Immunosensing Based on Site-Directed Immobilization ofAntibody Fragments and Polymers that Reduce

Nonspecific Binding

Inger Vikholm-Lundin

Information Technology, Technical Research Centre of Finland, P.O. Box 14021,FIN-33101 Tampere, Finland

Received December 7, 2004. In Final Form: April 21, 2005

Antibody Fab′-fragments can be directly coupled onto gold, and the space between the fragments canbe filled with protein repellent disulfide bearing polymers. Coupling of the antibody Fab′-fragments, andthus both the amount of nonspecific binding and antigen binding but also the ability to regenerate thelayer, is dependent on the immobilization procedure. First, the immobilization has taken place by couplingthe Fab′-fragments to the surface and thereafter attaching the polymer in the remaining space betweenthe antibodies. Second, the Fab′-fragments have been added after the surface has been coated by polymer.Third, the Fab′-fragments and polymer have been added onto the surface from the same solution. Up to80% of the antigen could be removed during regeneration, if proper concentrations of polymer and Fab′-fragments were immobilized onto the gold surface. Only about 60% of the antigen could be removed, whenthe fragments were coupled directly onto a clean Au surface before the polymer or if low concentrationsof polymer were attached onto gold before the Fab′-fragments. The first immobilization method, however,showed the highest response to antigen.

1. Introduction

The use of immunoassay technology in clinical, foodsafety, and environmental analysis will continue to grow.Traditionally, radioimmunoassay, solid-phase enzymeimmunoassay, and fluoroimmunoassay have been usedto measure the very specific and tight binding betweenantibody and antigen. All approaches, however, rely ona marker molecule, such as a radioisotope, an enzyme, ora fluorescent probe, that allows quantification of theantibody-antigen complex. In most cases, the result isnot obtained until several incubations, washing, andseparation steps have been carried out. Surface plasmonresonance, SPR, on the other hand, does not require alabeled reagent for detection of the molecular interaction.SPR detects changes in refractive index and thicknessdue to the binding of molecules to a gold surface.1

In immunoassays, antibodies are mostly adsorbed onthe sensor surface or covalently coupled via functionalgroups that are not site-specific. When immobilizingantibodies via amino and carboxyl groups, the orientationof the protein molecule on the sensor surface will berandom. The lack of control over the orientation of theantibodies limits the proportion of available binding sites,whereas site-specific immobilization leads to higheractivity.2 Options for controlled immobilization of anti-bodies include specific binding of the Fc region of theantibody to a layer of protein A or protein G.3 Fab′-fragments can, moreover, be covalently attached ontosupported lipid layers through the free sulfhydryl groupopposite the antigen binding domain, and biotinylatedantibodies can be coupled onto a surface by biotin/(strept)-avidin chemistry.4-9 Assays often suffer of interference

from other molecules that nonspecifically bind to thesurface. Tween 20, bovine serum albumin (BSA), casein,fat-free milk, or serum have been used to block nonspecificbinding, NSB sites and restrict conformational changesof the immobilized antibodies.10 Yet instead of blockingthe surface, it would be preferred to use a host matrixthat reduces NSB. The phosphatidyl headgroup is knownto suppress NSB.5 Polymers such as poly(ethylene oxide),dextran, and poly(ethylene glycol) have been used toimprove the biocompatibility of surfaces and provide anattractive option for producing surfaces that preventNSB.11-13

To obtain a site-specific orientation of antibodies andprevent nonspecific adsorption, our concept has been toimmobilize Fab′-fragments directly onto a gold film andto block the remaining free space by attaching a nonionichydrophilic polymer of N-[tris(hydroxy-methyl)methyl]-acrylamide, pTHMMAA, between the antibodies.14,15 Thepolymer possesses low NSB and can be covalently attachedonto the gold surface by disulfide anchors.16 The im-mobilization procedure results in a significant reduction

* Corresponding author. Tel.: +358 3 3163363. Fax: +358 33163319. E-mail: inger.vikholm@vtt.fi.

(1) Jonsson, U.; Fagerstam, L.; Ivarsson, B.; Karlsson, R.; Lund, K.;Lofås, S.; Persson, B.; Roos, H.; Ronnberg, I.; Sjolander, S.; Stenberg,E.; Ståhlberg, C.; Malmqvist, M. BioTechniques 1991, 11, 620.

