Reducing Nonspecific Adhesion on Cross-Linked Hydrogel Platforms

Nov 12, 2005 - Shawn D. Carrigan and Maryam Tabrizian* ... on a cross-linked layer of poly(ethylene glycol) (PEG) and PEI and employs an anti-IgG Fc...
0 downloads 0 Views 553KB Size
12320

Langmuir 2005, 21, 12320-12326

Reducing Nonspecific Adhesion on Cross-Linked Hydrogel Platforms for Real-Time Immunoassay in Serum Shawn D. Carrigan and Maryam Tabrizian* McGill University, Biomedical Engineering Department, 3775 University St., Duff Medical Building, Room 316, Montreal, QC, Canada H3A 2B4 Received July 27, 2005. In Final Form: October 4, 2005 Biointerfaces that limit nonspecific adhesion of serum proteins have been developed by relying solely on cross-linked hydrogels. In addition to being characterized for adhesion of serum proteins, immunoassay sensitivity was also investigated through a sandwich assay for rhIL-1ra. Among the compositions developed, the optimal surface is comprised of pre-cross-linked carboxymethylcellulose (CMC) and polyethyleneimine (PEI) overlaid on a cross-linked layer of poly(ethylene glycol) (PEG) and PEI and employs an anti-IgG Fc specific ligand for oriented antibody immobilization; viscoelastic modeling provides a thickness estimate of 5 nm for the hydrogel alone, rising to 33 nm after the deposition of antibodies. Alternate compositions employing a Protein A ligand and PEG at the exposed surface of the biointerface were disfavored due to an 8-fold increase in serum adhesion and retarded immobilization kinetics, respectively. Through the rapid deposition provided by hydrogels, construction of the entire biointerface, including receptor immobilization, can be completed in 1 h. Based on QCM-D measurements, estimated nonspecific serum adsorption using these compositions is as low as 1.1 ng/mm2. The immunoassay as developed requires 10 min, providing a detection limit of 500 ng/mL rhIL-1ra in 25% human serum using only 5 µg of the secondary antibody.

Introduction Continuing development of real-time nonlabeled immunoassay protocols aims to improve sensitivity and specificity performance so that rapid sample analysis is feasible with detection limits similar to those of common fluorescent and incubated assays. Ultimately, such methods could provide clinicians with an affordable bed-side diagnostic system for on-line monitoring of disease biomarkers. Quartz crystal microgravimetry (QCM) and surface plasmon resonance (SPR), acoustic and optical mass sensors respectively, are the prevalent real-time transduction technologies used in immunoassay. Despite the advanced maturity of SPR-based methods, contributable to its earlier development as a viable biosensing platform, recent direct comparisons of both technologies indicate equivalent performance for DNA hybridization and protein detection assays.1,2 Present research on the design of biointerfaces for use in real-time methods is currently divided between development of hydrogel platforms and functionalized surface coatings. In affinity sensing, the biointerface needs to provide a viable immobilization platform for a specific capture molecule while resisting nonspecific adsorption of biomolecules present in the sample being analyzed. As recently as one year ago, developments to this end focused primarily on the employ of functionalized self-assembled monolayers (SAM) of sulfur-bearing molecules or plasma polymerized films (PPF);3-6 continuing investigations remain prevalent in the field.7-9 These two-dimensional coatings result in fewer active receptor sites for biomol* Corresponding author. E-mail: [email protected]. (1) Laricchia-Robbio, L.; Revoltella, R. P. Biosens. Bioelectron. 2004, 19, 1753-1758. (2) Su, X.; Wu, Y. J.; Knoll, W. Biosens. Bioelectron. 2005, in press. (3) Ostuni, E.; Grzybowski, B. A.; Mrksich, M.; Roberts, C. S.; Whitesides, G. M. Langmuir 2003, 19, 1861-1872. (4) Herrwerth, S.; Rosendahl, T.; Feng, C.; Fick, J.; Eck, W.; Himmelhaus, M.; Dahint, R.; Grunze, M. Langmuir 2003, 19, 18801887. (5) Wang, H.; Li, D.; Wu, Z. Y.; Shen, G. L.; Yu, R. Q. Talanta 2004, 62, 201-208.

