Analytical Biochemistry 456 (2014) 6–13

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/locate /yabio

Coordination complexes as molecular glue for immobilizationof antibodies on cyclic olefin copolymer surfaces

http://dx.doi.org/10.1016/j.ab.2014.03.0230003-2697/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author at: Australian Institute for Bioengineering and Nano-technology, University of Queensland, Brisbane, 4072 QLD, Australia. Fax: +61 73365 3833.

E-mail address: a.whittaker@uq.edu.au (A.K. Whittaker).1 Abbreviations used: COC, cyclic olefin copolymer; PS, polystyrene; PMMA,

poly(methyl methacrylate); ELISA, enzyme-linked immunosorbent assay; COP, cyclicolefin polymer; XPS, X-ray photoelectron spectroscopy; AFM, atomic force micros-copy; Mes, 2-(N-morpholino)ethanesulfonic acid; PBS, phosphate buffer saline;Hepes, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; BSA, bovine serumalbumin; T20, Tween 20; TnI, troponin I; MAb, monoclonal antibody; TNFa, tumornecrosis factor alpha; Strep–HRP, streptavidin–horseradish peroxidase conjugate;TSH, thyroid-stimulating hormone; GAM–HRP, goat anti-mouse–horseradish perox-idase conjugate; TMB, 3,30 ,5,50-tetramethylbenzidine; NMR, nuclear magnetic reso-nance; CDCl3, deuterated chloroform; OD, optical density; M&G–COC, Mix&Go–cyclicolefin copolymer.

Huey Wen Ooi a,b, Shaun J. Cooper a, Chang-Yi Huang a, Dean Jennins a, Emma Chung a, N. Joe Maeji a,Andrew K. Whittaker b,c,⇑a Anteo Diagnostics, Eight Mile Plains, 4113 QLD, Australiab Australian Institute for Bioengineering and Nanotechnology, University of Queensland, Brisbane, 4072 QLD, Australiac Centre for Advanced Imaging, University of Queensland, Brisbane, 4072 QLD, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 27 January 2014Received in revised form 26 March 2014Accepted 29 March 2014Available online 8 April 2014

Keywords:ImmunoassayCyclic olefin copolymersSurface functionalizationMetal complexesEnzyme-linked immunosorbent assay

A novel metal-based chelating method has been used to provide an order of magnitude increase in immu-noassay performance on cyclic olefin copolymer (COC) plastics compared with passive binding. COCs arehydrophobic, and without surface modification they are often unsuitable for applications where proteinadhesion is desired. When interacting with the bare plastic, the majority of the bound proteins will bedenatured and become nonfunctional. Many of the surface modification techniques reported to daterequire costly equipment setup or the use of harsh reaction conditions. Here, we have successfully dem-onstrated the use of a simple and quick metal chelation method to increase the sensitivity, activity, andefficiency of protein binding to COC surfaces. A detailed analysis of the COC surfaces after activation withthe metal complexes is presented, and the immunoassay performance was studied using three differentantibody pairs.

� 2014 Elsevier Inc. All rights reserved.

Cyclic olefin copolymers (COCs)1 are amorphous thermoplasticsthat have recently been used extensively for various applicationssuch as biosensors [1,2], biodiagnostic chips, microfluidic devices[3,4], micro total analysis systems [5], DNA immobilization, immu-noassays, and microarrays [6]. The clarity and optical resistanceagainst commonly used sterilization regimes [7] makes COCs moreattractive for many more applications than polystyrene (PS), whichtends to cloud when sterilized with ethylene oxide or radiation. Inaddition, COCs offer high chemical resistance to acids, bases, andpolar organic solvents (e.g., dimethyl sulfoxide, methanol, acetone)

as well as thermal resistance by selecting COCs with high glass tran-sition temperatures [8]. Furthermore, COCs are photosensitive,which allows the writing of gratings for selective and label-free bio-sensing [9]. In contrast to poly(methyl methacrylate) (PMMA), COCsare less sensitive to humidity [10] and have orders of magnitudeslower loss at terahertz frequencies [11]. The density of COCs isapproximately half the density of glass, and COCs are less brittle,which makes them an attractive alternative to glass in optical com-ponents when weight or durability becomes important. By compar-ison, polycarbonate and PMMA, both used as alternative materials inlife sciences, are brittle, are prone to chemical degradation, and haveinferior optical properties compared with COCs [12].

Despite COCs offering excellent bulk properties for many appli-cations as described above, surface modification of these materialsremains a challenge. These thermoplastic materials are preparedby chain copolymerization of cyclic monomers such as norbornenewith ethylene or by ring-opening methathesis polymerization ofcyclic monomers followed by hydrogenation [13]. Their purehydrocarbon composition means that they lack readily accessiblefunctional groups. The surface hydrophobicity of COC promotesfouling by proteins, a complicating factor when adhesion of pro-teins is unwanted. In cases where adhesion of proteins is desired,nonspecific passive immobilization of proteins often leads to lossof function. The introduction of hydrophilic polar functional groups

Immobilization of antibodies on COC surfaces / H.W. Ooi et al. / Anal. Biochem. 456 (2014) 6–13 7

through various surface pretreatments, to increase the surface freeenergy, has been suggested as a method to minimize such difficul-ties. Among the surface modification techniques reported on COCsare plasma treatment [3,14–17], ultraviolet (UV) [5], gamma orelectron beam radiosterilization [18], and chemical treatment[19,20] to oxidize the nonreactive surface.

