Oriented Surface Immobilization of Antibodies at the ConservedNucleotide Binding Site for Enhanced Antigen DetectionNathan J. Alves,† Tanyel Kiziltepe,†,‡ and Basar Bilgicer*,†,‡,§

†Department of Chemical and Biomolecular Engineering, ‡Advanced Diagnostics and Therapeutics, and §Department of Chemistryand Biochemistry, University of Notre Dame, Notre Dame, Indiana 46556, United States

*S Supporting Information

ABSTRACT: The conserved nucleotide binding site (NBS),found on the Fab variable domain of all antibody isotypes,remains a not-so-widely known and unutilized site. Here, wedescribe a UV photo-cross-linking method (UV-NBS) thatutilizes the NBS for oriented immobilization of antibodiesonto surfaces, such that the antigen binding activity remainsunaffected. Indole-3-butyric acid (IBA) has an affinity for theNBS with a Kd ranging from 1 to 8 μM for different antibodyisotypes and can be covalently photo-cross-linked to theantibody at the NBS upon exposure to UV light. Using theUV-NBS method, antibody was successfully immobilized on synthetic surfaces displaying IBA via UV photo-cross-linking at theNBS. An optimal UV exposure of 2 J/cm2 yielded significant antibody immobilization on the surface with maximal relativeantibody activity per immobilized antibody without any detectable damage to antigen binding activity. Comparison of the UV-NBS method with two other commonly used methods, ε-NH3

+ conjugation and physical adsorption, demonstrated that the UV-NBS method yields surfaces with significantly enhanced antigen detection efficiency, higher relative antibody activity, andimproved antigen detection sensitivity. Taken together, the UV-NBS method provides a practical, site-specific surfaceimmobilization method, with significant implications in the development of a large array of platforms with diverse sensor anddiagnostic applications.

■ INTRODUCTIONThe conserved nucleotide binding site (NBS), which is foundin the variable region of the Fab arm of all antibody isotypes,remains a not-so-widely known and unutilized site. This paperdescribes a method for UV photo-cross-linking antibodies ontosurfaces at the NBS, in an oriented manner such that theantigen binding activity is preserved. Antibodies are conjugatedto surfaces in a large array of platforms including sensor anddiagnostic applications developed for the detection ofpathogens, disease biomarkers, water contaminants, drugdiscovery, and laboratory-based immunoassays.1−4 For all ofthese applications, antibody’s antigen binding activity is acritical parameter that governs the sensitivity, dynamic range,and reproducibility of the detection tools.The current standard method of immobilizing antibodies to

surfaces involves noncovalent physical adsorption to a detectionsurface through nonspecific hydrophobic interactions (physicaladsorption method).5−12 This method, however, results inrandomly oriented antibody molecules on the detection surfaceand yields ∼90% antibody that is in an inactive orientation dueto steric blocking of the antigen binding sites.13−15 A commonalternate method involves nonspecific chemical immobilizationto amine reactive surfaces by utilizing the lysine side chainamino groups present on the surface of antibodies (ε-NH3

+

method).16−23 In this method, control over sites of conjugationis not possible, which results in an inhomogeneous surface with

reduced antibody activity similar to that of the random physicaladsorption method.24,25 Despite resulting in a significant loss ofantibody activity, these methods are still commonly used due totheir simplicity of execution. Site-specific immobilizationmethods are also currently available; however, these methodshave either complicated chemical procedures or are detrimentalto the antibody activity. For example, one method involvesselective reduction of the disulfide bonds between the twoheavy chains and uses the available thiol side chains as reactivesites for conjugation to gold or maleimide functionalizedsurfaces.26−30 This method, albeit site specific, yieldsmonovalent capture antibodies and inactive antibody fragmentsdue to unintentional reduction of the disulfide bonds thatmaintain the structure of the antigen binding frameworkregions. An alternate method utilizes the carbohydrate chainspresent on antibodies originating from post-translationalprotein modifications. These side chains can be oxidized byvarious chemical approaches to form reactive aldehyde groupsthat can be used to selectively couple to hydrazine function-alized surfaces.31−35 This technique provides a high couplingyield, but the denaturing conditions and oxidative chemicalsused can result in a loss of antibody activity.36 In addition,vastly different degrees of post-translational modifications on

Received: January 25, 2012Published: May 21, 2012

Article

pubs.acs.org/Langmuir

© 2012 American Chemical Society 9640 dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−9648

antibodies cause this approach to deliver highly variableoutcomes from antibody to antibody.37,38 The primary concernwith many of the currently available antibody immobilizationmethods is the loss of antigen recognition due to sterichindrance of the antigen binding sites or partial denaturation ofthe antibody due to chemically harsh reaction conditions.13,24,25

These outcomes result in a loss of antibody activity that directlyequates to a loss in antigen capture efficiency of thefunctionalized surface. Taken together, these points highlightthe need for the development of a practical and reproduciblemethod for site specific immobilization of antibodies to surfacessuch that the antigen binding activity remains unaffected.Here we describe a photochemistry-based NBS specific

antibody immobilization (UV-NBS) method for orientedimmobilization of antibodies onto surfaces. In an earlierpublication, we extensively characterized the NBS usingmolecular modeling and showed that it is a highly conservedbinding pocket located in the “conserved” region of the variabledomain of all antibody isotypes from various species (Figure1A).39,40 This characterization was achieved by performing aleast-squares root-mean-square deviation superposition of allFab domain crystal structures of >260 immunoglobulins in theRCSB Protein Data Bank.39 Specifically, our analysis revealedthat four residues, namely two tyrosine residues on the variableregion of the light chain [framework region 2 (FR2), position42, and FR3, position 103 based on IMGT numbering]41 andone tyrosine and one tryptophan residue on the variable regionof the heavy chain [FR3, position 103, and junction region,position 118, respectively], are conserved. In some instances aphenylalanine is observed in place of the tyrosine at either ofthe conserved positions on the VL (Figure S1 in SupportingInformation). Hence, the NBS is rich with amino acid residues