ecule recognition than their three-dimensional hydrogel counterparts, aiming instead to increase assay sensitivity by reducing the nonspecific protein adhesion known to occur with hydrogels. However, current trends indicate renewed interest in hydrogel platforms due to their favorable performance and regeneration characteristics,10-12 with a majority of reported platforms continuing to rely on lengthy SAM under-layer preparations for attachment to the sensor substrate. In addition to developments focusing on improved biointerfaces through material functionality, immunoassay methodologies are beginning to benefit from nanoscale architecture manipulation. Both monolayer and PPFbased sensors are now beginning to employ nanoscale features in their designs to control the density of immobilized capture molecules, thereby directly impacting subsequent antigen recognition.7,8 Protein engineering further demonstrates the potential of nanotechnology, with the reduced size of single-chain fragment variable (scFv) antibodies and combinatorially constructed affinity proteins allowing greater immobilization density of active recognition sites while minimizing adsorbed mass, which in turn maximizes mass-based assay sensitivity.9,13 These new avenues in the field of biointerface research likely represent the future of affinity biosensing. However, (6) Zhou, C.; Friedt, J. M.; Angelova, A.; Choi, K. H.; Laureyn, W.; Frederix, F.; Francis, L. A.; Campitelli, A.; Engelborghs, Y.; Borths, G. Langmuir 2004, 20, 5870-5878. (7) Haes, A. J.; Chang, L.; Klein, W. L.; Van Duyne, R. P. J. Am. Chem. Soc. 2005, 127, 2264-2271. (8) Wang, H.; Wu, J.; Li, J.; Ding, Y.; Shen, G.; Yu, R. Biosens. Bioelectron. 2005, 20, 2210-2217. (9) Shen, Z.; Stryker, G. A.; Mernaugh, R. L.; Yu, L.; Yan, H.; Zeng, X. Anal. Chem. 2005, 77, 797-805. (10) Masson, J. F.; Battaglia, T. M.; Davidson, M. J.; Kim, Y. C.; Prakash, A. M. C.; Beaudoin, S.; Booksh, K. S. Talanta 2005, in press. (11) Stigter, E. C. A.; de Jong, G. J.; van Bennekom, W. P. Biosens. Bioelectron. 2005, 21, 474-482. (12) Wu, T. Z.; Su, C. C.; Chen, L. K.; Yang, H. H.; Tai, D. F.; Peng, K. C. Biosens. Bioelectron. 2005, in press. (13) Renberg, B.; Shiroyama, I.; Engfeldt, T.; Nygren, P. K.; Karlstrom, A. E. Anal. Biochem. 2005, 341, 334-343.

10.1021/la052046h CCC: $30.25 © 2005 American Chemical Society Published on Web 11/12/2005

Serum Adhesion Resistance on Biointerfaces

Langmuir, Vol. 21, No. 26, 2005 12321

Figure 1. Schematic representation of the various biointerface compositions and construction sequences employed. Table 1. Construction and Mass Ratio Compositions of the Various Hydrogel Biointerfaces Studied composition

sublayer

1

0.5 mg/mL 1800 Mw PEI

2

0.5 mg/mL 1800 Mw PEI

3

0.5 mg/mL 1800 Mw PEI cross-linked to 10 mg/mL PEG (base layer) 0.5 mg/mL 1800 Mw PEI

4

emerging methods drastically increase the complexity and associated cost of immunoassays in the short term. Considering an ultimate application of affordable realtime monitoring of disease biomarkers, our research prioritizes design parameters to consider simplicity, and thereby cost, in conjunction with standard immunoassay performance parameters. Previously, we reported the development of hydrogel-based biointerfaces for QCM-D which offer strong immunoassay characteristics while providing the advantages of rapid and facile surface preparation, and affordability of materials.14,15 These biointerfaces immobilize monoclonal antibodies for the immunoassay of sepsis-related biomarkers using watersoluble hydrogel biointerfaces that require as little as three minutes for deposition. To further improve upon these biointerfaces, we focused on development characteristics that would lead toward clinically applicable platforms. Specifically, the main improvement sought over previous biointerfaces was an ability to detect small molecules in serum, where an abundance of arrayed protein sizes and shapes would normally confound measurements through a combination of blocking immobilized receptors and nonspecific mass adsorption to the biointerface. Given the facile surface preparation previously obtained through the employ of hydrogels alone, we sought to develop biointerfaces which could be constructed using equally simple methods while (14) Carrigan, S. D.; Scott, G.; Tabrizian, M. Biomaterials 2005, 26, 7514-7523. (15) Carrigan, S. D.; Scott, G.; Tabrizian, M. Langmuir 2005, 21, 5966-5973.