Typically during the surface modification process, the polymersurface is modified through oxidation, degradation, and crosslink-ing, which inevitably causes structural alterations to the first fewmolecular layers on the polymer surface. Plasma treatment derivedfrom oxygen, ammonia, and noble gases is one of the commonlyused surface pretreatment methods. Although COC surfaces aresuccessfully rendered hydrophilic by the plasma, these surfacesare unstable because polymer chains on the surface tend to rear-range and return the surface to the native unreactive hydrophobicform. This phenomenon has been widely confirmed through incre-ment in contact angle measurements [21] and f-potential decay[22], which often occurs over a period of days. Furthermore, thesource of surface charge after pretreatment with plasma, gamma,or electron beam radiosterilization is unclear. To optimize thehydrophilic functionality and energy of COC surfaces, pretreatmentfor long duration and high power plasma source has been con-ducted. Although this may increase shelf life, the surface roughnesscan be compromised, thereby rendering the materials incompati-ble with applications that require homogeneous surface roughness.

It is well known from several decades of experience in enzyme-linked immunosorbent assay (ELISA) that plain PS microtiter plateswithout surface modification are suitable for only a small percent-age of proteins. More hydrophilic proteins such as antibodiesrequire polar groups on the plate surface, and as a consequencethere are many types of surface-modified microtiter plates. Follow-ing this learning, increasing the polarity of COC surfaces is likely tobe essential to improve its ability to bind antibodies. The differenttypes of cyclic olefin polymers (COPs) and COCs available commer-cially have varying abilities to bind proteins. Some workers havereported problems associated with nonspecific binding to COC sur-faces [23]; however, there have also been reports of COP materialsthat are inert to biological molecules. Niles and Coassin reportedthe use of COP (Zeonor 1420R) as the base material for a high-den-sity multiwall plate (Aurora Biotechnologies) [12]. These plateswere reported to be inert to biological molecules except smallhydrophobic polypeptides less than 10 kDa. Based on thesereports, it is difficult to achieve desirable antibody binding proper-ties on these surfaces without applying surface treatments.

Where passive binding is undesirable or covalent linkages aredifficult to achieve, a completely different approach to binding pro-teins via the use of metal complexes can be used [24]. Using a high-throughput surface discovery approach, many thousands of metalcomplexes were screened for their efficiency to bind antibodies,and several particular complexes were found to be highly efficientin binding proteins. These particular metal complexes, calledMix&Go [25–27], depend on two basic characteristics for binding.Using slow exchanging metal complexes such as chromium(III) inoligomeric form, there is avidity or multicomponent chelation tothe synthetic surface while retaining potential to similarly bindproteins. A single metal–ligand interaction may readily break; how-ever, the odds of multiple interactions breaking simultaneously arelow. However, COC plastics are hydrophobic and should not haveelectron donating groups to chelate to metal complexes.

It has been previously shown that versions of such metal com-plexes can bind to plain non-irradiated PS surfaces. Presumably,these metal complexes are in coordination with the p-electronsof the phenyl ring, and although one interaction may be weak, amultiplicity of interactions is sufficient to create a completelydifferent surface over the previously hydrophobic nonfunctionalsurface. Considering their structure, COP/COC surfaces do not have

electron donating potential to bind metal complexes. However,work on enhancing adhesion of metals onto such COC/COP surfaceshas been reported [8,15]. Nikolova and coworkers studied theeffects of plasma treatment on the adhesive strength of aluminumand copper on COC surfaces [15]. The adhesive strength in COC–metal composites increased with the intensity of plasma treat-ment; however, the untreated controls demonstrated the abilityto bind these metals as well. If a weak potential to bind metalson COC surfaces exists, a basic strategy of using slowly exchangingmetal complexes and avidity of such metals using polymeric metalconstructs will potentially allow the formation of a surface thatpromotes protein binding. Once formed, the residual coordinationsites remaining after forming metal complexed surfaces are che-lated to small ligands such as water that undergo exchange witha specific half-life. In other words, the Mix&Go activated COC sur-faces can be stored but remain active indefinitely. Importantly, inthe presence of many classes of biomolecules (e.g., antibodies,streptavidin, protein A or G), coordination forces bind such pro-teins firmly onto the surface.

In this study, the potential of the metal complexes to bind toCOC surfaces to form a hydrophilic chelating surface was investi-gated. Three different sandwich assays were performed using oneparticular formulation called Mix&Go Biosensor, which was knownto perform well on PS surfaces as well as silica and other metaloxide surfaces. Two assays were known to work well by passivebinding on PS microtiter plates, and the third sandwich assay gavepoor results under the same conditions. To better understand theeffects of metal coordination on such hydrophobic surfaces, surfaceanalytical techniques such as X-ray photoelectron spectroscopy(XPS), atomic force microscopy (AFM), and contact angle studieson these metal complex-activated surfaces were performed.

