QCM-D Analysis of the Performance of Blocking ... - ACS Publications

Jul 22, 2008 - Kristina Reimhult, Karolina Petersson, and Anatol Krozer*. The IMEGO Institute, ArVid HedValls Backe 4, SE-411 33 Gothenburg, Sweden...
0 downloads 0 Views 106KB Size
Langmuir 2008, 24, 8695-8700

8695

QCM-D Analysis of the Performance of Blocking Agents on Gold and Polystyrene Surfaces Kristina Reimhult, Karolina Petersson, and Anatol Krozer* The IMEGO Institute, ArVid HedValls Backe 4, SE-411 33 Gothenburg, Sweden ReceiVed January 23, 2008. ReVised Manuscript ReceiVed May 20, 2008 With today’s developments of biosensors and medical implants comes the need for efficient reduction of nonspecific binding. We report on a comparison of the ability of traditionally used blocking agents and poly(ethylene glycol) (PEG) derivatives to prevent protein adsorption on both gold and polystyrene surfaces. The adsorption kinetics of blocking molecules and proteins was monitored gravimetrically using quartz crystal microbalance with dissipation (QCM-D). The resistance to nonspecific adsorption was evaluated on gold and polystyrene surfaces coated with bovine serum albumin (BSA) or casein, gold coated with three different 6-11 ethylene glycol (EG) long hydroxyl- or methoxyterminated PEG-thiolates and polystyrene blocked with a PLL-g-PEG or three different 12 EG long benzyl-PEGderivatives. The prevention of protein adsorption on the coated surfaces was evaluated by monitoring the mass uptake at the addition of both pure prostate specific antigen (PSA) and seminal plasma. We demonstrate that on pure gold the PEG-thiols are superior to the other blocking molecules tested, with the end group and length of the PEG-thiols used being of minor importance. On polystyrene surfaces blocking with PLL-g-PEG, BSA and casein gave the best results. These results have an impact on further development of an optimized immunoassay protocol.

Introduction The minimization of nonspecific binding (NSB) is vital not only for applications where detection limit and accuracy is an issue, like for example within clinical diagnostics and the immunoassay based detection of diseases.1 Optimization of NSB is also relevant when biological molecules interact with nonbiological surfaces like e.g. during the inflammatory response to implants and to prevent biofouling and microbial adhesion associated failure of biomedical devices.2 Usually the surfaces are exposed not only to cells, bacteria or proteins but also to smaller molecules like DNA fragments or peptides (and even amino acids). Given this complexity it is therefore not surprising that NSB has been and still is a subject of vigorous research. Well performing blocking agents should of course be inert to the adsorption of debris in a biological fluid (e.g., blood or seminal plasma). They should also bind strongly to the substrate surface to prevent exchange reactions. In addition, when blocking nonspecific adsorption for sensing application the blocking agents should not interfere with further reactions. For example, the size of the blocking agent should be small enough not to hinder the targeted analyte to approach and react with the catching molecule, usually an antigen approaching an antibody. Materials to block found interesting in the literature are for example Au or Si and its compounds, like Si3N4 or SiO2, that are the surfaces-of-choice for biosensor applications and poly(dimethylsiloxane) (PDMS), that is a common material used in construction of microfluidic systems.3 Strong coupling of polymeric blocking agents to metal surfaces is achieved via metal-sulfur (thiolate) or Si-H bonds. Blocking molecules that enable multiple bonds to hydrophobic polymeric substrates are * To whom correspondence should be addressed. E-mail: anatol.krozer@ imego.com. Tel: +46 31 750 18 06. Fax: +46 (0)31 750 18 01. (1) Gobi, K. V.; Iwasaka, H.; Miura, N. Biosens. Bioelectron. 2007, 22, 1382– 1389. (2) Harris, L. G.; Tosatti, S.; Wieland, M.; Textor, M.; Richards, R. G. Biomaterials 2004, 25, 4135–4148. (3) Marie, R.; Beech, J.; Vo¨ro¨s, J.; Tegenfeldt, J.; Ho¨o¨k, F. Langmuir 2006, 22, 10103–10108.