hydrogel composition

anti-hIL-1ra ligand

50:1 pre-cross-linked CMC: 600 Mw PEI 50:1 pre-cross-linked CMC: 600 Mw PEI 50:1 pre-cross-linked CMC: 600 Mw PEI 331/3:331/3:1 pre-cross-linked CMC:PEG:600 Mw PEI

anti-mouse IgG Fc specific Protein A anti-mouse IgG Fc specific anti-mouse IgG Fc specific

providing the necessary resistance to nonspecific serum protein adhesion. In this work, we have characterized new hydrogel-only compositions based on their ability to provide sensitive immunoassay platforms which resist serum fouling of the biointerface. Specifically, compositions have been developed which incorporate two separate ligands for oriented antibody immobilization and three hydrogel compositions consisting of polyethyleneimine (PEI), carboxymethylcellulose (CMC), and poly(ethylene glycol) (PEG). Materials and Methods Quartz Crystal Microgravimetry with Dissipation (QCMD). Characterization of the biointerface preparation and functionality was assessed using QCM-D (D300, Q-Sense). Frequency and dissipation measurements were acquired using the 3rd through 7th overtones on 5 MHz gold crystals. All experiments were performed in triplicate at 25 °C using 10 mM HEPES, 0.15 M NaCl buffer solution (Biacore International AB). Biointerface Preparation. Four hydrogel biointerface compositions were constructed using combinations of carboxymethylcellulose (CMC) (Carbomer), poly(ethylene glycol) (PEG) diacid (Mw ) 600) (Fluka), and polyethyleneimine (PEI) (Mw ) 600 & 1800) (Alfa Aesar), as defined in Table 1 and depicted in Figure 1. The initial composition, representing a combination of previously developed hydrogel chemistries,14,15 employs a PEI (1800 Mw) sublayer which is subsequently cross-linked upon deposition of the pre-cross-linked CMC:PEI (600 Mw). The pre-cross-linked CMC:PEI solution is prepared by mixing CMC and PEI with N-hydroxysuccinimide (NHS; Fluka) and 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC; Avocado Research Chemicals Ltd.),16,17 which activates the functional groups