Materials and methods

Materials

2-(N-Morpholino)ethanesulfonic acid (Mes), phosphate buffersaline (PBS), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid(Hepes), bovine serum albumin (BSA), Tween 20 (T20), sulfuric acid,and Greiner Bio-One COC plates (cat. no. M3812) were obtained fromSigma–Aldrich. Troponin I (TnI) monoclonal antibody (MAb, 19C7),human cardiac TnI antigen, and biotinylated TnI MAb (16A11) werepurchased from HyTest (Finland). MAb1 tumor necrosis factor alpha(TNFa) antibody, biotinylated MAb11 TNFa antibody, and streptavi-din–horseradish peroxidase conjugate (Strep–HRP) were purchasedfrom BD Biosciences. Recombinant human TNFa was purchasedfrom R&D Systems (USA). Clone 057-11003 thyroid-stimulating hor-mone (TSH) MAb was purchased from Meridian Life Science. Clone5403 biotinylated TSH MAb was purchased from Medix Biochemica.Goat anti-mouse–horseradish peroxidase conjugate (GAM–HRP)was obtained from Jackson ImmunoResearch (USA). 3,30,5,50-tetra-methylbenzidine (TMB) substrate was purchased from Thermo Sci-entific. Mix&Go Biosensor was obtained from Anteo Diagnostics.Buffers prepared were 50 mM Mes at pH 5.2 (coating buffer for TNFaand TSH assays), 10 mM PBS with 1% BSA (assay buffer for TNFa andTSH assays), 50 mM Mes with 5% BSA (blocker for TNFa and TSHassays), 10 mM PBS with 0.05% T20 (wash buffer for TNFa, TSH,and TnI assays), Hepes at pH 7.4 (coating buffer for TnI assay),10 mM PBS with 0.1% T20 (assay buffer for TnI), and PBS with 1%casein (blocker for TnI assay).

Characterization

13C nuclear magnetic resonance (NMR) spectra were obtainedon a Bruker AVANCE 400-MHz instrument operating at

8 Immobilization of antibodies on COC surfaces / H.W. Ooi et al. / Anal. Biochem. 456 (2014) 6–13

400.13 MHz with a standard Bruker 5-mm broadband observedgradient probe. Deuterated chloroform (CDCl3) was used as the sol-vent. Spectra were referenced to solvent signal at d13C = 77.0 ppm.All spectra were processed using Bruker TOPSPIN.

Contact angle measurements were carried out on an apparatuscomposed of an adjustable stage and lens assembly fitted with acamera and linked to Scion imaging software. The sample wasplaced on the stage, and 5 ll of Milli-Q water was transferred bya 50-ll glass flat-tip syringe onto the sample surface. The needletip was removed from the water droplet before capturing an imageof the droplet. Contact angles were measured using the imagingsoftware. All measurements were taken five times.

A Kratos Axis Ultra X-ray Photoelectron Spectrometer withmonochromatic Al Ka X-rays (1486.6 eV) at 150 W (15 kV,10 mA) was used to collect XPS spectra. Photoelectron data werecollected at a takeoff angle of theta = 90�. Survey scans were takenat an analyzer pass energy of 160 eV. Survey scans were performedover a range of 1200 to 0 eV binding energies with 1.0-eV steps anda dwell time of 100 ms. The base pressure in the analysis chamberwas set at 1.0 � 10�9 Torr and during sample analysis was set at1.0 � 10�8 Torr. Processing was performed using CasaXPS. Bindingenergy corrections were made by referencing spectra to the carbon1s fixed at 285 eV [28]. Survey scans were taken of an area of0.7 � 0.3 mm2.

An MFP-3D (Asylum Research) atomic force microscope wasused for all of the measurements. The cantilevers used wereHA_NC (Etalon) from NT–MDT (Russia) having a nominal springconstant of 4.5 N/m and a nominal resonant frequency of145 kHz. All of the images were obtained by employing the tappingmode of the atomic force microscope in air. The microscope wasmounted on an anti-vibrational table (Herzan) and operated withinan acoustic isolation enclosure (TMC, USA).

Immunoassays

The plates were treated by incubating each well with Mix&GoBiosensor solution for 1 h. Wells were then washed thoroughlywith deionized water, followed by a wash with coating buffer.For passive binding experiments, the plates were used as received.

For the loading assays, capture antibodies were diluted in coat-ing buffer at 1 lg/ml and incubated in plates for 30 min beforewashing with wash buffer twice on a Tecan 96 Plate Washer. Plateswere then blocked with blocking buffer for 1 h. The washing stepwas carried out three times to remove excess blocker. This was fol-lowed by the incubation of GAM–HRP in assay buffer (0.1 lg/ml)for 30 min and then washing five times with wash buffer. The sub-strate, TMB, was reacted with HRP for 7 min before the reactionwas terminated by the addition of 2 M sulfuric acid, which resultedin a yellow product.

For the sandwich assays, capture antibodies were diluted incoating buffer at desired concentrations and incubated in platesfor 30 min (unless stated otherwise) before washing with washbuffer twice on a Tecan 96 Plate Washer. Plates were then blockedwith blocking buffer for 1 h. The washing step was carried outthree times to remove excess blocker. Antigen in assay bufferwas incubated in plates for 1 h before washing five times withwash buffer to remove excess analytes. This was followed by theincubation of biotinylated detection antibodies in assay buffer(0.5 lg/ml) for 30 min and then washing five times with wash buf-fer. Strep–HRP (0.1 lg/ml in assay buffer) was then reacted withthe biotinylated antibodies for 15 min and washed five times withwash buffer. The substrate, TMB, was reacted with HRP for 7 minbefore the reaction was terminated by the addition of 2 M sulfuricacid, which resulted in a yellow product.