of importance since traditional immunoassays are often performed using more or less hydrophobic polystyrene as the substrate.4 Traditionally used blocking agents include bovine serum albumin (BSA), dextran and casein. The blocking abilities of wells in commercially available polystyrene ELISA-plates by BSA, casein and other natural biological blocking agents have been extensively reported in the literature.5–7 The blocking efficiency was usually inferred indirectly by first blocking the surface and then monitoring subsequent binding (or rather the absence of binding) of labeled proteins, and comparing this signal with the one measured on native polystyrene. The results reported in the literature are somewhat inconclusive. For example, Vogt et al.7 reported that the blocking performance of casein was better than that of BSA. ELISA wells exposed to casein at concentrations higher than 30 µg/mL resulted in 95% blocking efficiency while the highest BSA blocking efficiency was 90%, which was reached only after exposures to BSA concentrations in excess of 800 µg/mL. Sentandreu et al.,5 on the other hand, reported much lower blocking efficiencies for both casein and BSA despite using higher concentrations of each of the molecules. Furthermore, Feiler at al.8 showed that exposing polystyrene coated quartz electrodes (similar to the substrates used in the present work) to BSA at as small concentrations as 50 µg/mL was enough to efficiently block the surface against subsequent adsorption of bovine submaxillary gland mucin (consisting mainly of glycoproteins). Surprisingly, we did not find any investigations of the blocking efficiencies of natural biological agents like BSA or casein on gold, although the adsorption of BSA on Au has been studied.9 (4) Shmanai, V. V.; Nikolayeva, T. A.; Vinokurova, L. G.; Litoshka, A. A. BMC Biotechnol. 2001, 1, 4. (5) Sentandreu, M. A.; Aubry, L.; Toldra´, F.; Ouali, A. Eur. Food Res. Technol. 2007, 224, 623–628. (6) Kenna, J. G.; Major, G. N.; Williams, R. S. J. Immunol. Methods 1985, 85, 409–419. (7) Vogt, R. F.; Phillips, D. L.; Henderson, L. O.; Whitfield, W.; Spierto, F. W. J. Immunol. Methods 1987, 101, 43–50. (8) Feiler, A. A.; Sahlholm, A.; Sandberg, T.; Caldwell, D. K. J. Colloid Interface Sci. 2007, 315, 475–481. (9) Zhang, Y.; Wang, M.; Xie, Q.; Wen, X.; Yao, S. Sensors Actuators B 2005, 105, 454–463.

10.1021/la800224s CCC: $40.75  2008 American Chemical Society Published on Web 07/22/2008

8696 Langmuir, Vol. 24, No. 16, 2008

In recent years the research focus has been on blocking using synthetic polymeric molecules and the surfaces that to date are considered the most resistant to protein adsorption are the ones presenting poly(ethylene glycol) (PEG).10,11 Typically, linear PEG molecules are modified to include functionality for binding to the surface of interest. The research include blocking of Au by using PEG-thiols12 and PEG-functionalized alkanethiols,13–15 blocking of oxidized PDMS and negatively charged metal oxide surfaces such as TiO2, SiO2, Nb2O5 and Ta2O5 using PLL-gPEG,3,10,16–18 blocking via condensation of PEG-silanols with silanols on silicon surfaces19 and blocking of hydrophobic surfaces using PEO/PPO/PEO (Pluronic).20–22 Another polymeric compound, pTHMMAA, which is a nonionic hydrophilic disulfide that performs well when used for blocking in immunoassays on gold, has been reported by Vikholm and co-workers.23–25 We present a comprehensive evaluation of the binding of blocking molecules and subsequent adsorption of proteins on surfaces using quartz crystal microbalance with dissipation (QCM-D). We compare the performance of BSA and casein with the performance of several PEG-derivatives on both gold and polystyrene surfaces. BSA and casein were chosen because of their standard use as blocking molecules in immunochemistry applications, while PEG derivatives are more recently identified blocking molecules showing very good performance.10,11 As a test of the blocking ability we monitor the uptake of proteins from seminal plasma and the uptake of prostate specific antigen (PSA) in pure phosphate buffered saline (PBS). Human seminal plasma contains over 30 proteins, including coagulation proteins (that are known to be sticky) and proteins like prealbumin, albumin, globulins, mucin, nucleoprotein, immunoglobulins and transferrin,26 and can therefore be considered a good reference for nonspecific adsorption. Our focus here is two-fold. First we wanted to find simple and cheap blocking agents suitable to apply in an immunoassay protocol on Au and polystyrene substrates, respectively. In assays with antibodies immobilized directly on surfaces without linking chemistries, small blocking molecules are preferable in order not to introduce steric hindrance for further binding of antigens and detecting antibodies. With this in mind, we have chosen short PEG-molecules for our experiments. Second, we wanted to compare how the PLL-g-PEG copolymer performs on polystyrene as compared to traditional blocking molecules and to short PEG strands. The blocking molecules giving the largest (10) Chapman, R.; Ostuni, E.; Takayama, S.; Holmlin, R.; Yan, L.; Whitesides, G. J. Am. Chem. Soc. 2000, 122, 8303–8304. (11) Lee, S.; Vo¨ro¨s, J. Langmuir 2005, 21, 11957–11962. (12) Unsworth, L. D.; Sheardown, H.; Brash, J. L. Biomaterials 2005, 26, 5927–5933. (13) Prime, K. L.; Whitesides, G. M. Science 1991, 252, 1164–1167. (14) Harder, P.; Grunze, M.; Dahint, R.; Whitesides, G. M.; Laibinis, P. E. J. Phys. Chem. B 1998, 102, 426–436. (15) Menz, B.; Knerr, R.; Go¨pferich, A.; Steinem, C. Biomaterials 2005, 26, 4237–4243. (16) Huang, N.; Michel, R.; Vo¨ro¨s, J.; Textor, M.; Hofer, R.; Rossi, A.; Elbert, D.; Hubbell, J.; Spencer, N. Langmuir 2001, 17, 489–498. (17) Kenausis, G. L.; Vo¨ro¨s, J.; Elbert, D. L.; Huang, N.; Hofer, R.; RuizTaylor, L.; Textor, M.; Hubbell, J. A.; Spencer, N. D. J. Phys. Chem. B 2000, 104, 3298–3309. (18) Pasche, S.; Vo¨ro¨s, J.; Griesser, H. J.; Spencer, N. D.; Textor, M. J. Phys. Chem. B 2005, 109, 17545–17552. (19) Zhang, M.; Desai, T.; Ferrari, M. Biomaterials 1998, 19, 953–960. (20) Razatos, A.; Ong, Y.; Boulay, F.; Elbert, D.; Hubbell, J.; Sharma, M. L.; Georgiou, G. Langmuir 2000, 16, 9155–9158. (21) Amiji, M.; Park, K. Biomaterials 1992, 13, 682–692. (22) Liu, V. A.; Jastromb, W. E.; Bhatia, S. N. J. Biomed. Mat. Res. 2002, 60, 126–134. (23) Vikholm, I. Sens. Act. B 2005, 106, 311–316. (24) Vikholm-Lundin, I. Langmuir 2005, 21, 6473–6477. (25) Vikholm, I.; Albers, W. M. Biosens. Bioelectron. 2006, 21, 1141–1148. (26) Lay, M. F.; Richardson, M. E.; Boone, W. R.; Bodine, A. B.; Thurston, R. J. J. Assisted Repr. Gen. 2001, 18, 144–150.