12322

Langmuir, Vol. 21, No. 26, 2005

to allow cross-linking through amine-coupling. The presence of NHS and EDC in this solution also activates the sublayer, resulting in cross-linking between the CMC:PEI and sublayer. Herein, a colon will be used to denote hydrogels that are precross-linked prior to deposition (e.g., CMC:PEI). A rabbit antimouse IgG Fc specific ligand (Pierce Biotechnology) is then immobilized for subsequent oriented immobilization of monoclonal mouse antibodies. The second biointerface composition is identical to the initial chemistry, with the substitution of Protein A (Pierce Biotechnology) to provide oriented antibody immobilization in place of the anti-mouse IgG antibody. The final two compositions, in an attempt to further minimize nonspecific adsorption, employ PEG chains with terminal carboxylic acid groups which can be cross-linked to the other hydrogels through amine-coupling chemistry. The third composition employs a base layer of PEG that is then activated with NHS-EDC, with the remaining construction following the sequence as defined for the initial composition. To differentiate from the pre-cross-linked notation, a dash will be used to denote that PEI is subsequently cross-linked to PEG for this composition (i.e., PEG-PEI). The final composition employs the same construction sequence as the initial composition, with the exception that PEG is included in the pre-cross-linked solution with CMC and PEI in a 331/3:331/3:1 mass ratio. Both PEG-based compositions employed the anti-mouse IgG Fc specific antibody ligand. Following deposition of the antibody capture ligand, monoclonal mouse anti-hIL-1ra antibodies (R&D Systems) are immobilized through either the anti-mouse IgG or Protein A ligand. All compositions are then deactivated using ethanolamine hydrochloride pH 8.0 (Alfa Aesar), and passivated using Superblock (Pierce Biotechnology) to further prevent nonspecific adsorption. Biointerface preparation is completed by regenerating the surfaces using injections of 100 mM glycine hydrochloride (Sigma-Aldrich) followed by a 2:1 solution of running buffer and a mixture of 0.6 M potassium thiocyanate, 0.9 M urea, 1.8 M magnesium chloride, and 1.8 M guanidine hydrochloride.18 Biointerface Characterization. The performance of each biointerface composition was characterized with regards to a series of parameters. Antigen response was assessed based on the QCM-D response to a sandwich assay using a secondary goat anti-IL-1ra antibody (R&D Systems). Concentrations of rhIL1ra (R&D Systems) from 25 to 500 ng/mL in buffer were injected (150 µL) and incubated for 5 min, followed by 50 µg/mL of the secondary antibody (100 µL) for an additional 5 min. Nonspecific adhesion of the secondary antibody to the biointerfaces was also characterized both before and after the series of antigen exposures to ensure that responses measured sandwiched rhIL-1ra. Serum adhesion was also assessed, through exposure to 25% human serum in buffer, following both a buffer rinse and surface regeneration. Serum was diluted 1:3 with running buffer and filtered in 0.22 µm cellulose acetate centrifuge tube filters (Corning). Following these characterizations, the third composition employing the PEG sublayer was subsequently employed to test antigen sensitivity in the range of 0.5-25 ng/mL in buffer, and 25-500 ng/mL in serum. Film Parameter Modeling. Estimates for the effective thickness and density, as well as storage and loss moduli, of the adhered biointerfaces were calculated using Q-Tools software (Q-Sense). Based on continuum mechanics calculations for a Voight film, the viscoelastic parameters were estimated using a fitting procedure to minimize differences between measured and calculated values for the 3rd through 7th frequency overtones and the 5th dissipation overtone (D5) for 10 data points at each step of the biointerface preparation.19 To improve model fitting, minimal frequency dependence factors for both the layer loss (0.005) and storage moduli (0.1) were employed.20 Statistics. Frequency and dissipation shifts were quantified using MATLAB software (The Mathworks) to calculate average (16) Cuatrecasas, P.; Parikh, I. Biochemistry 1972, 11, 2291-2299. (17) Lofas, S.; Johnsson, B.; Edstrom, A.; Hansson, A.; Lindquist, G.; Hillgren, R. M. M.; Stigh, L. Biosens. Bioelectron. 1995, 10, 813-822. (18) Andersson, K.; Hamalainen, M.; Malmqvist, M. Anal. Chem. 1999, 71, 2475-2481. (19) Voinova, M. V.; Rodahl, M.; Jonson, M.; Kasemo, B. Phys. Scripta 1999, 59, 391-396. (20) Larsson, C.; Rodahl, M.; Hook, F. Anal. Chem. 2003, 75, 50805087.

Carrigan and Tabrizian

Figure 2. Impact of different rabbit anti-mouse IgG ligands on QCM-D frequency adsorption characteristics, and on subsequent anti-hIL-1ra adsorption, on CMC:PEI composition with and without the PEI sublayer. The former anti-IgG ligand, which prohibited repeat measurements through incomplete antigen release, provides greater ligand and subsequent antibody adsorption. The replacement ligand provides more rapid antibody adsorption and permits complete antigen release. values at specified time points in raw data files. Bar charts plot average F5/5 and D5 values, with error bars representing standard deviations. Statistically significant differences were identified using two-tailed Student’s T-tests forp < 0.05.

Results and Discussion Biointerface Construction. The biointerfaces developed herein were conceived with the goal of advancing previously developed cross-linked hydrogel immunoassay platforms by providing for more rapid antibody immobilization during preparation and for serum testing capability. Based on previous findings, which demonstrate the utility of a PEI sublayer in resisting nonspecific protein adsorption, two initial compositions are characterized that employ a pre-cross-linked hydrogel composition overlaid on and cross-linked to PEI. These compositions employ oriented antibody immobilization through either an antiIgG Fc specific or Protein A ligand. Separately, two compositions employing the anti-IgG ligand also incorporate short chain (Mw ) 600) PEG in an attempt to further reduce nonspecific protein interaction. The first of these compositions involves deposition of an initial PEG sublayer, which upon activation is cross-linked to PEI. The second PEG-based composition relies on the standard PEI sublayer and incorporates PEG in the pre-cross-linked composition (Figure 1). During preliminary characterizations to determine the impact of the PEI sublayer on a previously developed CMC: PEI composition,15 it was discovered that antigen release during surface regeneration was incomplete using the antimouse IgG (rabbit) ligand. This hindrance to developing a multi-use real-time platform was overcome by substituting an anti-mouse IgG (rabbit) ligand purchased from an alternate supplier and by including a strong ionic solution in the regeneration sequence.18 In comparing adsorptions of both ligand and anti-hIL-1ra using the problematic ligand, the inclusion of a PEI sublayer is found to retard ligand adhesion kinetics while simultaneously resulting in an unidentified “hump” in the adsorption profile (Figure 2, curves b and c). This trend is similarly evident with the replacement ligand, with a notable reduction in the adsorbed mass of ligand (Figure 2, curves a and b). However, the replacement ligand results in substantially improved anti-hIL-1ra adhesion kinetics,

