteins like BSA and, especially, lysozyme, it takes more time to diffuse into the grafted layer and to fit into the adsorption sites. This can also be proven by the jump of the SPR angles observed for lysozyme adsorption on high grafting density sensor samples (cf. Figure 7). The small protein can penetrate into the polymer layer easily and almost immediately upon contact of the solution with the sensor. However, after equilibration of the sensor layer with the protein solution a slightly lower value is reached. Moreover, we explain the relatively lower protein uptake values in the grafted layers with low to moderate grafting density for lysozyme compared to the larger BSA with its much higher net charge than for BSA under the experimental conditions; i.e., the electrostatic repulsion of protein from the layer by already bound protein is stronger than its weak binding in the glycopolymer layer. The schematic illustration in Figure 9 may explain the different adsorption behavior of the three proteins on surfaces with different grafting density of glycopolymer. At a surface with a MUD content of 20%, the largest amount of places is available for the proteins to adsorb on the hydrophobic end groups of 1-undecanethiol. Nevertheless, fibrinogen adsorbs less than the other two proteins both reversibly and irreversibly because of its larger dimensions. In other words, this relatively low surface coverage is already enough to effectively resist the adsorption of fibrinogen. When increasing initiator density to 50 and 80%, the space between glycopolymer chains becomes more narrow and the protein more hindered to reach hydrophobic sites on the SAM so that there is a decrease in adsorption for both BSA and lysozyme. There is still protein adsorption within the glycopolymer layer which causes the SPR signal to increase right after injection of the protein solution. However, this fraction of adsorbed protein is only weakly bound and can be washed out by the subsequent rising with the binding buffer. For 80% initiator density the chains are so densely packed that even lysozyme is
5752 DOI: 10.1021/la903895q
weakly adsorbed protein irreversibly adsorbed protein with washed off with buffer (%) respect to bare gold (%) 58.0 94.1 56.5
4.8 0.5 0.2
sufficiently prevented from entering the layer. This chain density has the biggest influence on fibrinogen, and only very little protein adsorbs. An overview of data for the different proteins on 80% MUD density grafted sensors is presented in Table 3. Fibrinogen has the lowest values for both total and irreversibly bound protein with respect to bare gold of the three proteins as well as a similar percentage value for weakly adsorbed protein as BSA. For 100% MUD content, adsorption of all these proteins is suppressed. The grafted glycopolymer formed a dense water-swollen layer on the surface and showed minimal binding and maximal steric repulsion to the proteins.
Conclusions SPR sensors were modified with mixed SAMs of different composition, and after initiator immobilization poly(GAMA) was grafted via surface-initiated ATRP. Contact angle measurements showed the successful synthesis of glycopolymer layers with different grafting densities. Ellipsometry measurement demonstrated that the thickness of the poly(GAMA) brushes, grown from SAMs at highest initiator density, was a linear function of polymerization time up to 120 min, and thereafter the growth rate slowed down. The grafted glycocalyx-like poly(GAMA) layer exhibits very promising protein resistance to all the proteins (BSA, lysozyme, and fibrinogen) used in this work, and the amount of adsorbed protein was dependent on the length and density of the glycopolymer chains and also on the size of the proteins. With the increase of chain length, a relatively small protein like BSA showed a decrease in the adsorbed amounts and an increase in the percentage of weakly bound protein. Increasing grafting density caused lower adsorbed amounts for all three proteins. However, the reduction was clearly most pronounced for the largest protein fibrinogen. For the highest grafting density, the amounts of irreversibly bound protein were lower than 0.1% of the monolayer adsorption capacity of a planar surface. Furthermore, the repeated adsorption and desorption experiments with grafted layers of medium density showed that the protein binding can be definitely divided into reversible and irreversible parts and the reversibly bound fraction can be quantitatively desorbed and readsorbed. These properties make the grafted synthetic glycopolymer layers a very promising platform for specific binding of proteins or other bioparticles. Ongoing work using the functionalization schemes established in the present study is now focused onto grafted layers from other glycomonomers with specific binding properties for lectins, other proteins, or cells. Acknowledgment. The Alexander von Humboldt Foundation, Bonn, Germany, is thanked for the postdoctoral fellowship of Dr. Qian Yang.
Langmuir 2010, 26(8), 5746–5752