Reimhult et al.

reduction of nonspecific adsorption were then used in PSA sandwich immunoassay experiments on gold and polystyrene, respectively, that will be the subject of a forthcoming publication.

Experimental Section Reagents. The ability to block adsorption on pure gold was evaluated using bovine serum albumin (BSA) (Fluka), casein lysate (Sigma) and three different PEG-thiols with hydroxyl- or methoxylhead groups. The PEG-thiols: 99% pure (HO(CH2CH2O)7CH2CH2SH) (referred to as hPEG7), 98% (HO(CH2CH2O)11CH2CH2SH) (hPEG11) and 99% H3CO(CH2CH2O)6CH2CH2S2CH2CH2(CH2CH2O)6OCH3 (mPEG6) were purchased from Polypure, Norway. The size of BSA and fragmented casein was estimated using photon correlation spectroscopy (PCS) (Zetasizer Nano ZS, Malvern Instruments). The ability to block adsorption on polystyrene was evaluated using BSA, casein lysate, PLL(20)-g[3.5]-PEG(2)/PEG(3.4)biotin18% (poly(L-lysine)-graft-PEG copolymer, Surface Solutions, Schweiz, referred to as PLL-g-PEG) and three different PEG derivates obtained from Polypure. The PLL-g-PEG is composed of a poly(L-Lysine) backbone of 20 kDa to which 2 kDa poly ethylene glycol side chains and 3.4 kDa biotinylated PEG chains are grafted. The three short PEG derivatives were (HO(CH2CH2O)12CH2CH2OH) (referred to as PEG12), (C6H5CH2O(CH2CH2O)12CH2CH2OH) (monobenzyl ether PEG12, mbPEG12) and (C6H5CH2O(CH2CH2O)12CH2CH2OCH2C6H5) (dibenzyl ether PEG12, dbPEG12). The PEG molecules had either none, one or two hydrophobic benzyl ether end groups and correspondingly two, one or no hydroxyl head groups. PSA in seminal plasma was obtained from CanAg Diagnostics (now Fujirebio), Sweden, and a 95% PSA Tris-solution was obtained from Scripps laboratories, USA. Phosphate buffered saline (PBS, pH 7.4) (Medicago) was used as the running buffer in all the experiments. All solutions were prepared using ultrapure water (Milli-Q System, Millipore, Sweden). 30% hydrogen peroxide was obtained from Scharlau and 30% ammonium hydroxide was obtained from Merck. Quartz Crystal Microbalance with Dissipation monitoring. The experiments were performed using a Q-Sense D300 system (Q-Sense, Sweden). The equipment is based on Quartz Crystal Microbalance with Dissipation monitoring (QCM-D). The sensor chips are composed of disk-shaped, highly polished, AT-cut piezoelectric quartz crystals (fundamental resonance frequency of 5 MHz) with thin gold film electrodes deposited on its two faces, one of which has been grounded (Q-Sense, Sweden). For adsorption reactions onto gold (Au) we used the grounded electrodes as substrates. The results for polystyrene (PS) were obtained on similar quartz wafer covered by a thin PS film spin coated by Q-Sense onto the grounded electrode of the sensor chip. All experiments were performed at 23 °C ((0.05 °C), with the temperature being controlled by the instrument. With the QCM-D technique one can follow every addition of chemicals and measure the mass and viscoelasticity of the layer associated with molecules adsorbed on the sensor surface. The received signal upon adsorption consists of a shift in frequency (F) that is proportional to the negative of the adsorbed mass and a shift in dissipation (D) that is proportional to the viscoelasticity of the adsorbed layer. The relation between the shift in frequency (∆F) and the mass of the adsorbed layer (∆m) is for thin adsorbed layers well approximated by the Sauerbrey relation27