Optical density (OD) of the colorimetric substrate forimmunoassays was measured using a Tecan Infinite 200 PRO.

The measurements were conducted at 450 nm with the referencewavelength at 620 nm. The bandwidth at both wavelengths wasset to 9 nm. The temperature of the measurements was 23 or 24 �C.

Results and discussion

Chemistry of the surfaces

The class of commercial materials known as COCs encompassesa large number of potential cyclic monomers prepared using vari-ous polymerization methods. The chemical structure and composi-tion of the COC polymers are important because they determinethe physical and surface properties of the materials. Changes inthe copolymer content and chemical structure of the cyclic mono-mers can modify physical properties such as microstructure, glasstransition temperature, and surface energy [13,22,29]. Shin andcoworkers reported comprehensive 13C NMR studies of the chem-ical structures of a series of commercial COCs and COPs and relatedthese to the thermal and surface properties [13]. They demon-strated that with increasing amounts of bulky cyclic monomerunits, the glass transition temperature of the copolymers increasesand the presence of ester or ether groups increases the surfaceenergy. The chemical composition of the COC plates used in thesestudies was determined by 13C NMR, as shown in Fig. 1, and thepolymer was found to be composed of ethylene and norbornenemonomers. The peaks in the NMR spectra were assigned accordingto the extensive NMR studies that have been reported in the liter-ature [13,30,31]. Resonances from the ethane-1,2-diyl units (car-bons 8 and 9) and the cyclic monomer (carbons 5 and 6)overlapped in the region between 29.8 and 31.8 ppm. Other peaksdue to the norbornanediyl units are at 32.7 ppm (carbon 7) andbetween 40.7 and 41.6 ppm (carbons 1 and 4). The main peaks ofinterest lie in the region of 46 to 48 ppm and are assigned to themethine carbons (2 and 3) of the norbornanediyl units. Thesepeaks provide information on the composition and structure ofthe copolymer. Rische and coworkers showed that, in general,COCs display either of two distinct NMR spectra depending onthe norbornene content [32]. Well-separated and narrow peakswere observed when the fraction of norbornene in the COC usedwas less than 50%, as opposed to broader and less separated peaksfor higher norbornene contents. These observations indicate thatthe COC studied here has less than 50% norbornene units and con-sists mainly of alternating ethanediyl/norbornanediyl and longsequences of ethylene units. NMR does not indicate the presenceof oxygen or other groups having electron donating potentialwithin this COC material.

The COC plates were treated with Mix&Go simply by exposingthe surfaces to the metal polymer solution for 1 h under ambientconditions (M&G–COC). The presence of bound metal complexeson the COC surface was confirmed through the detection of chro-mium peaks by XPS. Fig. 2shows the XPS spectra of the COC sur-faces before and after treatment with Mix&Go. Measurementswere conducted on three different areas of four treated wells,which were distributed randomly in a plate (see online Supple-mentary material). Table 1 shows the percentages of Cr relativeto other elements measured in each well. The variability in the dis-tribution of Cr between wells was relatively high, ranging from0.43 to 2.67% in the four wells tested. The Cr complex was stableto multiple washing steps, indicating that by some mode of actionmetal complexes could be bound to COC surfaces [8,15]. Theuntreated COC surface showed the presence of zinc, silicone, andoxygen, which may be attributed to the presence of molding agentsand contaminants or possibly oxidation of the surface introducedduring the injection molding process. It may be possible that theseoxygen species are the anchoring points for avidity binding of

Fig.1. 13C NMR spectrum of the COC plate in CDCl3.

Fig.2. XPS spectra of untreated COC (left) and M&G–COC (right).

Table 1Distribution of chromium on different COC wells.

Sample Average % of Cr Standard deviation

1 2.67 0.152 0.69 0.123 0.43 0.014 1.32 0.24

Immobilization of antibodies on COC surfaces / H.W. Ooi et al. / Anal. Biochem. 456 (2014) 6–13 9

these metal oligomers. Once the metal complex is deposited onthese sites, the complexes may aggregate or diffuse on the surfaceor through the bulk polymer bulk via random diffusion processes[33]. This may contribute to the variation in level of Cr detectedin the wells by XPS.

The roughness of the COC surfaces before and after treatmentwith Mix&Go was measured by AFM, as shown in Fig. 3. On

magnification (5 � 5 lm2 area), the metal complexed COC surfaceslooked distinctly different from the untreated surface, confirmingthat a metal complex film has coated onto the COC. The roughnessof the COC surfaces was measured by AFM to be 4.7 ± 0.5 nm andwas 4.5 ± 1.0 nm after treatment with Mix&Go. These results indi-cate that treatment with the metal complex solution did not alterthe topography of the surfaces. In contrast, alternative surfacetreatments can significantly increase surface roughness. COC sur-faces are commonly rendered hydrophilic through exposure tohigh-energy sources such as plasma and gamma irradiation. Thesetreatments produce unstable hydrophilic surfaces and, over pro-longed exposure time, surface etching occurs. Roy and coworkersobserved an increase in roughness from 11 to 23 nm after a120-s exposure to argon plasma treatment [3]. Further exposureup to 180 s produced COC surfaces with higher roughness, andthe use of oxygen instead of argon produced even rougher surfaces.Similarly, Nikolova and coworkers observed an increase in

Fig.3. AFM images of untreated COC (A) and M&G–COC (B) surfaces.