with aromatic side chains. When aromatic rings are exposed toa specific wavelength (254 nm) of UV light and are in closeproximity, reactive radicals form, resulting in the formation ofnew covalent bonds between the aromatic rings.42,43 Therefore,the NBS provides a useful site for selective conjugation ofantibodies to small molecule ligands that contain aromaticrings. To identify such small molecule ligands with a highbinding affinity and selectivity for the NBS, we performed an insilico screening by docking various small molecules from theZINC database at the NBS.39 The top scoring molecules(2-(2-benzimidazolylamino)-1-ethanol, 7-methyltryptamine, sine-fugin, indole-3-butyric acid, tryptophan, and 5-methylindole-3-carboxaldehyde)were then experimentally investigated fortheir binding affinity to the NBS. Indole-3-butyric acid (IBA)emerged as the highest affinity binding nucleotide analogue,with Kd values ranging between 1 and 8 μM depending on theantibody, and was therefore chosen for the applicationdescribed in this study.The UV-NBS method necessitates IBA conjugated surfaces

for antibody cross-linking. When antibody is introduced to thewell, surface conjugated IBA associates with the antibody viabinding to its NBS. Upon UV exposure, a covalent bond formsbetween the IBA and the antibody, permanently tethering theantibody to the surface and thereby generating the antibodyfunctionalized surface (Figure 1B). The described method,being site-specific, enables immobilization in an orientedmanner such that antigen binding activity of the antibody ispreserved for enhanced antigen detection efficiency of thefunctionalized surface.

Figure 1. (A) Location of the nucleotide binding site (NBS) in antibodies is shown in a cartoon representation and on the crystal structure of theantibody Fv region. (B) This is a schematic representation of the method for UV photo-cross-linking of antibodies onto surfaces at the NBS. IBAfunctionalized surfaces are generated by reacting IBA-EG11-amine with maleic anhydride-coated plates. Antibodies associate with IBA moiety at theNBS, and upon UV exposure a covalent bond forms between IBA and the antibody, permanently tethering the antibody to the surface. The oriented,site-specific conjugation of the antibody through its NBS preserves antibody’s antigen binding activity.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489641

■ EXPERIMENTAL DETAILSSynthesis of IBA-EG11-amine Conjugates and Generation of

IBA-Coated Surfaces. The IBA-EG11-amine ligand was synthesizedby coupling indole-3-butyric acid (IBA, Sigma) to mono-N-t-boc-amido-dPEG11-amine (Quanta Biodesign) following activation using2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium hexafluorophos-phate (HBTU, Novabiochem) in dimethylformamide (DMF, Sigma)and N,N-diisopropylethylamine (DIEA, Sigma) at room temperaturefor 3.5 h while agitating. After rotate evaporating DMF, the t-bocprotecting group was removed in a solution of 4% triisopropylsilane(Sigma), 4% DI water (Millipore Integral 10 Milli-Q system), and 92%trifluoroacetic acid (TFA, Sigma) for 45 min at room temperature.IBA-EG11-amine was purified via reverse phase high pressure liquidchromatography (RP-HPLC) on a Zorbax C18 (Agilent) column, andits mass was verified via matrix-assisted laser desorption ionizationtime-of-flight mass spectrometry (MALDI-TOF MS) using 2,5-dihydroxybenzoic acid (DHB, Sigma) as matrix (Figure S2). Thepurity was confirmed using RP-HPLC on an analytical Zorbax C18column (>95%), and the yield was 75%. IBA functionalized platesurface was generated by coupling IBA-EG11-amine to amine-reactive,maleic anhydride polymer-coated 96-well plates (Thermo Scientific).Coupling was achieved by incubating 100 μL of 0.625 μM IBA-EG11-amine in each well for 2 h, in PBS at pH 8.0, at room temperature. Anyremaining reactive maleic anhydride sites were then blocked byincubating the plate surface with 50 mM Tris buffer with 100 mMNaCl at pH 8.0 for 1 h.Synthesis of IBA-Biotin. IBA-biotin was synthesized using

standard solid phase synthesis protocols on a NovaPEG Rink Amideresin (Novabiochem) and Fmoc chemistry. First, HBTU activatedFmoc-Trp(Boc)-OH in DMF, and DIEA was coupled to the resin for3 h while agitating. Fmoc was deprotected using 20% piperidine inDMF. Fmoc-N-amido-dPEG2-acid (Quanta Biodesign) was thenactivated using HBTU and coupled in the next step. After Fmocdeprotection, HBTU activated Biotin (Sigma) was coupled to the resinconjugated molecule. Kaiser tests were performed between couplingsteps to monitor synthesis progress. IBA-biotin was cleaved from theresin in 95% TFA, 2.5% TIS, and 2.5% DI water, purified via RP-HPLC on a Zorbax C18 column, and characterized using MALDI-TOF MS (Figure S3). The purity was confirmed using RP-HPLC onan analytical Zorbax C18 column (>95%), and the yield was 82%.Synthesis of DNP-Biotin. DNP-biotin was synthesized using

standard solid phase synthesis protocols on a NovaPEG Rink Amideresin and Fmoc chemistry. Fmoc-Lys(ivDde)-OH (Novabiochem) wasactivated with HBTU in DMF and DIEA, followed by coupling to theresin using agitation for 3 h. The Fmoc protecting group was removedwith 20% piperidine in DMF. Boc-Lys(Fmoc)-OH (Novabiochem)was then activated and coupled to the deprotected amine followed byFmoc deprotection and coupling of biotin. The ivDde protectinggroup on lys was then removed using 2% hydrazine in DMF followedby activation and coupling of Fmoc-N-amido-dPEG2-acid (QuantaBiodesign). The ethylene glycol linker was then deprotected andincubated with 2,4-dinitro-1-fluorobenzene (Aldrich) in DMF withDIEA. Kaiser tests were performed between coupling steps to monitorsynthesis progress. Product was cleaved from the resin in 95% TFA,2.5% TIS, and 2.5% DI water and purified via RP-HPLC on a ZorbaxC18 column. The product was characterized using MALDI-TOF MS(Figure S4). The purity was confirmed using RP-HPLC on ananalytical Zorbax C18 column (>95%), and the yield was 40%.Western Blot Analysis of UV-NBS Photo-Cross-Linking of