(2) Attili, B.; Suleiman, A. Microchem. J. 1996, 54, 174.(3) Turkova, J. J. Chromatogr. 1999, 722, 11.(4) Morgan, H.; Taylor, D. M. Biosens. Bioelectron. 1992, 7, 405.

(5) Vikholm, I.; Albers, W. M. Langmuir 1998, 14, 3865.(6) Fischer, B.; Heyn, S. P.; Egger, M.; Gaub, H. E. Langmuir 1993,

9, 136.(7) Schonhoff, M.; Losche, M.; Meyer, M.; Wilhelm, C. Prog. Colloid

Polym. Sci. 1992, 89, 243.(8) Krull, U. J.; Brown, R. S.; Vandenberg, E. T.; Heckl, W. M. J.

Electron Microsc. Tech. 1991, 8, 212.(9) Heyn, S. P.; Egger, M.; Gaub, H. E. J. Phys. Chem. 1991, 94,

5073.(10) Yang, Z. P.; Li, Y. B.; Balagtas, C.; Slavik, M.; Paul, D.

Electroanalysis 1998, 10, 913.(11) Akkoyun, A.; Bilitewski, U. Biosens. Bioelectron. 2002, 17, 655.(12) Pasche, S.; De Paul, S. M.; Voros, J.; Spencer, N. D.; Textor, M.

Langmuir 2003, 9, 9216.(13) De Groot, C. J.; Van Luyn, M. J. A.; Van Dijk-Wolthuis, W. N.

E.; CadeH, J. A.; Plantinga, J. e A.; Den Otter, W.; Hennink, W. E.Biomaterials 2001, 22, 1197.

(14) Vikholm, I.; Sadowski, J. Method and Biosensor for Analysis,Patent Application US2003059954.

(15) Vikholm, I. Sens. Actuators, B 2005, 106, 311.(16) Mangeney, C.; Ferrage, F.; Aujard, I.; Marchi-Artzner, V.; Jullien,

L.; Olivier Ouari, O.; Djouhar, Rekaı, D.; Laschewsky, A.; Vikholm, I.;Sadowski, J. W. J. Am. Chem. Soc. 2002, 124, 5811.

6473Langmuir 2005, 21, 6473-6477

10.1021/la046992u CCC: $30.25 © 2005 American Chemical SocietyPublished on Web 05/27/2005

of the NSB of biological macromolecules, because thesurface between the Fab′-fragments will be highly hy-drophilic, and, moreover, a proper orientation of theantibody fragment will result in a high binding of antigen.

The purpose of this study has been to optimize theimmobilization procedure by attaching the antibodies andthe polymer onto the gold surface in three different ways.First, the antibodies have been attached directly onto thegold surface and the free remaining space has been coatedwith pTHMMAA. Second, the immobilization order hasbeen the opposite. The polymer has been attached ontothe surface before the antibody fragments. Third, the Fab′-fragments and the polymer have been attached onto thesurface from the same solution. Various concentrations ofboth Fab′-fragments and polymer have been studied toobtain on optimum amount of antibodies on the sensorsurface.

2. Experimental Section

Materials.Human IgG (hIgG) and polyclonal goat anti-humanF(ab′)2 (from Jackson ImmunoResearch) were used as the modelsystem. The F(ab′)2 fragments were split into Fab′-fragmentswith dithiotreitol (DTT, Merck) in a 150 mM NaCl, 10 mMHEPES, 5 mM EDTA buffer pH 6.0, typically overnight in amicro-dialysis tube.17 Briefly, F(ab′)2 fragments at a concentrationof 0.02-0.5 µg/mL were mixed with HEPES/EDTA buffer and6.25 mM DTT solution in a micro-dialysis tube. The dialysis tubewas immersed in 250 mL of argon-purged HEPES/EDTA bufferand dialyzed overnight at room temperature under argon. TheFab′-fragments were maintained under argon and used im-mediately.