10 Immobilization of antibodies on COC surfaces / H.W. Ooi et al. / Anal. Biochem. 456 (2014) 6–13

roughness from 4.3 to 8.2 nm on their COC surfaces after 60 s ofoxygen plasma treatment [15].

The hydrophilicity of the COC surfaces was also determined viacontact angle measurements. The contact angle of the hydrophobicCOC surface was measured to be 95.6 ± 3.9�, which was similar topreviously reported values [5]. The treatment with Mix&Go did notsignificantly alter the hydrophobicity of the COC surfaces giventhat the contact angle was found to be 104.5 ± 2.6�. The detectionof chromium and the underlying surface using XPS and the mini-mal changes in roughness in the AFM images are suggestive thatthe metal complex coatings are very thin films and might nottotally cover the underlying surface. The absence of significantchanges in the contact angle further indicates that the metal com-plex coating applied was insufficient to alter the hydrophilicity ofthe surfaces. This shows that the application of Mix&Go retainsthe original properties of the COC surfaces but is adequate to pro-vide superior improvements to the immunological performance ofthe materials, as discussed in the following section.

Immunological performance

Binding of biomolecules such as antibodies and other proteinson untreated COC surfaces is driven by hydrophobic interactions,which often leads to conformational damage and subsequent lossof functionality. From the surface analysis studies, we have shownthat treatment with Mix&Go has created a very thin film of metalcomplexes that might not totally cover the underlying hydropho-bic surface (by contact angle measurements). This is consistentwith the lack of electron donating groups on the COC surface,leading to a new surface that may still include the potential forhydrophobic binding but now augmented by metal coordinationforces. This could be likened to PS microtiter plates, which havebeen surface treated to improve passive binding of more hydro-philic proteins such as antibodies.

To test antibody binding performance on COC plates treatedwith Mix&Go, three sandwich ELISAs (TNFa, TnI, and TSH) wereselected. The TNFa antibody pair was recommended for ELISA bythe manufacturer. The TSH antibody pair has been confirmed byprevious work to perform adequately on microtiter plates. TheTnI antibody pair was known to work when the capture antibodywas covalently coupled on particles but not by passive bindingon microtiter plates. Fig. 4A shows the loading assay data of mouseanti-human TNFa, TnI, and TSH antibodies using GAM–HRP as thesecondary antibody. The binding efficiency of all three capture

antibodies on untreated and Mix&Go-treated COC surfaces was dif-ferent, as determined by anti-species antibody binding and consis-tent with the differences observed by surface analysis. Using acapture antibody concentration of 1 lg/ml, TNFa and TnI antibod-ies can be successfully bound to untreated COC surfaces by passivebinding. However, the TSH antibody either did not bind or wastotally damaged such that it could not be detected by the anti-species antibody. In contrast, there was an increase in loading forall three capture antibodies on Mix&Go-treated COC surfaces.However, the absolute loading was different for each antibody,suggesting some antibody-specific factors in this loading assay.The results of sandwich ELISAs (1000 pg/ml TNFa and TnI antigenand 5 lIU/ml TSH antigen) at the same capture antibody concen-tration are shown in Fig. 4B. The amount of functional antibodies,as evidenced by antigen binding on untreated and Mix&Go-treatedCOC surfaces, clearly shows increased assay performance under theexperimental conditions. Both TnI and TSH assays did not give anysignal with passive binding, whereas TNFa assays on Mix&Go-treated surfaces gave approximately nine times greater signal.

To further understand the differences between untreated andMix&Go-activated COC surfaces, titrations of capture antibody aswell as antigens for all three ELISAs were also conducted. Fig. 5shows the titration curves for the TNFa system. Capture antibodytitration was conducted at an antigen concentration of 1000 pg/ml, and antigen titration with capture antibody was conducted at1 lg/ml. On M&G–COC, maximum signal was obtained at approx-imately 2 lg/ml capture antibody, with no further increase in per-formance with increasing concentrations of capture antibody. Incomparison, more than double that concentration (�5 lg/ml) isneeded to reach maximum capacity of a passive well. The maxi-mum OD values for the passive surface are approximately half ofthe OD values measured on the M&G–COC surface. The signal-to-noise ratio in the antigen titration study also significantlyimproved on the M&G–COC surfaces in comparison with theuntreated COC surfaces (see Fig. 5B). In addition, the percentagecoefficient of variation (%CV) of a plate improved significantly aftertreatment with Mix&Go from 16% (untreated) to 6% (treated) (seeSupplementary material). It is interesting to note that this 6% inter-well CV was achieved on Mix&Go-treated COC plates having chro-mium ion distributions varying from 0.43 to 2.67%. This suggestsgreater surface uniformity to the incoming capture antibody andvariations in coating thickness as opposed to percentage coverage.