IBA to the Antibody. IgG antibody (1.7 μM) was incubated withexcess IBA-biotin (100 μM) in PBS buffer at pH 7.4 and then exposedto increasing UV (254 nm) energy in a Spectroline UV Select SeriesCross-linker (Spectronics). The samples were run on a 10% SDS-PAGE gel with a Tris-glycine running buffer under reducingconditions at 110 V for 1 h and were transferred to a nitrocellulosemembrane at 110 V for 90 min in a 10% MeOH transfer buffer(Boston Bioproducts). The membrane was blocked with 10% dry milkin Tris buffered saline (TBS) for 1 h and was then blotted with 1:10000 dilution of Streptavidin-HRP (Jackson ImmunoResearch) 1 h atroom temperature (RT). A chemiluminescent HRP substrate

(Invitrogen) was used to detect the location where IBA-biotin wascovalently conjugated on the antibody. To verify transfer of all proteincontent to the membrane, both the SDS-PAGE gel (post transfer) andnitrocellulose membrane were Coomassie blue stained in a solution of10% acetic acid, 20% methanol, 0.15% Coomassie R-250 (EMDChemicals) for 30 min and destained in a solution of 20% acetic acid,20% methanol, 60% DI water for 1.5 h. Control experiment performedin the absence of UV exposure or in the absence of IBA-biotin did notyield any detectable bands. Similarly, control experiments performedwith only biotin did not yield any detectable bands. Mouse anti-FITCIgG1 (clone: DE3) was purchased from Millipore, rat anti-DNP IgG1

(clone: LO-DNP-2) was purchased from Invitrogen, and mouse anti-Streptavidin IgG1 (clone: S8C12) was purchased from Meridian LifeScience, Inc.

UV-NBS Antibody Immobilization Method. IBA-coated 96-wellplates were generated as described above and were blocked with 50mM Tris, 100 mM NaCl, pH 8.0 buffer for 1 h. Antibody was added toIBA-coated wells in a total volume of 100 μL PBS buffer with 0.1%Tween20 at pH 7.4 for 1.5 h at RT and was then exposed to theindicated amount of UV light using a Spectroline Select Series UVcross-linker (Figure S5). Unbound antibody was then washed using anautomated plate washer (MDS Aquamax 2000), three cycles of 200 μLPBS with 0.05% Tween20 at pH 7.4. Antibody immobilized wells wereblocked with BSA blocking buffer (2.5 g BSA in 50 mL of PBS bufferwith 0.05% Tween20 at pH 7.4) for 1 h to prevent nonspecificadhesion/interactions.

ε-NH3+ Antibody Immobilization Method. Antibody was

incubated in an amine reactive maleic anhydride 96-well plate for 2h at RT in PBS buffer at pH 8.0. Unbound antibody was washed usingan automated plate washer (three cycles of 200 μL PBS with 0.05%Tween20 at pH 7.4). Any remaining reactive sites on the plate surfacewere then quenched using 50 mM Tris buffer with 100 mM NaCl atpH 8.0 for 1 h. The surface was then blocked using BSA blockingbuffer for 1 h.

Physical Adsorption Method. Antibody was adsorbed to a highbinding 96-well plate (Costar) in 0.05 M carbonate−bicarbonatecoating buffer at pH 9.6 for 2 h at RT (Figure S6). Unbound antibodywas washed using an automated plate washer (three cycles of 200 μLPBS with 0.05% Tween20 at pH 7.4). Plate surface was then blockedwith BSA blocking buffer for 1 h.

Determination of Antigen Detection Efficiency of AntibodyImmobilized Surfaces. The antigen detection efficiency of theantibody immobilized surfaces generated using the above-describedmethods were determined by an enzymatic assay, where biotinconjugated versions of antigens (e.g., DNP-biotin, FITC-biotin) weresynthesized to enable detection using streptavidin-HRP. Briefly, IgGimmobilized surfaces were incubated with saturating concentrations ofantigen−biotin in a total volume of 100 μL PBS with 0.05% Tween20at pH 7.4 for 1.5 h. After washing unbound antigen−biotin, the wellswere incubated with a 1:10 000 dilution of streptavidin−HRP (1.0mg/mL stock) in BSA blocking buffer for 1 h. HRP substrate (AmplexRed, Invitrogen) was added, and fluorescent product formation wasobserved on a Molecular Devices SpectraMax M5 plate reader (ex 570nm, em 592 nm) (Figure S7). The results are reported as relativefluorescence units (RFU). Control experiments performed withoutantigen−biotin, and immobilized antibody were used as backgroundfor the antigen detection measurements.

Determination of Total Antibody Content of AntibodyImmobilized Surfaces. Quantification of the total surface immobi-lized antibody for each of the three immobilization methods wasperformed using HRP conjugated Fc domain specific secondaryantibodies from goat purchased from Jackson ImmunoResearch(Figure S8). Briefly, wells were incubated with a 1:5000 dilution ofanti-Fc antibody (1.0 mg/mL stock) in BSA blocking buffer for 1 h.After washing unbound anti-Fc antibody, HRP substrate (Amplex Red,Invitrogen) was added, and fluorescent product formation wasobserved (ex 570 nm, em 592 nm). The results are reported asRFU. Control experiments performed without immobilized antibodywere used as background.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489642

Determination of Antigen Detection Sensitivity. A doseresponse curve was created using each of the immobilizationtechniques by varying the antigen concentration while keeping thetotal surface immobilized antibody constant. The resulting plots(antigen detection signal vs antigen concentration) were fit by linearregression. Sensitivity was determined from the slope of the linearregression.