∆m ) -C∆Fn ⁄ n

(1)

where C is the mass sensitivity constant (C ) 17.7 ng · cm-2 · Hz-1 for 5 MHz crystal) and n is the overtone number (n ) 1, 3,...). Typically the ∆F and ∆D of the fundamental frequency and of the three lowest overtones are all simultaneously monitored. All data presented are collected from the third overtone with values for the frequency divided by three (∆F3/3). (27) Sauerbrey, G. Z. Phys. 1959, 155, 206–222.

Blocking Agents on Gold and Polystyrene Surfaces

Langmuir, Vol. 24, No. 16, 2008 8697

Also water molecules coupled to and trapped by the adsorbed molecules contribute to the sensor signal, which means that one actually measure an effective mass load, and not a real mass of the adsorbed molecules alone. The water contribution to the total measured mass increases with increased molecular weight of the protein.9,28,29 For BSA the water bound to an adsorbed protein layer produces an additional mass increase of 15%.9 It is instructive to use the concept of the surface area per adsorbed molecule, Aad, in the forthcoming discussion of the quality of blocking given by

Aad ) Mw ⁄ (∆mNA)

(2)

where ∆m is the mass of the adsorbed layer calculated using eq 1, NA is the Avogadro constant and Mw is the molecular weight of the added molecule. Note that when using eq 1 one underestimates Aad since the water bound within the adlayer is included into the mass change. Using the results mentioned above, for BSA this leads to an underestimation of the Aad by at least 15%. Cleaning of the Sensor Surfaces and the Cell. A clean gold surface gets covered with contaminants from the air within minutes in a laboratory atmosphere, rendering the hydrophilic surface hydrophobic.30–33 This affects the reproducibility of the results and complicates for example binding of PEG-thiols to gold surfaces. New gold covered or polystyrene on gold covered quartz crystal microbalance (QCM) crystals were used for each experiment. The gold surfaces were thoroughly cleaned immediately before use. The cleaning procedure started with 5 min in a hot (about 80 °C) 5:1:1 mixture of ultrapure water, hydrogen peroxide (30%) and ammonium hydroxide (30%). After rinsing the gold surfaces with ultrapure water and drying them in a flow of 5N5 pure nitrogen gas, they were subjected to UV-ozone treatment for at least 20 min and thereafter immediately mounted in the QCM cell. The polystyrene surfaces were only rinsed with ultrapure water and dried in a flow of nitrogen gas immediately before use. The QCM cell was thoroughly rinsed with PBS buffer between each measurement. Experimental Procedure. The experiments were performed by sudden exchange of the solutions in the experimental chamber. The uptake kinetics was, most probably, mass transport limited and did not describe the true reaction kinetics. However the differences in time evolution of the sensor response to adsorption of similar molecules in solution were obviously due to different reaction kinetics. After stabilization of the QCM-D baseline in phosphate buffered saline (PBS), blocking molecules in PBS were added at the same mass concentration, 100 µg/mL, to a clean sensor surface (see Figure 1). For BSA uptake on polystyrene, also mass concentrations of 1 and 5 mg/mL were used (see Figure 3). After the uptake of each of the blocking molecules had reached steady state for the first addition, extra additions of the blocking molecules solutions were performed. Loosely bound molecules were then rinsed off the surface by several washing steps using the PBS solution. Thereafter the level of nonspecific binding was tested by first adding 10 µg/mL of 95% pure prostate specific antigen (PSA) and then, after rinsing with buffer, 1:70 dilutions of PSA-containing seminal plasma in PBS. After several additions of each of the protein solutions the samples were rinsed by adding the buffer solution. The total exposure time of each of the protein solutions varied from 30 min to 1 h depending on the kinetics.

Results and Discussion As mentioned in the Introduction we have chosen relatively short PEG-molecules for our blocking experiments in order for (28) Ho¨o¨k, F.; Rodahl, M.; Brzezinski, P.; Kasemo, B. Langmuir 1998, 14, 729–734. (29) Caruso, F.; Furlong, D. N.; Kingshott, P. J. Colloid Interface Sci. 1997, 186, 129–140. (30) Bain, C. D.; Troughton, E. B.; Tao, Y.-T.; Evall, J.; Whitesides, G. M.; Nuzzo, R. G. J. Am. Chem. Soc. 1989, 111, 321–335. (31) Gaines, G. L., Jr J. Colloid Interface Sci. 1981, 79, 295. (32) Smith, T. J. Colloid Interface Sci. 1980, 75, 51–55. (33) Schneegans, M.; Menzel, E. J. Colloid Interface Sci. 1982, 88, 97–99.