TNFa sensitivity on the M&G–COC surfaces was significantlybetter, implying that the functionality of capture antibodies on

TNF TnI TSH0

1

2

3

OD

COCM&GC10

A

TNF TnI TSH0.0

0.2

0.4

0.6

0.8

1.0

OD

COCM&GC10

B

α α

Fig.4. (A) Loading assay of mouse anti-human TNFa, TnI, and TSH antibodies (1 lg/ml) with GAM–HRP as detection antibody (0.1 lg/ml) on untreated COC and M&G–COCplates. (B) Sandwich assay of mouse anti-human TNFa, TnI, and TSH antibodies (1 lg/ml), antigens (TNFa and TnI at 1000 pg/ml and TSH at 5 lIU/ml), and biotin mouse anti-human as detection antibody.

0 2 4 6 8 100.0

0.5

1.0

Concentration of cAb (µg/ml)

OD

A

0 1000 2000 30000.0

0.5

1.0

1.5

Concentration of Ag (pg/ml)

OD

B

Fig.5. TNFa sandwich assay: Titration of cAb (Ag = 1000 pg/ml) (A) and antigen (cAb = 1 lg/ml) (B) on COC (d) and M&G–COC (N) surfaces. cAb, capture antibody; Ag,antigen.

Immobilization of antibodies on COC surfaces / H.W. Ooi et al. / Anal. Biochem. 456 (2014) 6–13 11

M&G–COC was maintained, whereas passive binding on untreatedCOC must have led to functional damage. Similar improvementswere also observed for the TnI and TSH sandwich assays. TheTSH capture antibody titration curve reached a plateau earlier onthe M&G–COC surfaces at approximately 5 lg/ml, whereas thepassive curve increased up to the concentrations studied(Fig. 6A). There was a clear antigen titration on the M&G–COC sur-face, whereas the same assay on the untreated surface failed(Fig. 6B). Similarly, the TnI assay (Fig. 7) required fewer antibodiesto reach a maximum with higher antibody loading capacity andobtained a measurable antigen titration on the M&G–COC surfacescompared with the untreated COC surfaces. It should be noted thatthe measured OD values for the TnI assays are relatively low, butthis is not surprising considering that this capture antibody wasknown to be sensitive to passive binding. TnI sandwich assaysusing this particular capture antibody on commercially availablePS microtiter plates (MaxiSorp and Greiner Bio-One) gave very

0 2 4 6 8 100

1

2

3

Concentration of cAb (µg/ml)

OD

A

Fig.6. TSH sandwich assay: Titration of cAb (Ag = 50 lIU/ml) (A) and antigen (cAb = 1 lg

poor performance (see Supplementary material). Interestingly, onboth treated and untreated surfaces, there were relatively largequantities of immobilized antibodies by loading assay, indicatingthat the poor ELISAs were likely due to degradation of the TnIantibodies.

In addition to the improved assay performance, the binding ofantibodies to M&G–COC surfaces occurred at a much faster ratecompared with passive binding, suggesting a completely differentmode of binding. Typically, antibody coating on PS plates requiresan overnight incubation [34]. Fig. 8 shows the coating capture anti-body time profiles for the three sandwich assay systems. On theM&G–COC surfaces, a dramatic increase occurs during the first10 min of incubation for all three systems. As expected, the passiveCOC surfaces required longer incubation periods for antibodies tofully coat the wells. To demonstrate the efficiency of M&G–COCover untreated COC surfaces, a 24-h incubation time point, whichis considered to be the commonly used incubation period for

0 20 40 600.0

0.5

1.0

1.5

Concentration of Ag (µIU)

OD

B

/ml) (B) on COC (d) and M&G–COC (N) surfaces. cAb, capture antibody; Ag, antigen.

0 2 4 6 8 100.00

0.05

0.10

0.15

0.20

Concentration of cAb (µg/ml)

OD

A

0 500 10000.00

0.05

0.10

0.15

Concentration of Ag (pg/ml)

OD

B

Fig.7. TnI sandwich assay: Titration of cAb (Ag = 1000 pg/ml) (A) and antigen (cAb = 1 lg/ml) (B) on COC (d) and M&G–COC (N) surfaces. cAb, capture antibody; Ag, antigen.

0 20 40 60 80 1000.0

0.5

1.0

Time (min)

OD

A

0 20 40 60 80 1000.0

0.2

0.4

0.6

Time (min)

OD

B

0 20 40 60 80 1000.00

0.05

0.10

0.15

0.20

Time (min)

OD

C

Fig.8. Coating capture antibody (1 lg/ml) time profiles for TNFa (A), TSH (B), and TnI (C) measured from sandwich assay on COC (d) and M&G–COC (N) surfaces.

TNF TnI TSH0.0

0.2

0.4

0.6

0.8

OD

COCM&GC10

α

Fig.9. Performance of TNFa, TnI, and TSH sandwich assays on COC and M&G–COCsurfaces over capture antibody incubation time of 24 h.

12 Immobilization of antibodies on COC surfaces / H.W. Ooi et al. / Anal. Biochem. 456 (2014) 6–13

passive binding, was also measured for all three systems and isreported in Fig. 9. The measured signals after 24 h show that theincreased incubation period on untreated COC did not improvethe activity of the assays and was incapable of reaching the effi-

ciencies demonstrated by the M&G–COC surfaces. The lower ODvalues measured for the 24-h time points compared with theshorter time points may be attributed to the deterioration of thebiomolecules over time and the different kinetics observed at 4 �C.