■ RESULTS AND DISCUSSION

The UV-NBS method for surface immobilization of antibodiesrequires an IBA functionalized surface (Figure 1). Therefore,we synthesized an IBA derivative, IBA-EG11-amine, andcoupled it to an amine reactive maleic anhydride polymercoated 96-well plate to yield IBA functionalized surfaces (foroptimization see Figure S9). To generate antibody immobilizedsurfaces, IgGDNP antibody (0.1 pmol) was incubated in thewells for 1.5 h to allow association of NBS and surface-conjugated IBA. The plate was then exposed to UV light forcovalent bond formation between IBA and the NBS residues.The site-specificity of the UV-NBS method leaves the antigenbinding site unaffected and the Fc domain available forsecondary antibody binding. Quantification of the total surfaceimmobilized antibody, using an HRP linked anti-Fc secondaryantibody, demonstrated that the UV-NBS method successfullygenerated antibody immobilized surfaces in a UV energydependent manner, reaching up to ∼4.5 × 103 RFU (Figure2A). Next, we evaluated the antigen detection efficiency of theantibody immobilized surface with an ELISA assay. Briefly,IgGDNP immobilized surfaces were incubated with saturatinglevels of DNP−biotin conjugate, and captured DNP-biotin wasquantified using streptavidin−HRP as a reporter. Our resultsdemonstrated that the antibody immobilized surfaces, gen-erated via the UV-NBS method, were effective in antigendetection. Maximum antigen detection efficiency was achievedwith surfaces produced using a UV exposure between 2 and 5J/cm2, reaching up to ∼21 × 103 RFU (Figure 2B).Importantly, the relative antibody activity, which is the ratioof the antigen detection efficiency signal to the total surfaceimmobilized antibody signal, reached a maximum at 2 J/cm2

(Figure 2C). The results of these experiments established theoptimal UV energy for the UV-NBS method as 2 J/cm2, wherewe observed (i) a high yield of antibody photo-cross-linking,(ii) the maximum antigen detection efficiency of the surface,and (iii) the maximum relative antibody activity. We alsoachieved similar results with an alternate antibody−antigensystem (mouse anti-FITC; IgGFITC) (results not shown). No

antibody cross-linking was observed in control experimentsperformed using plates that were not coated with IBA or in theabsence of UV exposure (results not shown). Furthermore,antibody cross-linking was dramatically inhibited in controlexperiments performed in the presence of competitivelybinding soluble IBA (results not shown).It is noteworthy that antigen detection efficiency reached a

plateau at 2 J/cm2 despite an increase in total surfaceimmobilized antibody (Figure 2A,B). This plateau effectcould be a reflection of a shortcoming of the HRP-basedenzymatic assay at the higher antigen concentrations, since therate of the enzymatic reaction is limited by substrate diffusionfrom the bulk solution to the surface. Alternatively, it couldresult from increased damage to the antigen binding site whencompared to the Fc region beyond 2 J/cm2. To confirm that nosignificant damage to the antibody’s antigen binding activitywas induced at the UV doses used, we performed additionalexperiments to test the antibody’s structural integrity post-UVexposure. For this, IgG antibody was cross-linked to a platesurface using the common ε-NH3

+ immobilization method andexposed to increasing amounts of UV energy (0−10 J/cm2).The damage to the antigen binding site was determined byincubating IgG immobilized surfaces with saturating levels ofantigen−biotin conjugates, which was quantified by streptavi-din−HRP in an ELISA assay. To evaluate the structuralintegrity of the Fc domain, a secondary anti-Fc antibody wasused. The anti-Fc antibody only recognizes the intact Fcstructure; therefore, loss in its ability to bind to UV exposedIgG correlates directly to Fc damage. Our results suggested thata UV dose of up to 2 J/cm2 did not have detectable impact onthe antibody’s integrity based on (i) its ability to recognize itsantigen and (ii) the secondary antibody’s ability to bind to theFc domain (Figure S10). We further confirmed this result bychallenging the antibody’s antigen binding activity by exposingit to increasing UV energies in solution and observing its abilityto recognize its surface immobilized antigen in an ELISA assay.With this method, we did not detect any significant damage toeither the antigen binding activity or the Fc recognition up to 5J/cm2 (Figure S11). Finally, in a separate experiment weevaluated if UV exposure caused any higher order aggregates orcross-linked products between antibody molecules by dynamiclight scattering and did not detect any structural changes afterUV exposure at 2 J/cm2 (Figure S12). Combined, these resultssuggest 2 J/cm2 as a well-tolerated UV dose that preservesstructural integrity and antigen binding activity of the antibody.

Figure 2. UV-NBS method for surface immobilization of IgGDNP antibody to IBA-functionalized surface using increasing UV energy. (A) Amount ofsurface immobilized antibody was measured using an HRP linked anti-Fc secondary antibody. (B) Antigen detection efficiency of the surface wasmeasured with DNP−biotin as the antigen and streptavidin−HRP as the reporter. In both experiments, y-axis is reported as the relative fluorescenceunits (RFU). (C) Relative antibody activity (the ratio of the signals from antigen detection efficiency to the total surface immobilized antibody) isplotted. All data represents means (±SD) of triplicate experiments.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489643

Furthermore, the Fc domain is also preserved, making itpossible for quantification of the total surface immobilizedantibody. Therefore, UV exposure of 2 J/cm2 was used inproducing antibody immobilized surfaces for antigen detectionin the rest of the study.Next, we investigated if the UV-NBS method yielded IBA

cross-linking selectively at the NBS by analyzing the products ofa UV-cross-linking reaction using Western Blot analysis. Briefly,we synthesized an IBA−biotin conjugate for easy detection bystreptavidin. IBA−biotin was incubated with IgG antibody inPBS buffer to allow binding and was then photo-cross-linked tothe NBS by exposure to a range of UV energies (0−2 J/cm2).The product, antibody conjugated IBA−biotin, was run on anSDS-PAGE under reducing conditions and transferred to anitrocellulose membrane, and biotinylated fragments wereprobed by using a streptavidin−HRP reporter molecule.While the NBS is located between the light and heavy chainsour results demonstrated a preferential covalent insertion of theIBA−biotin molecule selectively to the antibody light chain in aUV energy dependent manner as indicated by an increase inband intensity with increasing amounts of UV exposure (Figure3). This is likely due to the orientation of the IBA when

associated with the NBS prior to UV exposure. Controlexperiments performed with just biotin did not yield any bandsin the blotted membrane (results not shown). Although thisresult does not necessarily ensure that IBA cross-linking takesplace precisely at the NBS, it does confirm insertion of the IBAstrictly on the antibody light chains, indicating that the photo-cross-linking is not a random event and takes place at a specificsite. This result was confirmed by repeating this assay usingvarious antibodies from different species with different antigenspecificities (results not shown).We further evaluated the UV-NBS method by comparing it

to two other commonly employed antibody immobilizationmethods: the ε-NH3

+ and the physical adsorption methods. Forε-NH3

+ immobilization, antibody was allowed to react directlywith an amine reactive maleic anhydride functionalized plate.The ε-NH3

