Research Article

Received: 15 December 2009, Revised: 6 August 2010, Accepted: 23 August 2010, Published online in Wiley Online Library: 2010

(wileyonlinelibrary.com) DOI:10.1002/jmr.1095

Solid-phase preparation of protein complexesz

Paolo Pengoa*, Gianluca Veggiania, Kwanchai Rattanamaneea,by,Andrea Gallottaa, Luca Beneducea and Giorgio Fassinaa

Protein–protein conjugation is usually achieved by s

J. Mol. Rec

olution phase methods requiring concentrated protein solutionand post-synthetic purification steps. In this report we describe a novel continuous-flow solid-phase approachenabling the assembly of protein complexes minimizing the amount of material needed and allowing the repeateduse of the same solid phase. The method exploits an immunoaffinity matrix as solid support; the matrix reversiblybinds the first of the complex components while the other components are sequentially introduced, thus allowing thecomplex to grow while immobilized. The tethering technique employed relies on the use of the very mild syntheticconditions and fast association rates allowed by the avidin–biotin system. At the end of the assembly, the immobilizedcomplexes can be removed from the solid support and recovered by lowering the pH of the medium. Under theconditions used for the sequential complexation and recovery, the solid phase was not damaged or irreversiblymodified and could be reused without loss of binding capacity. The method was specifically designed to prepareprotein complexes to be used in immunometric methods of analysis, where the immunoreactivity of each componentneeds to be preserved. The approach was successfully exploited for the preparation of two different immunoaffinityreagents with immunoreactivity mimicking native squamous cell carcinoma antigen-immunoglobulin M (SCCA-IgM)and alphafetoprotein-immunoglobulin M (AFP-IgM) immune complexes, which were characterized by dedicatedsandwich enzyme-linked immunosorbent assay (ELISA) and immunoblot. Besides the specific application described inthe paper, the method is sufficiently general to be used for the preparation of a broad range of protein assemblies.Copyright � 2010 John Wiley & Sons, Ltd.

Keywords: protein complexes; solid-phase complex preparation; immune complexes

* Correspondence to: P. Pengo, Xeptagen SpA, Via delle Industrie 9, Margher-a-Venezia I-30175, Italy.E-mail: pengo@xeptagen.com

a P. Pengo, G. Veggiani, K. Rattanamanee, A. Gallotta, L. Beneduce, G. Fassina

Xeptagen SpA, Via delle Industrie 9, Marghera-Venezia I-30175, Italy

b K. Rattanamanee

Department of Pharmacy Practice, Faculty of Pharmaceutical Sciences,

Naresuan University, Pitsanulok, Thailand

y Supported by Naresuan and Chulalongkorn Universities within the ICS-UNIDOcollaborative framework of activities.

z This article is published in Journal of Molecular Recognition as a special issueon Affinity 2009, edited by Gideon Fleminger, Tel-Aviv University, Tel-Aviv, Israeland George Ehrlich, Hoffmann-La Roche, Nutley, NJ, USA.

Abbreviations: SCCA, Squamous Cell Carcinoma Antigen; AFP, alphafetopro-

tein; IgM, Immunoglobulin M; ELISA, Enzyme-linked Immunosorbent Assay; BSA,

Bovine Serum Albumin; IgG, immunoglobulin G; NHS, N-hydroxysuccinimide; HRP,

Horseradish Peroxidase; ABTS, 2,20-azino-bis(3-ethylbenzthiazoline-6-sulphonic

acid); SDS–PAGE, Sodium Dodecyl Sulphate Polyacrylamide Gel Electrophoresis. 5

INTRODUCTION

Chemical conjugation is the simplest approach to protein taggingor to the implementation of different biochemical properties inthe same protein complex. Though rather simplistic, thisapproach has allowed the development of the large range ofconjugates routinely used by virtually any immunometricmethods of analysis. Conjugation usually relies on the formationof covalent bonds between the different partners by exploitinghomo- or hetero-bifunctional reagents reacting with aminoacidside chains, usually lysine, cysteine or carboxylic acids con-veniently found on the protein surface. Alternatively, a reactivegroup, such as an aldehyde, can be introduced by oxidation ofthe sugar moieties at the protein glycosylation sites and thenmade to react with the free lysine side chains of another proteinspecies forming an imine or a secondary amine if a suitablereducing agent is present. These conjugation procedures are welldescribed in the literature and are routinely used in the practice(Wong, 1991); other approaches, such as the click-chemistryand the Staudinger ligation, are increasingly being pursued(Brennan et al., 2006; Gauthier and Klok, 2008; Best, 2009). Despitetheir widespread use, standard conjugation protocols presentmajor drawbacks related to the need of concentrated proteinsolutions, usually in the range of 1–10mg/ml or more, (Wong,1991) to ensure an efficient conjugation and the need ofpost-synthetic purification steps to remove large polymericspecies or unreactedmaterials. This feature limits the applicabilityof the approach to cheap proteins stable in concentratedsolutions and readily available in large amounts. Conversely,the conjugation of proteins available in small amounts or prone