Figure 1. Plot of the frequency shift as a function of time when adding 100 µg/mL of the blocking molecules mPEG6, hPEG7, hPEG11, BSA and casein to gold (Au) and PEG12, mbPEG12, dbPEG12, PLL-g-PEG, BSA and casein to polystyrene (PS), respectively. The V arrows indicate the second (and third) addition of blocking molecules while the v arrows mark the first rinse with buffer. Arrow markings for the PEG curves are omitted due to lack of space.

them to be suitable in an immunoassay protocol (PLL-g-PEG on polystyrene being an exception). With each EG unit being about 3.6 Å,34 the short PEG molecules of 6-12 EG moieties are about 2-4 nm long, which is to compare with the dimensions of IgG antibodies that are determined by X-ray diffraction to be about 13.1 nm × 10.5 nm × 7.3 nm.35 Blocking free space between antibodies attached on a surface using the short PEG molecules should therefore leave the antibodies free for further binding to antigens without steric hindrance. Binding of PEGs to gold was achieved using PEG derivatives containing thiol or disulfide-groups that form thiolate bonds with the surface. It is well-known that long-chain alkanethiols form ordered and oriented monolayers onto noble metals where the molecules adsorb with their head groups pointing away from the surface.30,36 PEG thiols adsorbed on Au most probably form similar oriented layers, even though not as well ordered. Binding of short PEG molecules to the hydrophobic polystyrene was achieved by using PEG derivatives containing hydrophobic benzyl ether end groups. The intention was that the hydrophobic interaction between the benzyl ether end group and the polystyrene would be strong enough to reduce possible exchange reactions with biomolecules from seminal plasma and would promote oriented adsorption of the PEG derivatives on the surfaces, enabling formation of layers that would expose the glycol groups away from the surface. To study the effect of PEG chain length on the blocking ability, but still use relatively short PEG chains to prevent possible steric hindrance in an immunoassay format, we compare the blocking ability of a hydroxy-PEG-thiol containing seven ethylene glycol residues (hPEG7) with a hydroxy-PEG-thiol being eleven ethylene (34) www.piercenet.com. (35) Hoare, H. L.; Sullivan, L. C.; Ely, L. K.; Beddoe, T.; Henderson, K. N.; Lin, J.; Clements, C. S.; Reid, H. H.; Brooks, A. G.; Rossjohn, J. J. Mol. Biol. 2008, 377, 1297–1303. (36) Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M. Chem. ReV. 2005, 105, 1103–1169.

8698 Langmuir, Vol. 24, No. 16, 2008

Reimhult et al.

Table 1. Results Presented for the Steady State Adsorption of 100 µg/mL (Unless Otherwise Stated) Blocking Molecules to Au and Polystyrene Surfaces, Respectively Mw b (kDa)

mass coveragec (ng/cm2)

Au Au Au Au Au PS

66 0.71 0.39 0.56

PS PS PS PS PS

0.55 0.64 0.73 ∼29

535 ( 92 191 ( 14 244 ( 11 320 ( 12 1505 ( 99 329 ( 27 (506) 32 113 113 1267 873

blocking molecule

surface

BSA mPEG6 hPEG7 hPEG11 Caseina BSA (BSA 5 mg/mL) PEG12 mbPEG12 dbPEG12 PLL-g-PEG Caseina

66

Aad d (nm2) 20.5 ( 0.6 0.62 ( 0.04 0.27 ( 0.01 0.29 ( 0.01 33.3 ( 3.6 (22) 2.9 0.9 1.1 3.8

a The mass of the casein fragments are not known. b Molecular weight. Calculated using experimentally determine saturation frequency shifts and eq 1. d Geometrical surface area per adsorbed molecule (Aad), calculated using experimental mass coverage and eq 2. c

glycol residues long (hPEG11). The effect of different head groups has been investigated by comparing the seven ethylene glycol long hydroxy-PEG-thiol with a methoxy-PEG-disulfide being six ethylene glycol long when adsorbed as a thiolate on the gold surface (mPEG6). Disulfides are known to adsorb as thiolates by reductive cleaving of the S-S bond on the gold surface.37 Figure 1 presents the shifts in frequency with time when exposing clean Au or polystyrene surfaces to solutions containing various blocking molecules. The desorption of blocking agents upon buffer rinsing, measured as positive frequency shifts, was virtually undetectable in all cases except for casein where approximately 10% of the adsorbed mass desorbed (see Figure 1). Figure 1 shows that the binding kinetics is faster for the small molecule PEG derivatives as compared to BSA, casein and PLLg-PEG. This is partly due to a higher molar concentration of small PEG containing solutions and in part also because of the smaller size (see Table 1) and the resulting faster diffusion of the small PEGs. Except for PLL-g-PEG, the addition of fragmented casein caused the largest shift in frequency, and thus the largest shift in adlayer mass of the blocking molecules tested on gold and on polystyrene. The larger shift of casein can be explained by larger molecular size and a corresponding thicker adlayer, if we assume monolayer coverage. Size measurements performed using photon correlation spectroscopy (PCS) indicated two size populations of fragmented caseinsone population having an effective molecular diameter centered around 20 nm and the other one with an effective diameter centered around 150 nm. For comparison we have determined the PCS size of BSA to be 8 nm. Apparently the casein lysate consists of a broad distribution of casein fragment sizes with an appreciable part of caseins being much larger than BSA. Since casein lysate contains molecules of vastly different and large sizes the adsorbed casein facilitate trapping of water within the layer, which adds to the measured mass of the adlayer and results in high dissipation values. The increase of the dissipation monitored during casein adsorption was high on both polystyrene and gold surfaces (see Figure 2). For example, it was more than twice as high as the dissipation increase per unit frequency change monitored during BSA adsorption. Using eqs 1 and 2 the experimentally measured frequency shifts at saturation can be converted to the mean area per adsorbed (37) Noh, J.; Hara, M. Langmuir 2000, 16, 2045–2048.