+ method conjugates the antibody to the platesurface arbitrarily through multiple lysine side chains, resultingin randomly oriented antibody molecules on the surface.Immobilization via the physical adsorption method is primarily

driven by nonspecific hydrophobic interactions between theantibody and plate surface. Therefore, for physical adsorption,the antibody was incubated on a high-binding ELISA platesurface. Similar to the ε-NH3

+ method, the physical adsorptionmethod also yields randomly oriented antibodies on thesurface. Both of these commonly used methods were comparedto the UV-NBS method for (i) antigen detection efficiency ofthe functionalized surface, (ii) total antibody immobilized tothe surface, and (iii) relative antibody activity.For this comparison, we functionalized 96-well plates using

the three respective methods with initial antibody amounts of0.01, 0.05, or 0.1 pmol of IgGDNP to generate the antibody-coated surfaces. Antigen detection efficiency of each surface wasdetermined via the aforementioned method by using saturatinglevels of DNP−biotin followed by HRP−streptavidin. Totalimmobilized antibody content in the wells was determined byan HRP conjugated Fc specific secondary antibody. Whencomparing the three methods, UV-NBS immobilization methodyielded functionalized surfaces with significantly higher antigendetection efficiency than both of the other methods (Figure4A). Specifically, the enhancement reached up to ∼10-foldwhen compared to the ε-NH3

+ method and up to ∼122-fold

Figure 3. UV-NBS photo-cross-linking site on the antibody wasinvestigated using Western Blot analysis. IBA−biotin was cross-linkedto the antibody (IgGDNP) by exposure to UV energy from 0 to 2 J/cm2

in PBS buffer. SDS-PAGE was run under reducing conditions, and theproteins were transferred to a nitrocellulose membrane. Both the gel(SDS−gel post-transfer) and membrane (nitrocellulose membrane)were stained by coomassie blue to verify efficiency of transfer.Streptavidin−HRP was used to probe for covalently attached IBA−biotin to the antibody. Blotted film shows that biotin tag only appearson the antibody light chain. Similar results obtained using IgGFITC

(data not shown).

Figure 4. Evaluation of the UV-NBS method in comparison to ε-NH3+

and physical adsorption methods using the rat IgGDNP/DNPantibody/antigen system. The 96-well plates were functionalizedwith IgGDNP using all three methods. (A) Antigen detection efficiencyof all three surfaces was detected with DNP−biotin as the antigen andstreptavidin−HRP as the reporter. (B) Total surface immobilizedantibody was quantified using an HRP linked anti-Fc secondaryantibody. In both experiments, y-axis is reported as relativefluorescence units (RFU), and the x-axis shows the starting amountof antibody used to generate the surface. (C) Relative antibody activityis plotted. Data represent means (±SD) of triplicate experiments.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489644

when compared to physical adsorption method at 0.01 pmol ofantibody. From another perspective, surfaces generated usingonly 0.01 pmol of antibody via the UV-NBS method deliveredbetter or comparable antigen detection efficiency (10.2 × 103

RFU) when compared to surfaces generated using even 10-foldmore antibody via the other two methods (2.4 × 103 and 11.0× 103 RFU for physical adsorption and ε-NH3

+ methods,respectively). Combined, these results demonstrate that theUV-NBS method provides us with an effective antibodyimmobilization method to generate functionalized surfaceswith the highest antibody detection efficiency.Next, we compared the amount of surface immobilized

antibody in all three methods as determined by a secondaryanti-Fc antibody. Interestingly, the amount of surfaceimmobilized antibody was significantly higher with the ε-NH3

+ method when compared to the physical adsorption andUV-NBS methods (Figure 4B). Despite the lesser amount ofsurface immobilized antibody with the UV-NBS method,antigen detection efficiency was significantly higher (Figure4A), demonstrating that the UV-NBS method yields enhancedpreservation of antigen binding activity per surface immobilizedantibody. This result strongly suggests that the UV-NBSmethod highly preserves antibody’s antigen binding activity as aresult of the site-specific immobilization of antibody on thesurface. It is noteworthy to mention that quantification ofsurface immobilized antibody, via an anti-Fc secondaryantibody, in both ε-NH3

+ and physical adsorption methods isan underestimate. This results from immobilization at randomsites, which leaves the Fc domain unavailable for binding in afraction of antibody molecules. Therefore, the enhancementobserved for relative antibody activity with the UV-NBSmethod is an underestimate and the actual enhancement islikely even higher.To further emphasize this point, we calculated the relative

antibody activity for all methods by using the ratio of the signalfrom antigen detection efficiency to the total surfaceimmobilized antibody signal (Figure 4C). As expected, UV-NBS resulted in a much higher relative antibody activity thanthe other two methods, reaching up to ∼43-fold and ∼91-foldenhancements in relative antibody activity when compared tothe ε-NH3

+ and physical adsorption methods, respectively.Taken together, these results strongly suggest that the UV-NBSmethod highly preserves antibody’s antigen binding activity as aresult of oriented immobilization onto surfaces, while activity issignificantly lost with the ε-NH3

+ and physical adsorptionmethods as a result of randomly orientated antibodies. Theseresults are summarized in Tables S1 and S3.Similar trends, with even more dramatic enhancement in

surface antigen detection efficiency and relative antibodyactivity, were observed with the IgGFITC/FITC antibody/antigen system (Figure 5). The UV-NBS method yieldedfunctionalized surfaces with significantly higher antigendetection efficiency, enhancements reaching up to ∼3.6-foldwhen compared to the ε-NH3

+ method at 0.01 pmol antibodyand up to ∼280-fold when compared to physical adsorptionmethod at 0.05 pmol of antibody. No antigen binding wasdetectable with the physical adsorption method at 0.01 pmol ofantibody. Despite lower amounts of surface immobilizedantibody, UV-NBS also resulted in a much higher relativeantibody activity than the other two methods, reaching to ∼94-fold and ∼674-fold enhancements when compared to the ε-NH3