Serum Adhesion Resistance on Biointerfaces

Langmuir, Vol. 21, No. 26, 2005 12323

Table 2. Comparison of the Overall Baseline Shifts Resulting from the Completed Hydrogel Biointerfaces Indicates No Significant Difference in Frequency or Dissipative Characteristics of the Hydrogels Alone cumulative shift from cross-linked hydrogels composition

F5/5 shift (Hz)

D5 shift (1E-6)

pre-cross-linked CMC:PEI (developed previously15) pre-cross-linked CMC:PEI on PEI sublayer pre-cross-linked CMC:PEI on PEG-PEI sublayer pre-cross-linked CMC:PEG:PEI on PEI sublayer

-33.3 ( 6.5 -43.7 ( 13.0 -39.4 ( 5.4 -33.8 ( 2.3

5.0 ( 0.9 5.8 ( 1.5 3.5 ( 1.8 4.2 ( 0.3

allowing the contact time during antibody immobilization to be reduced from 60 to 20 min. All further experiments employ the replacement anti-IgG ligand. Preparation for each biointerface composition remains rapid, with the longest hydrogel deposition sequence, involving the PEG-PEI sublayer, requiring 10 min. Complete preparation time for this biointerface, including deactivation and passivation, requires 1 h. Comparison with recent protocols utilizing functionalized surface SAM and PPF coatings, as well as nanoscale manipulations, indicate deposition times ranging from overnight to 48 h.7-9 Characterization by QCM-D of the various compositions indicates a strong degree of similarity between the various polymer constructions, as indicated by uniformity of both frequency and dissipation measurements (Table 2). Interestingly, neither the PEI sublayer nor the combined PEG-PEI sublayer, significantly alters the mass of subsequently adhered pre-cross-linked polymers. However, the various surface compositions do impact the ensuing biointerface preparation steps. In assessing the impact of different sublayers, there is no significant difference in either adsorbed mass or adhesion kinetics of the anti-IgG ligand or anti-hIL-1ra antibody when using the PEI or PEG-PEI sublayers, indicating that the effect of the PEG on surface preparation

Figure 3. Adsorption characteristics of the ligands and antihIL-1ra antibodies on the various biointerface compositions. Protein A (2) provides faster adhesion kinetics but reduced ligand and antibody adsorption. The composition with PEG at the surface (4) provides significantly slower ligand and antibody adhesion kinetics.

is likely overshadowed by cross-linking it with PEI and subsequently blanketing the PEG-PEI sublayer with CMC:PEI (Figure 3, curves 1 and 3, and Figure 4). Inclusion of PEG at the exposed biointerface surface, as with the CMC:PEG:PEI composition, however, does impact the adhesion rates of both ligand and antibody (Figure 3, curve 4). Although extending the binding times permitted the adsorbed masses of ligand and antibody to equal those achieved with compositions lacking PEG at the exposed surface, the significant mass desorption obtained during surface passivation and regeneration may result from noncovalently bound antibodies being desorbed from the hydrogel (Figure 4). Thus, the inclusion of PEG at the surface of the cross-linked biointerface is suspected to reduce the overall antibody immobilization density following completion of the biointerface preparation by electrostatically repelling the antibodies and preventing the necessary covalent attachment. Finally, in comparing the Protein A (42 kDa) and anti-IgG (150 kDa) ligands, significantly reduced mass adsorption of both the ligand and anti-hIL-1ra is found in the case of Protein A (Figure 4), despite more rapid adhesion kinetics for both molecules (Figure 3, curve 2). Although the diminished adsorbed mass of the ligand is attributable to the difference in molecular mass between the two ligands, and is consistent with previous findings,15 reasoning for the considerable difference in antibody immobilization remains undetermined. Modeling Biointerface Construction. To further characterize the hydrogel biointerface preparations, modeling was used to estimate film parameters based on a Voight viscoelastic model.19 Given the dependency of QCM frequency measurements on adhered mass, specification of the buffer density and viscosity, in conjunction with measured overtones, permits the model to estimate the effective film thickness and density. In general, fitted parameters (F3, F5, F7, and D5) deviated from the actual