Figure 2. Plot of the dissipation shift as a function of the frequency shift when adding 100 µg/mL BSA and casein to gold (Au) and polystyrene (PS), respectively.

molecule which is inversely proportional to the surface coverage. The results are shown in Table 1. Note that the calculated area per adsorbed molecule is the geometric surface area. The rootmean-square roughness of the Au electrode is small (e1 nm) and has not been taken into consideration. We find that the PEG thiols occupy the same area on the surface irrespective of the type of PEG. The PEG molecule with a disulfide group (mPEG6) covers twice the area of the PEG thiols (hPEG7 and hPEG11) (0.62 vs 0.27 and 0.29 nm2, respectively, see Table 1), which supports the idea that the disulfide splits up into two parts upon binding to the Au surface. Each of the disulfide PEG molecules may be regarded as two PEG thiolates on the surface, which agrees well with the data obtained from the experiments. In an idealized model for a PEG-terminated alkanethiolate monolayer on Au, with the PEG moieties adopting a helical gauche conformation, the molecules have a cross-section of 0.21 nm2.14 The obtained area of ≈0.28 nm2 for the PEG thiols is slightly larger, indicating either an incomplete monolayer coverage or adsorbate-adsorbate repulsion within the PEG adlayer that prevents higher density. It is also seen from Table 1 that the coverage of hydrophobic PEGs on polystyrene is much lower than the corresponding saturation coverage of PEG-thiolates on Au. However, the two PEGs that contain hydrophobic benzyl ether end groups (one group, mbPEG12; or two groups, dbPEG12) both form a denser layer compared to the PEG with no hydrophobic end group (PEG12). These trends can be rationalized by the expected trends in binding energies of each of the surface molecule complexes: the thiol-Au covalent bond is certainly stronger than the hydrophobic interaction between the styrene of polystyrene and the benzyl ether groups, which in turn is stronger than the passive adsorption of PEG12 to polystyrene. Binding of PEG to polystyrene via the benzyl group facilitates exposure of the EG groups outward from the surface. Since previous results on the blocking efficiency of BSA adsorbed on polystyrene indicated the need to use relatively high BSA concentrations to achieve efficient blocking,5–7 we have determined the BSA adsorption at concentrations much higher than the ones used for most of our blocking experiments. In Figure 3 we present the results obtained from the measurements of uptake of BSA on both Au and on polystyrene at different concentrations, including few results from the literature. The shift in frequency of BSA added to Au at 100 µg/mL used by us is only slightly lower than the previously reported BSA associated shift obtained at much higher concentration.9 The corresponding area per adsorbed molecule, Aad, decreases from 20.5 nm2 (Table 1) to about 18 nm2 at the higher concentration.9 The adsorption of BSA on polystyrene, on the other hand, increases substantially at higher BSA concentrations, but is

Blocking Agents on Gold and Polystyrene Surfaces

Figure 3. Plot of the frequency shift at BSA steady state uptake as a function of BSA concentration. Squares mark our data while circles mark data from refs 8 and 9, respectively, modified to correspond to a 5 MHz crystal.