+ and physical adsorption methods, respectively. Theseresults are summarized in Tables S2 and S3. These experiments

were also performed with a protein-based antigen detectionsystem, IgGstreptavidin/streptavidin, which yielded similar trends(Figure S13).Finally, we evaluated the antigen detection sensitivity of the

surfaces generated by the three immobilization methods.Sensitivity (S) is defined as the slope of the linear regressionline obtained by plotting detection signal versus antigenconcentration.35 A larger slope indicates a higher degree ofsensitivity. Sensitivity was calculated and compared for all threeimmobilization techniques using IgGDNP (0.1 pmol) as thecapture antibody, DNP−biotin conjugate as the antigen, andstreptavidin−HRP as the reporter. A fluorescent HRP substratewas employed to determine the dose−response curve of a rangeof standard antigen concentrations from 0 to 10 μM. The linearregression equation was obtained for all three methods withregression coefficients (R2) of 0.9511, 0.9486, and 0.9576 forthe UV-NBS, ε-NH3

+, and physical adsorption methods,respectively (Figure 6). The UV-NBS immobilization methoddisplayed overall higher signal intensity with the highestsensitivity (S = 1.796), ε-NH3

+ immobilization method withlower signal intensity and lower sensitivity (S = 1.224), andphysical adsorption with significantly lower signal intensity andthe lowest sensitivity (S = 0.227). The difference in the slope of

Figure 5. Evaluation of the UV-NBS method in comparison to ε-NH3+

and physical adsorption methods using the mouse IgGFITC/FITCantibody/antigen system. The 96-well plates were functionalized withIgGFITC using each respective method. (A) Antigen detectionefficiency of the surfaces was determined with saturating levels ofFITC−biotin as the antigen and streptavidin−HRP as the reporter.The y-axis shows the relative fluorescence units (RFU) obtained usingan HRP substrate, and the x-axis shows the starting amount ofantibody used to generate the surface. (B) Total surface immobilizedantibody was quantified using an HRP linked anti-Fc secondaryantibody. (C) Relative antibody activity is plotted. Data representmeans (±SD) of triplicate experiments.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489645

the regression curves can be explained by the enhanced bindingefficiency of the UV-NBS surface, which results from enhancedantibody activity. Combined, these results demonstrate thatsurfaces generated by the UV-NBS method produce a highersensitivity in antigen detection than the ε-NH3

+ and thephysical adsorption methods, 1.5- and 7.9-fold enhancement insensitivity, respectively.Despite the enhanced sensitivity achieved with the UV-NBS

method, an improvement at the lower limits of detection (LLD,3SD to the mean of the zero standard35) was not detectableemploying this antibody/antigen system. The LLD werecomparable at ≤10 nM for all three methods after the 20min reaction time. It is noteworthy that this estimated LLDvalue is in the same range as the dissociation constant of theantibody/antigen interaction (IgGDNP/DNP; Kd = 8 nM). Forthis reason, a reduction in the detection of analyte atconcentrations below the Kd value can be attributed to theloss in the antibody’s ability to bind to its antigen and is not areflection of a shortcoming of the UV-NBS method. On theother end of the spectrum, the upper limit of detection isgoverned by the inherent limitations of the HRP enzyme basedassay used in the UV-NBS method. Although the intrinsickinetic parameters of the enzyme and the substrate are notaltered, substrate diffusion through the bulk solution to thesurface where the enzyme is immobilized can become the rate-limiting factor.44 As a result, increasing the amount of HRP onthe surface beyond a certain threshold will not increase the rateof product formation, yielding a plateau in the signal intensity.This is a possible explanation for the plateau observed in Figure2B.The UV-NBS method may be further improved by

optimizing the conditions to increase the total amount ofantibody immobilized on the surface. IBA is a hydrophobicmolecule, which may cause it to associate with the maleicanhydride polymer, rendering it partially inaccessible forbinding to the antibody. Hence, a more hydrophilic ligandand a more uniform surface would make the NBS ligand moreaccessible to the antibody and may improve the yield of thesurface immobilized antibody. However, a ligand with a higheraffinity for the NBS and increased photoreactivity would have amuch greater impact on enhancing the coupling efficiency ofthe antibody to the surface immobilized ligand. These studiesare currently ongoing in our laboratory.

■ CONCLUSIONThe results presented in this study establish the UV-NBSmethod as a practical, gentle, and reproducible method for sitespecific antibody immobilization on surfaces. The site specificcross-linking at the conserved NBS provides preservation ofantibody’s antigen binding activity and has the potential to beapplied to various immunosensor platforms. In our experi-ments, surfaces functionalized with the UV-NBS methoddisplayed significantly enhanced antigen detection efficiency,higher relative antibody activity, and improved antigendetection sensitivity in comparison to two other commonlyused antibody immobilization methods.Finally, the UV-NBS method can be adapted to different

detection modalities other than enzyme-based assays such ascolorimetric, fluorescence, impedance, refractive index, andplasmon modalities. With the development of micro- andnanofluidic diagnostic chips the physical detection areas inthese devices are approaching length scales comparable to thesize of the detection molecules themselves.45,46 Because of thelimited detection area, and therefore decreased number ofantibodies that can be presented, it is critical that everyantibody on the detection surface maintains its ability torecognize and bind its specific antigen. The UV-NBS methodprovides a great advantage over other immobilizationtechniques by providing that each immobilized antibodypreserves its antigen binding activity, which is essential toprovide for reliable outcomes. Taken together, the UV-NBSmethod provides a site-specific, practical, and flexible surfaceimmobilization method, with significant implications in thedevelopment of a large array of immunosensor platforms.

■ ASSOCIATED CONTENT*S Supporting InformationFigures S1−S15 and Tables S1−S3. This material is availablefree of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Tel 574-631-1429, Fax 574-631-8366, e-mail bbilgicer@nd.edu.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors thank the Center for Environmental Science andTechnology for usage of the DLS and the Notre Dame MassSpectrometry and Proteomics Facility for usage of mass analysisinstrumentation.