The nonionic hydrophilic polymer, bearing hydroxyl groups,were grafted onto gold by disulfide anchors. The polymer wasbased on a nonionic monomer, N-[tris(hydroxymethyl)methyl]-acrylamide, pTHMMAA (Figure 1). A disulfide-functionalizeddiazo initiator was employed to incorporate a metal-anchoringmoiety in the hydrophilic polymer. pTHMMAA was obtained byfree radical homo-polymerization as previously described andhad an average molecular weight of 6000 g/mol.16

Measurements. Glass slides coated with a thin film of goldwere cleaned in a hot solution of H2O2:NH4OH:H2O (1:1:5) andrinsed with water. The slides were attached via index matchingoil to a SPR prism on a Surface Plasmon Resonance Device(SPRDevi, VTT, Tampere, Finland). A flow cell was assembledon the prism, and the flow cell was thoroughly rinsed with 10mM HEPES, 150 mM NaCl, pH 6.8 buffer solution. The Fab′-fragments were allowed to interact with the gold-coated surfacetypically for 10 min, followed by rinsing the surface with HEPESbuffer for 5 min. 1000-1500 µL of pTHMMAA at a concentrationof 0.15 g/L was allowed to interact with the surface for 5-10 min.Various concentrations of pTHMMAA in HEPES/EDTA werealso grafted onto the gold surface followed by the addition ofvarious concentrations of anti-goat hIgG Fab′-fragments. Al-ternatively, pTHMMAA was added into the anti-goat hIgG Fab′-fragment HEPES/EDTA buffer solution, and both were attachedsimultaneously onto the gold surface. NSB of the layers wasdetermined with 0.5 g/L bovine serum albumin, BSA. Theinteraction of the layer with human IgG was studied in a buffer

of 50 mM Na2HPO4/NaH2PO4, 150 mM NaCl, pH 7.4. The layerswere regenerated with a 0.1 M glycine-HCl solution at pH )2.8.

3. Results and Discussion

Coupling of Fab′-Fragments Directly onto Au andBlocking of the Remaining Free Surface withpTHMMAA. Goat anti-hIgG Fab′-fragments can becoupled directly onto Au with a high antigen bindingefficiency.14,15 In Figure 2, a typical SPR sensor-gramshows the initial fast increase in SPR intensity at a fixedangle on attachment of Fab′-fragments onto gold. Thesurface was saturated within 10 min, and no Fab′-fragments could be rinsed off with buffer. pTHMMAAcould be attached on the surface onto which Fab′-fragments already had been attached as observed by anadditional increase in SPR intensity. Antibody Fab′-fragment and polymer immobilization was highly repro-ducible with a relative standard deviation (RSD) of 2.3%(n ) 3). The increase was, however, dependent on thepolymer and Fab′-fragment concentration.15 The polymerwas supposed to be attached between the bound Fab′-fragments, thus shielding the gold surface from NSB.There was only a minor increase in SPR intensity onadsorption of BSA (0.002 ( 0.001 au). NSB to the pureFab′-fragment film was 10-fold (0.02 ( 0.005 au). Therewas a large increase in SPR intensity on interaction of 10µg/mL hIgG with the layer. This change in SPR intensitycorresponded to 0.12 ( 0.02 au. The intensity change wasthus higher than that of a film composed of only Fab′-fragments, which showed a SPR increase of 0.10 ( 0.005au when interacting with 100 µg/mL hIgG. Thus, apartfrom suppressing NSB, the pTHMMAA molecules, more-over, help to increase the response of the layer. Thepolymer most probably protects the antibody fragmentsfrom unfolding, but might also minimize the risk of NSBto the fragment itself. Fab′-fragments that are notcovalently bound might also be replaced by polymers.The layer was regenerated with a glycine-HCl solutionduring 2 min, a washing solution and regeneration timeoften used to dissociate antigen.18 40% hIgG remained onthe surface after regeneration of the Fab′-fragment/pTHMMAA layer. The disability to regenerate part of theantibodies seems to be closely connected with theirorientation. If antibodies are randomly immobilized on

(17) Ishikawa, E. J. Immunoassay 1983, 4, 209.

(18) Kandimalla, V. B.; Neeta, N. S.; Karanth, N. G.; Thakur, M. S.;Roshini, K. R.; Rani, B. E. A.; Pasha, A.; Karanth, N. G. K. Biosens.Bioelectron. 2004, 20, 903.

Figure 1. Schematic view of the N-[tris(hydroxy-methyl)-methyl]acrylamide, pTHMMAA.