generally lower than the BSA uptake on Au. The area occupied by BSA on polystyrene at 100 µg/mL is 33.3 nm2 (Table 1), while at 5 mg/mL Aad decreases to 22 nm2. The larger area occupied by BSA molecules on polystyrene can be explained by a larger repulsive adsorbate-adsorbate interaction on a low surface charge substrate (polystyrene), as compared with the interaction on Au where extra screening electrons from the metal are available. This observation is supported by AFM studies of BSA adsorption on SiO2 and polystyrene,38 and by the fact that BSA is charged at the neutral pH39 used here. The BSA size as determined by X-ray diffraction on albumin crystals is approximately 8 nm × 8 nm × 3 nm.40 which is in line with the effective BSA diameter as determined by us using PCS (8 nm). A closed packed monolayer of side-on adsorbed albumin would give a theoretical Aad ≈ 24 nm2. If one takes into account that the experimental saturation Aad value of 22 nm2 determined from Figure 3 includes also contribution from water (about 15% 9) we conclude that it is close to the theoretical saturation coverage of BSA on polystyrene. Therefore we believe that BSA adsorbs side-on on polystyrene. Note that the experimentally observed Aad determined from BSA uptakes on Au is substantially smaller than the theoretical Aad (by g30%). This can be explained either by strong denaturation of BSA adsorbed on Au, or by partial denaturation followed by the formation of multilayer aggregates,29,38 or both. Films formed by denatured BSA should be stiffer than those formed by the native BSA, which implies that on Au the dissipation changes per mass increase should be smaller than the corresponding values on polystyrene. This indeed has been observed during the experiments shown in Figure 1, see Figure 2 (note that the dissipation and the frequency are monitored simultaneously). Figure 4 summarizes the experimentally observed saturation levels of nonspecific binding, obtained after serial exposure of the blocked surfaces to the two test solutions: one containing 95% pure prostate specific antigen (PSA) and the other containing PSA in seminal plasma in PBS, respectively. It shows the results of saturation mass adsorption obtained on blocked Au and on blocked polystyrene surfaces, respectively, after several exposures to each of the test solutions and subsequent rinsing. During the rinsing some of the loosely bound seminal plasma desorbed in all cases. Smaller mass shifts correspond to lower nonspecific adsorption and thus to better blocking ability. The change in (38) Valle-Delgado, J. J.; Molina-Bolivar, J. A.; Galisteo-Gonzales, F.; GalvezRuiz, M. J.; Feiler, A.; Rutland, M. Phys. Chem. Chem. Phys. 2004, 6, 1482– 1486. (39) www.albumin.org. (40) Sugio, S.; Kashima, A.; Mochizuki, S.; Noda, M.; Kobayashi, K. Protein Eng. 1999, 12, 439–446.

Langmuir, Vol. 24, No. 16, 2008 8699

mass was calculated using the corresponding shifts in frequency during the QCM measurements and eq 1. In all the experiments we have observed that 95% pure PSA antigen show no or negligible binding to the blocked Au surfaces and also very little uptake onto polystyrene blocked by either BSA, PLL-g-PEG, or casein (Figure 4). The proteins and possibly other molecules present in seminal plasma on the other hand do bind to the blocked surfaces. The results presented in Figure 4 show that the Au surfaces that adsorb the smallest amount of proteins are those carrying a monolayer of PEG-thiols, closely followed by the casein coated surfaces. The mechanism with which PEGylated surfaces suppress protein adsorption is still under debate, but factors such as chain density, chain length, chain conformation and charge neutrality are considered important, as is the interaction of PEG with water.12,14,41,42 Unsworth and co-workers12 showed that the PEGthiol chain density, rather than the chain length (albeit for much longer PEG chains), is a key property for suppression of protein adsorption to modified gold surfaces. They also showed that the chemistry of the exposed end group may influence the ability to block nonspecific binding, with hydroxyl-terminated PEG-thiols giving better results than methyl-terminated. The same effect of the end group has been shown for hydroxyl- and methylterminated long-chain alkanethiolates on gold.13 We have detected only minor differences in blocking efficiency when comparing the different PEG thiols adsorbed onto Au, with hydroxylterminated PEG thiols giving slightly better blocking performance than the methyl-terminated PEG thiol (see, Figure 4). The most protein rejecting polystyrene surfaces are those covered by casein, BSA and PLL-g-PEG, respectively (Figure 4). The poor performance of BSA covered Au surfaces (Figure 4), despite the higher BSA adsorption as compared to the polystyrene surface (Table 1), is most probably due to (partial) denaturation of adsorbed BSA which reduces its blocking ability due to exposure of normally hidden hydrophobic regions. Casein shows a much better blocking than BSA on gold probably because it consists of much larger molecules and the outermost casein layer is shielded from the denaturing metal, preserving its blocking ability. The blocking ability of casein is however best on polystyrene where it performs at least equally well as, or better than, both BSA and PLL-g-PEG. Note however that we do detect some adsorption of biomolecules from seminal plasma for all blocking agents on both surfaces. For example, from Figure 4 we estimate the blocking efficiency of casein on polystyrene to be better than 95% and this of BSA to be better than 90% in agreement with the results reported earlier,7 but obtained after exposures of the substrate to substantially lower BSA concentrations (100 µg/mL as compared to 800 µg/mL7). Contrary to the results obtained using commercial ELISA wells5–7 our experiments show that the blocking efficiency of dense BSA films on polystyrene (Aad ≈ 22 nm2) formed by exposing polystyrene to solutions with BSA concentration of 5 mg/mL was virtually identical to the blocking efficiency obtained using less dense films (Aad ≈ 33,3 nm2, see Figure 4 and Table 1) formed by exposing the sample to solutions with BSA concentration of 100 µg/ml. There are two obvious differences between the surfaces of commercial ELISA plates5–7 and the QCM crystal surfaces investigated by us (and by Feiler et al.8). The QCM crystal surfaces are most probably cleaner than the well surfaces of the commercial products and, second, the surfaces used here are very smooth. (41) Malmsten, M.; Emoto, K.; Van Alstine, J. M. J. Colloid Interface Sci. 1998, 202, 507–517. (42) Jeon, S. I.; Lee, J. H.; Andrade, J. D.; De Gennes, P. G. J. Colloid Interface Sci. 1991, 142, 149–158.