■ REFERENCES(1) Caygill, R. L.; Blair, G. E.; Millner, P. A. A Review on ViralBiosensors to Detect Human Pathogens. Anal. Chim. Acta 2010, 681,8−15.(2) Leonard, P.; Hearty, S.; Brennan, J.; Dunne, L.; Quinn, J.;Chakraborty, T.; O’Kennedy, R. Advances in Biosensors for Detectionof Pathogens in Food and Water. Enzyme Microb. Technol. 2003, 32,3−13.(3) Liu, S.; Zhang, X.; Wu, Y.; Tu, Y.; He, L. Prostate-SpecificAntigen Detection by using a Reusable Amperometric ImmunosensorBased on Reversible Binding and Leasing of HRP-Anti-PSA fromPhenylboronic Acid Modified Electrode. Clin. Chim. Acta 2008, 395,51−56.

Figure 6. Linear regression analysis of dose−response curvescomparing the UV-NBS, ε-NH3

+, and physical adsorption methods.The 96-well plates were functionalized with 0.1 pmol of IgGDNP usingthe respective methods. Antigen capture was detected using increasingconcentrations of DNP−biotin (0−10 μM) as the antigen andstreptavidin−HRP as the reporter.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489646

(4) Warsinke, A. Point-of-Care Testing of Proteins. Anal. Bioanal.Chem. 2009, 393, 1393−1405.(5) Ghani, R.; Iqbal, A.; Akhtar, N.; Mahmood, A.; Malik, N. S.;Usman, M.; Ali, H.; Akram, M.; Asif, H. M. Identification of DifferentStages of Hepatitis B Infection with Enzyme Linked ImmunosorbantAssay (ELISA) and Polymerase Chain Reaction (PCR) Assay. J. Med.Plants Res. 2011, 5, 2572−2576.(6) Li, X.; Conklin, L.; Alex, P. New Serological Biomarkers ofInflammatory Bowel Disease. World J. Gastroenterol. 2008, 14, 5115−5124.(7) McNulty, C. A. M.; Lehours, P.; Megraud, F. Diagnosis ofHelicobacter Pylori Infection. Helicobacter 2011, 16, 10−18.(8) Moelans, C. B.; de Weger, R. A.; Van der Wall, E.; van Diest, P. J.Current Technologies for HER2 Testing in Breast Cancer. Crit. Rev.Oncol. Hematol. 2011, 80, 380−392.(9) Niitsu, T.; et al. Associations of Serum Brain-DerivedNeurotrophic Factor with Cognitive Impairments and NegativeSymptoms in Schizophrenia. Prog. Neuropsychopharmacol. Biol.Psychiatry 2011, 35, 1836−1840.(10) Shephard, G. S. Determination of Mycotoxins in Human Foods.Chem. Soc. Rev. 2008, 37, 2468−2477.(11) Xiong, Q.; Ge, F. Identification and Evaluation of a Panel ofSerum Biomarkers for Predicting Response to Thalidomide inMultiple Myeloma Patients. Expert Rev. Proteomics 2011, 8, 439−442.(12) Yotsumoto, H. Specific Immune-Based Diagnosis of Tuber-culosis Infection. Rinsho Byori 2008, 56, 1026−33.(13) Kausaite-Minkstimiene, A.; Ramanaviciene, A.; Kirlyte, J.;Ramanavicius, A. Comparative Study of Random and OrientedAntibody Immobilization Techniques on the Binding Capacity ofImmunosensor. Anal. Chem. 2010, 82, 6401−6408.(14) Butler, J. E.; Ni, L.; Brown, W. R.; Joshi, K. S.; Chang, J.;Rosenberg, B.; Voss, E. W., Jr. The Immunochemistry of SandwichELISAs: VI. Greater than 90% of Monoclonal and 75% of PolyclonalAnti-Fluorescyl Capture Antibodies (CAbs) are Denatured by PassiveAdsorption. Mol. Immunol. 1993, 30, 1165−1175.(15) Butler, J. E.; Ni, L.; Nessler, R.; Joshi, K. S.; Suter, M.;Rosenberg, B.; Chang, J.; Brown, W. R.; Cantarero, L. A. The Physicaland Functional-Behavior of Capture Antibodies Adsorbed onPolystyrene. J. Immunol. Methods 1992, 150, 77−90.(16) Mendoza, L. G.; McQuary, P.; Mongan, A.; Gangadharan, R.;Brignac, S.; Eggers, M. High-Throughput Microarray-Based Enzyme-Linked Immunosorbent Assay (ELISA). BioTechniques 1999, 27, 778.(17) Byeon, J.; Limpoco, F. T.; Bailey, R. C. Efficient Bioconjugationof Protein Capture Agents to Biosensor Surfaces using Aniline-Catalyzed Hydrazone Ligation. Langmuir 2010, 26, 15430−15435.(18) Song, F.; Chan, W. C. W. Principles of Conjugating QuantumDots to Proteins Via Carbodiimide Chemistry. Nanotechnology 2011,22, 494006.(19) Patel, N.; Davies, M.; Hartshorne, M.; Heaton, R.; Roberts, C.;Tendler, S.; Williams, P. Immobilization of Protein Molecules OntoHomogeneous and Mixed Carboxylate-Terminated Self-AssembledMonolayers. Langmuir 1997, 13, 6485−6490.(20) MacBeath, G.; Schreiber, S. Printing Proteins as Microarrays forHigh-Throughput Function Determination. Science 2000, 289, 1760−1763.(21) Jyoung, J.; Hong, S.; Lee, W.; Choi, J. Immunosensor for theDetection of Vibrio Cholerae O1 using Surface Plasmon Resonance.Biosens. Bioelectron. 2006, 21, 2315−2319.(22) Faye, C.; Chamieh, J.; Moreau, T.; Granier, F.; Faure, K.; Dugas,V.; Demesmay, C.; Vandenabeele-Trambouze, O. In Situ Character-ization of Antibody Grafting on Porous Monolithic Supports. Anal.Biochem. 2012, 420, 147−154.(23) Bhatia, S.; Shriverlake, L.; Prior, K.; Georger, J.; Calvert, J.;Bredehorst, R.; Ligler, F. Use of Thiol-Terminal Silanes andHeterobifunctional Crosslinkers for Immobilization of Antibodies onSilica Surfaces. Anal. Biochem. 1989, 178, 408−413.(24) Vijayendran, R. A.; Leckband, D. E. A Quantitative Assessmentof Heterogeneity for Surface-Immobilized Proteins. Anal. Chem. 2001,73, 471−480.