Figure 2. Change in SPR intensity on attachment of (a) 30µg/mL goat anti-human IgG Fab′-fragments and (c) 0.15 g/LpTHMMAA onto a thin gold film. Rinsing with buffer is denotedwith (b), (d) marks the interaction with BSA, (e) 10 µg/mL hIgG,and (f) regeneration with a glycine-HCl solution.

6474 Langmuir, Vol. 21, No. 14, 2005 Vikholm-Lundin

the sensor chip, the number of effective sites is reducedand the layer is more difficult to regenerate. Part of theFab′-fragments was most probably physically adsorbedonto the gold surface, and antigen could for that reasonnot be removed. Antigen could not be removed from alayer composed of F(ab)2-fragments and pTHMMAAimmobilized under the same conditions. The F(ab)2-fragments were physically adsorbed and thus randomlyoriented on the surface. Biotinylated antibodies that aresite-directly immobilized have a functionality of 60%,whereas only 5% of antibodies immobilized by passiveadsorption are functional.19 This seems to be the case alsofor the Fab′-fragment/pTHMMAA layer onto gold.

Coupling of pTHMMAA and Subsequent Couplingof Fab′-Fragments. Various concentrations ofpTHMMAA were attached onto the gold surface, andantibody Fab′-fragments at a concentration of 60 µg/mLwere hereafter immobilized onto the surface. Monolayerformation of the polymer is shown in the inset of Figure3. Saturation of the surface was obtained at about 0.3 g/LpTHMMAA. The SPR intensity increase corresponds toa monolayer thickness of about 4.4 nm as earlier simulatedby fitting SPR data to a Fresnel equation.16 The size of aFab′-fragment is 7 × 5 × 4 nm3.20 If standing end-on onthe surface attached through the thiol group, the Fab′-fragments would protrude out from the polymer hostmatrix, which would enable antigen binding. The amountof Fab′-fragments that attached onto an Au film alreadycoated with pTHMMAA decreased with pTHMMAAconcentration as could be expected (Figure 3). If the Ausurface was coated with pTHMMAA > 0.015 g/L, no Fab′-fragments could be attached. At lower pTHMMAA con-centrations, Fab′-fragments attach onto Au probablybetween the pTHMMAA molecules. Yet as pTHMMAAmolecules start to be more closely packed, attachment ofFab′-fragments was hindered. No binding occurred atpTHMMAA concentrations where the polymer starts toform a monolayer. It is known that pTHMMAA givespolymers that are nonionic, very hydrophilic, biocompat-ible, and, moreover, inert to biological fouling.15 Therewas no NSB of BSA to the pTHMMAA monolayer. Thetotal change in SPR intensity of layers composed of 60

µg/mL Fab′-fragments below a polymer concentration of0.01 g/L pTHMMAA corresponded to 0.09 ( 0.01 au. Yetthe SPR intensity increased to 0.12 ( 0.005 au at 0.3 g/LpTHMMAA.

Figure 3 shows that the interaction of hIgG with themonolayer decreased above pTHMMAA concentrationsof 0.001 g/L. NSB of BSA decreased with pTHMMAAconcentration to zero at concentrations above 0.003 g/LpTHMMAA. Human IgG corresponding to about 0.045 (0.002 au could be bound at pTHMMAA concentrations<0.001 g/L. At a pTHMMAA concentration of 0.01 g/L, avery low amount of antibody fragments were bound andsubsequently hIgG binding was also low. The amount ofhIgG that could be removed from the layer by theregeneration solution decreased with pTHMMAA con-centration (Figure 4, inset). The Fab′-fragments mostprobably need more space for coupling through the thiolbond, and the polymer might be partly coiled on thesurface. Still, a high degree of the Fab′-fragments seemedto be site-directly coupled onto the surface at a lowpTHMMAA concentration, as about 60% of the antigencould be removed from the surface with the regenerationsolution. The polymer might protect antibodies fromunfolding and prevent conformational changes on thesurface at low polymer concentrations. At low polymerconcentration, the NSB was, however, low.