Figure 4. Stepwise shifts demonstrate the frequency changes resulting from the individual steps involved in preparing the various biointerface compositions. “f” indicates significant difference from the anti-IgG on CMC:PEI composition (p < 0.05), (n )3).

12324

Langmuir, Vol. 21, No. 26, 2005

Carrigan and Tabrizian

Figure 6. Nonspecific adsorption of the secondary antibody employed in sandwich immunoassays, as monitored by frequency (top bars, left ordinate) and dissipation (bottom bars, right ordinate) measurements, before and after antigen testing on the various biointerface compositions. Testing was conducted to ensure that antigens were released during surface regeneration. “f” indicates significant difference from the anti-IgG on CMC:PEI composition (p < 0.05), (n ) 3).

Figure 5. Modeling predictions for the biointerface thickness (upper) and calculated mass (lower) following each step of the respective constructions. Tabular stepwise changes to the aerial film density resulting from sequential preparation steps are inset in the bottom chart (n ) 3).

measurements by at most only a few tenths of one percent, though in a few instances deviations of a few percent could not be further reduced. Based on this modeling, the only noted difference in biointerface preparations using the various compositions lays in the reduced adhered mass of anti-hIL-1ra adsorption on the Protein A composition (Figure 5). Estimated thickness of the hydrogels prior to ligand and antibody adsorption is in the range of 5-10 nm. Though this figure is consistent with preliminary ellipsometry measurements (data not shown), accurate ellipsometric verification of such a thin film on polished gold QCM-D crystals is not feasible due to the roughness of the substrate. Although optically determined values of the hydrodynamic thickness of CMC adsorbed on inorganic particles indicates that these thicknesses may be slightly underestimated,21 no values are available for the crosslinked polymers used in this instance. Alternative verification, such as depositing the hydrogel on an ultra-flat gold surface for optical measurements, was not pursued given that previous experience demonstrated inconsistencies in the deposition of such thin films on surfaces whose roughness is not comparable. Inconsistencies in immunoassay performance have also been attributed to such seemingly minor changes in substrate characteristics.13 Following anti-hIL-1ra adsorption, the estimated film thicknesses rise to a range of 17-33 nm, resulting in net increases that correlate well with optically obtained values (21) Hoogendam, C. W.; Peters, J. C. W.; Tuinier, R.; de Keizer, A.; Stuart, M. A. C.; Bijsterbosch, B. H. J. Colloid Interface Sci. 1998, 207, 309-316.

for protein adsorption on hydrophobic silicon wafers22 and for IgG adsorption on functionalized polymers.23 However, corresponding estimates for aerial density increases range from 6.6 to 28.7 ng/mm2, which is considerably larger than radioactively measured values for IgG immobilization on an alternate polysaccharide composition.24 Verification of such high mass adsorptions would further justify the utility of these hydrogel compositions. Despite the strong correlation between estimates for the thickness and adsorbed mass (Figure 5), estimated masses provide more reliable figures given the inability of QCM-D modeling to sufficiently differentiate the extent to which the film density and thickness contribute to the adhered mass.20 Ultimately, however, until the hydrodynamic film thickness can be accurately determined using the QCM-D crystal substrate, caution need be employed in interpreting these estimated film parameters. Nonspecific Interaction. Given the noted difficulties in antigen release during preliminary investigations, and to ensure that immunoassay responses to the secondary antibody were not inflated by nonspecific interactions, each composition was characterized for binding of the secondary antibody prior to and following antigen testing. Relative to the anti-IgG ligand on the PEI sublayer composition, only the CMC:PEG:PEI pre-cross-linked composition demonstrated significantly elevated nonspecific interaction between the secondary antibody and the biointerface (Figure 6). Subsequent verification demonstrated an affinity of the anti-mouse IgG ligand for the secondary antibody (goat), the likely source of the minimal nonspecific interaction between the antibody and the various biointerface compositions (data not shown). To assess the suitability of these platforms for serum testing, each composition was exposed to 25% human serum to determine the level of nonspecific protein adsorption. Relative to the CMC:PEI composition on a PEI sublayer, only the composition utilizing Protein A demonstrates substantially elevated protein adhesion (Table 3). Serum exposure results in an 8-fold increase in (22) Hook, F.; Kasemo, B.; Nylander, T.; Fant, C.; Sott, K.; Elwing, H. Anal. Chem. 2001, 73, 5796-5804. (23) Preininger, C.; Clausen-Schaumann, H.; Ahluwalia, A.; de Rossi, D. Talanta 2000, 52, 921-930. (24) Caelen, I.; Gao, H.; Sigrist, H. Langmuir 2002, 18, 2463-2467.