Conclusions

In this study, a metal chelation system called Mix&Go Biosensorwas tested on COC plastics, a surface that was expected not to haveelectron donating groups for coordination of metal ions. However,surface analysis clearly shows effective binding of these metalcomplexes to the surface by simple addition of these aqueous solu-tions to the wells of the COC microtiter plate followed by incuba-tion for 1 h. These surface treatments were stable in theiractivated form and did not lead to any measurable change in sur-face roughness or hydrophobicity. For the three test assays, thistreatment led to significant increases in sensitivity, capture anti-body savings, and improved inter-well reproducibility to the pointwhere it enabled assays that were not possible by passive coating.

Immobilization of antibodies on COC surfaces / H.W. Ooi et al. / Anal. Biochem. 456 (2014) 6–13 13

COCs are highly hydrophobic and are known to lead to nonspe-cific binding of proteins, which often render them nonfunctional.However, their superior optical properties over other commonlyused polymers have increased their popularity in various applica-tions, and where needed different surface modification techniqueshave been applied to change the surface properties in order toimprove binding of proteins. Methods based on covalent bindingare usually preferred because they are more controllable and min-imize protein denaturation. However, covalent binding ofteninvolves syntheses that require synthetic skills. The method of sur-face treatment with Mix&Go implemented here is simple and eco-nomical and does not require in-depth chemistry laboratory skillsor the use of any sophisticated instruments. This preliminary studydid show variations in the amount of metal complexes in individ-ual wells of a microtiter plate detected by XPS. However, thesevariations were not reflected in the immunoassay performance ofthe activated COC surfaces. The improved performance of M&G–COC compared with COC was demonstrated through assays con-ducted on two antibody pairs (TNFa and TSH) known to work onstandard PS microtiter plates (by passive coating) and anotherantibody pair (cardiac biomarker, TnI) known not to work on stan-dard PS microtiter plates (by passive coating). All of these captureantibodies have been shown to bind much faster and reach maxi-mum coating at lower antibody concentrations, indicating minimalfunctional damage.

Acknowledgments

The authors thank the Australian National Fabrication Facility(ANFF), Queensland Node, for access to equipment. The authorsalso thank Barry Wood (Centre for Microscopy and Microanalysis,University of Queensland) and Elena Taran (ANFF) for discussionsregarding XPS and AFM, respectively. Financial support to H.W.O.in the form of an internship sponsored by AMSI Intern is gratefullyacknowledged.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ab.2014.03.023.

References

[1] G. Emiliyanov, P.E. Høiby, L.H. Pedersen, O. Bang, Selective serial multi-antibody biosensing with TOPAS microstructured polymer optical fibers,Sensors 13 (2013) 3242–3251.

[2] G. Emiliyanov, J.B. Jensen, O. Bang, P.E. Hoiby, L.H. Pedersen, E.M. Kjær, L.Lindvold, Localized biosensing with Topas microstructured polymer opticalfiber, Opt. Lett. 32 (2007) 460–462.

[3] S. Roy, C.Y. Yue, Y.C. Lam, Z.Y. Wang, H. Hu, Surface analysis, hydrophilicenhancement, ageing behavior, and flow in plasma modified cyclic olefincopolymer (COC)-based microfluidic devices, Sens. Actuators B 150 (2010)537–549.

[4] W.I. Jang, K.H. Chung, H.B. Pyo, S.H. Park, Self-operated blood plasmaseparation using micropump in polymer-based microfluidic device, in: J.-C.Chiao, A.S. Dzurak, C. Jagadish, D.V. Thiel (Eds.), Micro- and Nanotechnology:Materials, Processes, Packaging, and Systems III, SPIE Proceedings, vol. 6415,2006, p. 641511. doi: 10.1117/12.693475.

[5] C. Li, Y. Yang, H.G. Craighead, K.H. Lee, Isoelectric focusing in cyclic olefincopolymer microfluidic channels coated by polyacrylamide using a UVphotografting method, Electrophoresis 26 (2005) 1800–1806.

[6] G.A. Diaz-Quijada, R. Peytavi, A. Nantel, E. Roy, M.G. Bergeron, M.M. Dumoulin,T. Veres, Surface modification of thermoplastics: towards the plastic biochipfor high throughput screening devices, Lab Chip 7 (2007) 856–862.

[7] R.R. Lamonte, D. McNally, Cyclic olefin copolymers, Adv. Mater. Processes 159(2001) 33–36.

[8] Z.-R. Lu, In vivo imaging of dosage forms, Drugs Pharm. Sci. 188 (2009) 181–201.

[9] I.P. Johnson, W. Yuan, A. Stefani, K. Nielsen, H.K. Rasmussen, L. Khan, D.J. Webb,K. Kalli, O. Bang, Optical fibre Bragg grating recorded in TOPAS cyclic olefincopolymer, Electron. Lett. 47 (2011) 271–272.

[10] W. Yuan, L. Khan, D.J. Webb, K. Kalli, H.K. Rasmussen, A. Stefani, O. Bang,Humidity insensitive TOPAS polymer fiber Bragg grating sensor, Opt. Express19 (2011) 19731–19739.

[11] K. Nielsen, H.K. Rasmussen, A.J. Adam, P.C. Planken, O. Bang, P.U. Jepsen,Bendable, low-loss Topas fibers for the terahertz frequency range, Opt. Express17 (2009) 8592–8601.