Coupling of pTHMMAA and Fab′-Fragments fromthe Same Solution. The SPR intensity increased linearlywith Fab′-fragment concentration when pTHMMAA atconcentrations of 0.012 and 0.025 g/L were immobilizedon the gold surface from the same solution (Figure 5). Theincrease was also dependent on polymer concentration(Figure 6). Fab′-fragments and polymers are expected tocompete for attachment onto the surface. This competitionshould depend on the size and concentration of themolecules. The size of the Fab′-fragment is about 8 timesthat of the polymer, and the concentration of the polymerwas in the same order as that of the fragments. Fab′-fragments are expected to attach on the surface throughthe thiol group, whereas the polymer attaches throughdisulfide groups. Thiols and disulfides form stable self-assembled monolayers (SAMs) on gold via a polar covalentbond. With thiols, the reaction is assumed to take placeas an oxidative addition to gold with release of hydrogen,whereas in the case of disulfides, a cleavage of the S-Sbond occurs. Disulfides, however, adsorb approximately

(19) Davies, J.; Dawkes, A. C.; Haymes, A. G.; Roberts, C. J.;Sunderland, R. F.; Wilkins, M. J.; Davies, M. C.; Tendler, S. J. B.;Jackson, D. E.; Edwards, J. C. J. Immunol. Methods 1994, 167, 263.

(20) Sarma, V. R.; Silverton, E. W.; Davies, D. R.; Terry, W. D. J.Biol. Chem. 1971, 246, 3753.

Figure 3. Change in SPR intensity on the attachment of 60µg/mL Fab′-fragments to a surface coated with various amountsof polymer. The inset shows the change in SPR intensity on theattachment of various concentrations of the polymer pTHMMAAonto Au.

Figure 4. Change in SPR intensity on the attachment of Fab′-fragments (9) onto a gold surface coated with various concen-trations of pTHMMAA. NSB of 0.5 g/L BSA (3) and interactionwith 100 µg/mL hIgG (O). The inset shows the amount of hIgGremoved from the pTHMMAA/Fab′-fragment layers afterregeneration.

Immunosensing Based on Antibody Fragments Langmuir, Vol. 21, No. 14, 2005 6475

40% slower than thiols on gold.21 Above a Fab′-fragmentconcentration of 40 µg/mL, the SPR increase correspondedto 0.14 ( 0.02 au (n ) 9) (Figure 5). When the polymerwas attached onto the gold surface before the fragment,the total increase was only 0.09 ( 0.01 au (as shown inthe previous section). It therefore seems like the Fab′-fragments predominantly were attached on the surfacebefore the polymers. If a large amount of polymers wouldbe attached before the fragments, these would hinder Fab′-fragments from reaching the surface and even a lowamount of polymers on the surface could be expected tocause a response similar to that of polymers attached onthe surface before the fragments as in the previous section.

NSB of BSA was dependent on the polymer concentra-tion (Table 1). At a very low polymer concentration (0.001g/L), a low NSB of BSA could be observed. At 0.006 g/LpTHMMAA, BSA caused a small increase in SPR responsewhen the Fab′-fragment concentration was below 5 µg/mL or above 70 µg/mL, but between this concentrationrange there was no NSB of BSA. As the pTHMMAAconcentration further increased, it seems like materialwas removed from the surface. At a polymer concentrationof 0.025 g/L, the decrease in response corresponded to 4%of the Fab′/pTHMMAA layer. The decrease in SPR

intensity could only be due to a detachment of notcovalently bond fragments or polymers from the surface.An exchange with BSA cannot be excluded. Thus, atoptimum Fab′-fragment concentration, it seems like NSBto the mixed layer was reduced to -0.004 ( 0.005 au atpolymer concentrations in the range of 0.006-0.012 g/L.The NSB binding was remarkably reduced by the polymeras compared to a layer composed of only Fab′-fragmentsthat have a NSB of 0.02 ( 0.005 au, which correspondedto 25% of the specific binding.

The response of the mixed Fab′/pTHMMAA monolayerto hIgG was almost independent of Fab′-fragment andpolymer concentration in the concentration range studied(Figure 6). The response of hIgG to the layers was 0.08( 0.01 au. The response was lower than that of a layerwhere the antibodies were attached onto the surface beforethe polymer. A hIgG concentration of 10 µg/mL showeda response of 0.12 ( 0.02 au, but higher than that of alayer where the polymer was attached before the antibodyfragments (0.045 ( 0.002 au).