Serum Adhesion Resistance on Biointerfaces

Langmuir, Vol. 21, No. 26, 2005 12325

Table 3. Adhesion Characteristics on Various Biointerface Compositions Following Exposure to 25% Human Serum and Rinsing, and Remaining Adhered Proteins Following Surface Regenerationa

a The shaded row represents adhesion results on the PEG-PEI sublayer composition with the ligand and antibodies omitted. “*” indicates significant difference from the anti-IgG on CMC:PEI composition (p < 0.05), (n ) 3).

Figure 7. Frequency (top bars, left ordinate) and dissipation (bottom bars, right ordinate) responses of the various biointerface compositions to the secondary antibody following antigen exposure in buffer. “f” indicates significant difference from the anti-IgG on CMC:PEI composition (p < 0.05), (n ) 3).

adhesion on the Protein A surface, as well as a 5-fold increase in residual adherent proteins following surface regeneration. Overall, the PEG-PEI sublayer provides the most favorable characteristics, with no residual proteins remaining following surface regeneration. Further testing on this composition also reveals that additional rinsing at higher flow rates following serum exposure detaches roughly half of the adherent proteins, resulting in reduced residual adhesion levels prior to regeneration (data not shown). Modeling adsorption at this reduced level of adhesion estimates nonspecific mass adsorption of 1.1 ng/ mm2, comparing well with recent investigations of both various SAM and alternate hydrogel configurations.10,11 Characterization of this composition in the absence of ligand and antibodies reveals that the hydrogel itself is resistant to nonspecific interaction, indicating that nonspecific adhesion is likely the result of immunoglobulins interacting with the antibody-orienting ligands or antibodies. Interference of this type has been noted by others and requires similar assay-specific investigation to minimize these interactions.25 Antigen Response. Antigen testing, conducted over a range of 25-500 ng/mL rhIL-1ra and characterized solely by the response to secondary antibodies following antigen exposure, indicates that the CMC:PEI composition with a PEI sublayer results in the lowest antigen response (Figure 7). Comparison of the Protein A composition to previous data that excluded the PEI sublayer finds no significant difference in measured response at an antigen concentration of 25 ng/mL, indicating that the PEI sublayer does not diminish sensitivity. Although the (25) Martins, T. B.; Pasi, B. M.; Litwin, C. M.; Hill, H. R. Clin. Diagn. Lab Immunol. 2004, 11, 325-329.

Figure 8. Sandwich immunoassay response to secondary antibodies over an expanded concentration range in buffer using the anti-IgG composition on CMC:PEI composition employing the PEG:PEI sublayer, (n ) 3). The indicated responses to blank are the secondary antibody adsorptions following buffer injections (n ) 9). Fitted curves are 2nd order polynomials.