8700 Langmuir, Vol. 24, No. 16, 2008

Reimhult et al.

Figure 4. Calculated mass shifts when adding 95% pure PSA and PSA in seminal plasma to gold (Au) and polystyrene (PS) surfaces blocked with 100 µg/mL (unless otherwise stated) BSA, PLL-g-PEG, Casein or PEG-derivatives.

In addition, the experiments reported by Vogt et al.7 have been performed at a higher temperature (37 °C) compared to the ones reported here (23 °C). Both the higher impurity levels, larger surface roughness and the higher temperature employed would contribute to shift monolayer completion to higher concentrations, and would also partly explain why the results obtained using commercial products vary from one product to another.5,7 The mass shift associated with our PLL-g-PEG coating (1267 ng/cm2, Table 1) was significantly higher than the mass shift corresponding to 1 monolayer.17 The blocking efficiency of the copolymer applied to polystyrene-substrates was however inferior to casein (Figure 4), and to similar polymer films formed on oxides17,18 where it blocked virtually all protein adsorption, which may signal the intricate interplay between the substrate charge state and the copolymer morphology.18 Blocking the polystyrene surface with the small molecule PEG derivatives gave the poorest result with large nonspecific uptakes of both pure PSA antigen and of biomolecules from seminal plasma. There are several possible explanations for why short PEGs perform poorer on polystyrene than on Au. One possibility may be that the PEGs adsorb randomly and lying down on polystyrene, which results in less flexibility of the PEG chains and fewer hydrogen bonding water molecules due to multiple binding points of the PEG chain to the surface. High flexibility and the binding of interfacial water have been shown to be of great importance for the blocking effect of PEG.14 Another explanation of the increased nonspecific binding is the low surface coverage of PEG on polystyrene. The PEG surface density has been shown to be of importance in preventing protein adsorption.12,42 On the polystyrene surface the benzyl ether PEG derivatives are far less crowded than the corresponding PEG on Au, with each molecule occupying more than three times as much space as the PEG thiols do on gold (see Table 1). This leaves the polystyrene surface largely unprotected between the PEG chains, which may explain the higher nonspecific binding on benzyl ether PEG covered polystyrene surfaces. Yet another possibility is that the strength of the hydrophobic interaction is not sufficient to prevent exchange of adsorbed PEGs by biomolecules on the surface.

Conclusions Our aim with the study was to find a molecule that is suitable to use for blocking of surfaces directly coated with antibodies

without any spacer layer. The functionalized surfaces are after blocking to be used in a PSA sandwich immunoassay. To summarize the results, the PEG thiols proved to be better for blocking gold than the in immunochemistry commonly used BSA and casein. Apart from efficient prevention of nonspecific protein adsorption their linear dimensions are certainly smaller than the extent of an antibody, which rules out the possible steric hindrance for the subsequent antigen adsorption in an immunoassay. We therefore conclude all PEG thiols tested to fulfill our requirements for use to block nonspecific adsorption on antibody-functionalized Au surfaces. Casein, BSA or PLL-g-PEG are to prefer as blocking molecules on polystyrene surfaces compared to the evaluated benzyl ether PEG derivatives. To obtain dense BSA films on polystyrene high albumin concentrations, in excess of 1 mg/mL, are needed. Despite this finding, the blocking efficiency of BSA on polystyrene does not improve appreciably with coverage for the concentrations tested. It is virtually identical for films formed at 100 µg/mL as compared with adlayers formed at 5 mg/mL. The poor result of the short PEG derivatives on polystyrene is most probably due to the low PEG coverage and an insufficient binding strength of single hydrophobic PEGs to polystyrene. In future experiments dealing with optimization of immunoassays BSA will be chosen as the preferred blocking molecule on polystyrene mainly for two reasons: (i) it is a commonly used blocking agent and is much cheaper than PLL-g-PEGs are and (ii) it is certainly smaller and more uniform in size as compared to casein. The latter decreases the danger of steric hindrance for the specific adsorption of biomolecules to blocked and functionalized surfaces. For the blocking molecules chosen for coming PSA assay experiments the adsorption of pure PSA has been negligible while the mass increase due to nonspecific adsorption of biomolecules from seminal plasma was limited. Acknowledgment The authors thank CanAg Diagnostics AB (now Fujirebio) for supplying PSA in seminal plasma, and particularly Dr. Olle Nilsson for many illuminating discussions. We also thank Polypure AS for supplying the different PEG molecules and for valuable discussions. LA800224S