Only 34% of the antigen could be removed with aglycine-HCl regeneration solution when the layer wasprepared without pTHMMAA from a 5 µg/mL Fab′-fragment solution (Figure 7). When the layer was madefrom higher fragment concentrations, 62 ( 1% of theantigen could be removed, indicating that a higher amountof the fragments were site-directly orientated. A higheramount of fragments seemed to be adsorbed onto the goldsurface,andat lowFab′-fragmentconcentrationantibodiesmight be at least partly unfolded. There was a remarkableincrease in the ability to regenerate the layer whenpolymer was added into the immobilization solution(Figure 7). Even very low pTHMMAA concentrationsimproved the ability to regenerate the layer. Yet the abilityto regenerate the layer was also dependent on the amountof Fab′-fragments in the layer. When 5 µg/mL Fab′-fragments were present in the layer, 65 ( 6% of the antigen

(21) Jung, Ch.; Dannenberger, O.; Yue, X.; Buck, M.; Grunze, M.Langmuir 1998, 14, 1103.

Figure 5. Change in SPR intensity on the coupling ofpTHMMAA and various concentrations of goat anti-hIgG Fab′-fragments onto gold; n ) 3.

Figure 6. Binding of 100 µg/mL human IgG (O) to a mixedmonolayer of Fab′-fragments and pTHMMAA (9).

Table 1. Nonspecific Binding of BSA to LayersComposed of PTHMMAA and Fab′-Fragments in a

Concentration Range between 5 and 75 µg/mL (n ) 6)

pTHMMAA [g/L] BSA [au]

0.001 0.003 ( 0.0020.006 -0.004 ( 0.004

0.002 ( 0.005a

0.012 -0.003 ( 0.0060.025 -0.006 ( 0.003

a Fab′-fragment concentration e5 and g70 µg/mL.

Figure 7. Amount of hIgG removed during regeneration ofmixed monolayers of 5 µg/mL (O), >20 µg/mL (b) Fab′-fragmentsand pTHMMAA onto gold.

6476 Langmuir, Vol. 21, No. 14, 2005 Vikholm-Lundin

could be removed at higher polymers concentrations(Figure 7). 81 ( 1% of the layer could be regenerated whenthe layer was prepared from low amounts of polymer and20-40 µg/mL Fab′-fragments. Yet as the pTHMMAAconcentration increased, the regeneration efficiency de-creased. Above a pTHMMAA concentration of 0.003 g/L,73 ( 2% of the antigen could be removed from the layer.Too high concentrations of pTHMMAA might hinder Fab′-fragments from coupling to the surface through the thiolbond, thereby increasing the amount of fragments ad-sorbed onto the surface.

4. ConclusionsA simple and fast method for immobilizing antibodies

onto a gold sensor surface has been optimized for humanIgG antibody fragments. The method is generic and canbe used for coupling of any antibody in an oriented mannerto the sensor surface. Fab′-fragments are coupled directlyonto gold, and the remaining space between the fragmentsis blocked with nonionic, hydrophilic disulfide bearingpolymers, pTHMMAA, to suppress nonspecific binding.When the antibody Fab′-fragments were applied onto the

gold surface and the remaining free space between theantibodies was coated with polymer, a large part of thefragments seemed to be site-directly attached throughthe free thiol bonds. Part of the fragments was randomlyadsorbed onto the surface, and 60% of the antigen couldbe removed during regeneration of the layer. If pTHMMAAat very low concentrations was first applied on the surface,60% of the antigen could also be removed. At higherpTHMMAA concentration, the polymer molecules, on theother hand, prevent fragments from attaching onto thesurface. When Fab′-fragments and pTHMMAA wereimmobilized on the surface from the same solution, up to80% of the antigen could be removed, indicating a highdegree of site-directed immobilization of the antibodyfragments. The highest response to hIgG was, however,obtained when the antibody fragments were attached ontothe surface before the polymer. When the polymer wasattached on the surface before the antibody, the responsewas one-half of that when the antibodies and the polymerwere attached from the same solution.

LA046992U

Immunosensing Based on Antibody Fragments Langmuir, Vol. 21, No. 14, 2005 6477