greatest immunoassay response is generated by the CMC: PEG:PEI composition, these responses are likely inflated due to the previously noted nonspecific interaction between the secondary antibody and biointerface (Figure 6). Antigen sensitivity of these compositions remains unchanged relative to the previously developed hydrogel,15 despite a one-third reduction in the mass available for sandwiching the antigen. In the interest of achieving maximum antigen response while minimizing nonspecific interaction with the secondary antibody, the antigen concentration range was expanded for the PEG-PEI sublayer composition by testing in the range of 0.5-25 ng/mL rhIL-1ra. At an antigen concentration of 0.5 ng/mL, the secondary antibody frequency response is slightly larger than the average response due to nonspecific interaction alone but well within the indicated range of two standard deviations (Figure 8). Above this level, at concentrations ranging from 1 to 10 ng/mL, the secondary antibody response is greater than the 95% (2δ) boundary for nonspecific interaction. However, the response within this range is relatively constant, thereby preventing discernible concentration dependence. Dissipation measurements, considerably lower in magnitude, demonstrate a similar trend of responses being greater than the nonspecific interaction but having indiscernible concentration-dependence over the same range. Over the entire range of antigen concentrations analyzed, fitted parabolic curves show a strong correlation between antigen concentration and the secondary antibody responses of both frequency and dissipation measurements. Recognizing the favorable minimal serum adhesion characteristics of the PEG-PEI sublayer composition, antigen testing in serum was conducted over a range of

12326

Langmuir, Vol. 21, No. 26, 2005

Figure 9. Substantially reduced secondary antibody immunoassay frequency (top bars, left ordinate) and dissipation (bottom bars, right ordinate) responses result when switching from samples in buffer to samples in serum (n ) 3).

25-500 ng/mL in 25% human serum. Both frequency and dissipation responses to the secondary antibody are reduced relative to measurements in buffer, with frequency responses for serum samples ranging from 38 to 43% of the respective buffer values over the 100-500 ng/ mL range (Figure 9). However, concentration-dependent responses remain evident for serum samples. In considering the nonspecific adhesion of the secondary antibody to the biointerface, concentrations less than 500 ng/mL are not discernible from blank samples when measured in serum. Optimization using higher dilution rates, in conjunction with a ligand having no affinity for the secondary antibody and the addition of surfactants during sample preparation,26 should improve the sensitivity of the biointerface for serum samples. Despite the noted sensitivity reduction in serum, a ratio of nonspecific interaction to antibody response of 9.5:1 at 25 ng/mL compares favorably with a range of 10-100:1 for similar experiments focusing on the minimization of nonspecific interaction to SAM-based hydrogel biointerfaces.10 Conclusions The present research aims to improve the immunoassay performance of cross-linked hydrogel biointerfaces by considering clinical sample requirements. Though the new (26) Brogan, K. L.; Shin, J. H.; Schoenfisch, M. H. Langmuir 2004, 20, 9729-9735.

Carrigan and Tabrizian

compositions do not demonstrate improved assay sensitivity in buffer, several notable enhancements have been achieved. The biointerface construction time has been substantially condensed by reducing the anti-hIL-1ra binding time. Immunoassay sensitivity remains unchanged despite reducing the amount of secondary antibody employed, thereby reducing the overall cost of repeat cycles using this protocol. Nonspecific adsorption of serum proteins is minimal for the complete biointerface, with less adhesion to this antibody-immobilized surface than is reported for bare carboxymethyldextran (CMD), and other hydrogel and functionalized SAM coatings.10,11 This permits concentration-dependent immunoassay responses in serum, though in this instance concentrations below 500 ng/mL could not be discriminated due to the minimal interaction between the secondary antibody and immobilizing ligand. Future optimization of serum dilutions, antibody orienting ligand, and potential use of surfactants to reduce nonspecific adsorption should improve this detection limit.26 Overall, these cross-linked hydrogel biointerfaces provide more rapid, facile, and functional immunoassay performance than protocols relying on more toxic constituents and requiring greater care and preparation time for biointerface construction. Finally, additional characterization of these biointerfaces included viscoelastic Voight modeling to estimate the thickness and mass of adsorbed material resulting from each step in the biointerface preparation. Although these initial estimates correspond with data available in the literature, an inability to corroborate the film thickness prevents significant conclusions from being drawn using these estimates. Future work enabling monitoring of the thickness of films adhered to QCM-D crystals during measurement would greatly enhance modeling capabilities, thereby extending characterization of the biointerface, and potentially immunoassay response. Acknowledgment. We thank K. L. Douglas for continued invaluable suggestions on the manuscript. We also thank P. Prince and Dr. G. Scott (MDS Pharma Services) for advice on immunoassay protocols and for providing antibodies and human serum. This work was funded by the Natural Sciences and Engineering Research Council of Canada (NSERC), Fonds Que´be´cois de Recherche sur la Nature et les Technologies (FQRNT), and Ministe`re du De´veloppement E Ä conomique et Re´gional et de la Recherche (MDERR). LA052046H