(25) Peluso, P.; Wilson, D.; Do, D.; Tran, H.; Venkatasubbaiah, M.;Quincy, D.; Heidecker, B.; Poindexter, K.; Tolani, N.; Phelan, M.;Witte, K.; Jung, L.; Wagner, P.; Nock, S. Optimizing AntibodyImmobilization Strategies for the Construction of Protein Microarrays.Anal. Biochem. 2003, 312, 113−124.(26) Brogan, K. L.; Wolfe, K. N.; Jones, P. A.; Schoenfisch, M. H.Direct Oriented Immobilization of F(Ab ’) Antibody Fragments onGold. Anal. Chim. Acta 2003, 496, 73−80.(27) Lu, B.; Xie, J. M.; Lu, C. L.; Wu, C. G.; Wei, Y. OrientedImmobilization of Fab’ Fragments on Silica Surfaces. Anal. Chem.1995, 67, 83−87.(28) Vikholm-Lundin, I. Immunosensing Based on Site-DirectedImmobilization of Antibody Fragments and Polymers that ReduceNonspecific Binding. Langmuir 2005, 21, 6473−6477.(29) Lofas, S.; Johnsson, B.; Edstrom, A.; Hansson, A.; Lindquist, G.;Hillgren, R.; Stigh, L. Methods for Site Controlled Coupling toCarboxymethyldextran Surfaces in Surface-Plasmon ResonanceSensors. Biosens. Bioelectron. 1995, 10, 813−822.(30) Lee, W.; Oh, B.; Lee, W.; Choi, J. Immobilization of AntibodyFragment for Immunosensor Application Based on Surface PlasmonResonance. Colloids Surf., B 2005, 40, 143−148.(31) Quarles, R. H. Specific Conjugation Reactions of theOligosaccharide Moieties of Immunoglobulins. J. Appl. Biochem.1985, 7, 347−355.(32) Hoffman, W. L.; Oshannessy, D. J. Site-Specific Immobilizationof Antibodies by their Oligosaccharide Moieties to New HydrazideDerivatized Solid Supports. J. Immunol. Methods 1988, 112, 113−120.(33) Qian, W. P.; Xu, B.; Wu, L.; Wang, C. X.; Yao, D. F.; Yu, F.;Yuan, C. W.; Wei, Y. Controlled Site-Directed Assembly of Antibodiesby their Oligosaccharide Moieties Onto APTES Derivatized Surfaces.J. Colloid Interface Sci. 1999, 214, 16−19.(34) Zara, J. J.; Wood, R. D.; Boon, P.; Kim, C. H.; Pomato, N.;Bredehorst, R.; Vogel, C. W. A Carbohydrate-Directed Heterobifunc-tional Cross-Linking Reagent for the Synthesis of Immunoconjugates.Anal. Biochem. 1991, 194, 156−162.(35) Han, H. J.; Kannan, R. M.; Wang, S.; Mao, G.; Kusanovic, J. P.;Romero, R. Multifunctional Dendrimer-Templated Antibody Pre-sentation on Biosensor Surfaces for Improved Biomarker Detection.Adv. Funct. Mater. 2010, 20, 409−421.(36) Abraham, R.; Moller, D.; Gabel, D.; Senter, P.; Hellstrom, I.;Hellstrom, K. E. The Influence of Periodate-Oxidation on Mono-clonal-Antibody Avidity and Immunoreactivity. J. Immunol. Methods1991, 144, 77−86.(37) Leibiger, H.; Wustner, D.; Stigler, R.; Marx, U. VariableDomain-Linked Oligosaccharides of a Human Monoclonal IgG:Structure and Influence on Antigen Binding. Biochem. J. 1999, 338,529−538.(38) Qian, J.; Liu, T.; Yang, L.; Daus, A.; Crowley, R.; Zhou, Q.Structural Characterization of N-Linked Oligosaccharides on Mono-clonal Antibody Cetuximab by the Combination of OrthogonalMatrix-Assisted Laser desorption/ionization Hybrid Quadrupole-Quadrupole Time-of-Flight Tandem Mass Spectrometry andSequential Enzymatic Digestion. Anal. Biochem. 2007, 364, 8−18.(39) Handlogten, M. W.; Kiziltepe, T.; Moustakas, D. T.; Bilgicer, B.Design of a Heterobivalent Ligand to Inhibit IgE Clustering on MastCells. Chem. Biol. 2011, 18, 1179−1188.(40) Rajagopalan, K.; Pavlinkova, G.; Levy, S.; Pokkuluri, P. R.;Schiffer, M.; Haley, B. E.; Kohler, H. Novel Unconventional BindingSite in the Variable Region of Immunoglobulins. Proc. Natl. Acad. Sci.U. S. A. 1996, 93, 6019−6024.(41) Lefranc, M. P.; Pommieaaa, C.; Ruiz, M.; Giudicelli, V.;Foulquier, E.; Truong, L.; Thouvenin-Contet, V.; Lefranc, G. IMGTUnique Numbering for Immunoglobulin and T Cell Receptor VariableDomains and Ig Superfamily V-Like Domains. Dev. Comp. Immunol.2003, 27, 55−77.(42) Russ, M.; Lou, D.; Kohler, H. Photo-Activated Affinity-SiteCross-Linking of Antibodies using Tryptophan Containing Peptides. J.Immunol. Methods 2005, 304, 100−106.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489647

(43) Meisenheimer, K. M.; Koch, T. H. Photocross-Linking ofNucleic Acids to Associated Proteins. Crit. Rev. Biochem. Mol. Biol.1997, 32, 101−140.(44) Shuler, M. L.; Kargi, F. Immobilized Enzyme Systems. InBioprocess Engineering: Basic Concepts; Prentice-Hall: Englewood Cliffs,NJ, 2008; Vol. 2, p 79.(45) Ng, A. H. C.; Uddayasankar, U.; Wheeler, A. R. Immunoassaysin Microfluidic Systems. Anal. Bioanal. Chem. 2010, 397, 991−1007.(46) Hu, W.; Li, C. M. Nanomaterial-Based Advanced Immuno-assays. Wiley Interdiscip. Rev.: Nanomed. Nanobiotechnol. 2011, 3, 119−133.

Langmuir Article

dx.doi.org/10.1021/la301887s | Langmuir 2012, 28, 9640−96489648