[12] W.D. Niles, P.J. Coassin, Cyclic olefin polymers: innovative materials for high-density multiwell plates, Assay Drug Dev. Technol. 6 (2008) 577–590.

[13] J.Y. Shin, J.Y. Park, C. Liu, J. He, S.C. Kim, Chemical structure and physicalproperties of cyclic olefin copolymers, Pure Appl. Chem. 77 (2005) 801–814.

[14] S.-J. Hwang, M.-C. Tseng, J.-R. Shu, H.H. Yu, Surface modification of cyclic olefincopolymer substrate by oxygen plasma treatment, Surf. Coatings Technol. 202(2008) 3669–3674.

[15] D. Nikolova, E. Dayss, G. Leps, A. Wutzler, Surface modification of cycloolefiniccopolymers for optimization of the adhesion to metals, Surf. Interface Anal. 36(2004) 689–693.

[16] S. Roy, C.Y. Yue, Surface modification of COC microfluidic devices: acomparative study of nitrogen plasma treatment and its advantages overargon and oxygen plasma treatments, Plasma Processes Polym. 8 (2011) 432–443.

[17] H. Steffen, K. Schröder, B. Busse, A. Ohl, K.D. Weltmann, Functionalization ofCOC surfaces by microwave plasmas, Plasma Processes Polym. 4 (2007) S392–S396.

[18] H. Barakat, C. Aymes-Chodur, J. Saunier, N. Yagoubi, Effect of electron beamradio sterilization on cyclic olefin copolymers used as pharmaceutical storagematerials, Radiat. Phys. Chem. 84 (2013) 223–231.

[19] C.K. Dixit, S.K. Vashist, F.T. O’Neill, B. O’Reilly, B.D. MacCraith, R. O’Kennedy,Development of a high sensitivity rapid sandwich ELISA procedure and itscomparison with the conventional approach, Anal. Chem. 82 (2010) 7049–7052.

[20] C.K. Dixit, S.K. Vashist, B.D. MacCraith, R. O’Kennedy, Multisubstrate-compatible ELISA procedures for rapid and high-sensitivity immunoassays,Nat. Protoc. 6 (2011) 439–445.

[21] S. Laib, B.D. MacCraith, Immobilization of biomolecules on cycloolefin polymersupports, Anal. Chem. 79 (2007) 6264–6270.

[22] Y. Lu, J.S. Valentine, Engineering metal-binding sites in proteins, Curr. Opin.Struct. Biol. 7 (1997) 495–500.

[23] E.K.U. Larsen, N.B. Larsen, One-step polymer surface modification forminimizing drug, protein, and DNA adsorption in microanalytical systems,Lab Chip 13 (2013) 669–675.

[24] B.W. Muir, M.C. Barden, S.P. Collett, A.-D. Gorse, R. Monteiro, L. Yang, N.A.McDougall, S. Gould, N.J. Maeji, High-throughput optimization of surfaces forantibody immobilization using metal complexes, Anal. Biochem. 363 (2007)97–107.

[25] E. Chung, S.J. Cooper, B.T. Ohse, Y. Gao, D. Jennins, M.M. Koudijs, L. Yang, T.Ling, P. Vukovic, A. Wong, N.J. Maeji, Simple methods for forming proteinmonolayers, Lab-on-a-Chip European Congress, Barcelona, Spain, 2013.

[26] N. Abernethy, E. Chung, B.T. Fontanelle, Y. Gao, D. Jennins, M.M. Koudijs, D.Lim, L. Yang, T. Ling, P. Vukovic, A. Wong, N.J. Maeji, Metal polymers, a glue toimmobilise proteins onto synthetic surfaces, in: 44th Oak Ridge Conference:Emerging Technologies for 21st Century Diagnostics, San Jose, CA, 2012.

[27] B.W. Muir, M.C. Barden, D.B. Rylatt, N.J. Maeji, C.J. Hillyard, A.-D.J.-P. Gorse, R.L.Monteiro, Use of metal complexes, Patent application 20120214973, Bio-Layer, 2012.

[28] G. Beamson, D. Briggs, High Resolution XPS of Organic Polymers: The ScientaESCA300 Database, Wiley, Chichester, UK, 1992.

[29] C. Boiziau, G. Lecayon, Adhesion of polymers to metals: a review of the resultsobtained studying a model system, Surf. Interface Anal. 12 (1988) 475–485.

[30] J.E. Hale, Irreversible, oriented immobilization of antibodies to cobalt–iminodiacetate resin for use as immunoaffinity media, Anal. Biochem. 231(1995) 46–49.

[31] K.L. Mittal, The role of the interface in adhesion phenomena, Polymer Eng. Sci.17 (1977) 467–473.

[32] T. Rische, A.J. Waddon, L.C. Dickinson, W.J. MacKnight, Microstructure andmorphology of cycloolefin copolymers, Macromolecules 31 (1998) 1871–1874.

[33] F. Faupel, V. Zaporojtchenko, T. Strunskus, J. Erichsen, K. Dolgner, A. Thran, M.Kiene, Fundamental aspects of polymer metallization, in: E. Sacher (Ed.),Metallization of Polymers 2, Kluwer Academic/Plenum, New York, 2002.

[34] V. Mailaender, K. Landfester, Interaction of nanoparticles with cells,Biomacromolecules 10 (2009) 2379–2400.