ognit. 2010; 23: 551–558 Copyright � 20

to the formation of aggregates in concentrated solutions wouldbe difficult to achieve by this method. An approach which may, inprinciple, overcome some of these limitations is the developmentof efficient solid-phase protocols allowing the preservation ofprotein structure and functional activity reducing the need ofpost-synthetic purification processes. Some reports in theliterature have described the effort to devise solid-phaseapproaches to protein conjugation and functionalization indi-cating a growing attention towards these alternative protocols.

10 John Wiley & Sons, Ltd.

51

P. PENGO ET AL.

552

So far, all of these methods have displayed both advantages andlimitations related to the specific chemical approach being used.Among the most interesting methods, Russell et al. (2002; 2004)proposed a batch-mode solid-phase protein conjugation pro-cedure, exploiting the thiol-maleimide chemistry. However, thisbatch-wise method exploits the same chemistry for bothconjugation and protein immobilization and does not allowthe repeated use of the solid phase. Another approach to theimplementation of solid-phase protocols in protein conjugationhas been reported by Hu and coworkers (Hu et al., 2003; Hu andSu, 2003); the authors used an ion exchangematrix to temporarilyimmobilize bovine haemoglobin aiming to obtain dimers of theprotein; in this case glutaraldehyde was used as linker. The samegroup has also reported a method for haemoglobin pegylationrequiring immobilization of the protein on ion exchange columns(Suo et al., 2009). Here, we present a different solid-phaseapproach for the preparation of protein complexes where aspecial attention is paid to preserve the immune reactivity of thesingle components. The method is entirely based on the use ofstrong and specific non-covalent interactions between differentprotein building blocks in addition to the biotin–avidintechnology. The development of the reported method arosefrom the need to obtain immunoreactive molecules similar tonatural antigen-immunoglobulin M (IgM) immune complexes.The design of the approach was guided by practical consider-ations related to the limitations of the costs associated to thepurification of the final product.

MATERIALS AND METHODS

Materials

Bovine serum albumin (BSA), inorganic and organic materialsused for buffers preparation were obtained from standardsuppliers, the reagent grade water was obtained by using the Elix3 and the milli-Q Synthesis A10 systems (Millipore). All the buffersused were freshly prepared and filtered through 0.45mmnalgenefilters (Millipore) prior to use. Protein purifications by affinitychromatography were run on the AKTA Purifier system(Amersham Bioscience). Analytical gel filtration chromatographywere run using either a Biosep-SEC-S4000 column (Phenomenex)assembled on a waters 600 HPLC system equipped with an in-linedegasser and dual wavelength detector. sodium dodecylsulphate polyacrylamide gel electrophoresis (SDS–PAGE) andimmunoblot analysis were performed using the verticalelectrophoresis device Mini PROTEAN 3 cell (Bio-Rad) inconjunction with the power supply EPS 301 (AmershamBioscience). The molecular marker PageRulerTM Plus prestainedprotein ladder were obtained from Fermentas. The immunoblotanalysis were carried out using the Trans-Blot1 Transfer MediumPure Nitrocellulose Membrane (0.45mm) from Bio-Rad.Centrifugal filtrations were performed using either the

Amicon1 Ultra 10 000 MWCO or the microcon filtration devices(Millipore). Enzyme-linked immunosorbent assay (ELISA) assayswere run using high binding 96 wells plates (Greiner Bio-One).The washing procedures required in ELISA assays wereautomatically performed by using the Biotrak II Plate Washer(Amersham Biosciences). The ELISA assays required the use of thechromogenic substrate ABTS (2,20-azino-bis(3-ethylbenzthiazoline-6-sulphonic acid)) (Sigma Aldrich), the colour development wasmonitored at 405 nm using the Viktor3 TM 1420 MultilabelCounter plate reader (Perkin Elmer) or the Biotrak II Plate Reader

View this article online at wileyonlinelibrary.com Copyright �

(Amersham Biosciences). Rabbit polyclonal anti-squamous cellcarcinoma antigen (anti-SCCA) antibody is a product of XeptagenSpA, goat polyclonal anti-human IgM complexed to horseradishperoxidase (HRP) was obtained from Sigma-Aldrich. The totalprotein assay was performed in a microwell format according tothe Bradford method against bovine serum albumin or humanimmunoglobulin G (IgG) used as standards, optical densities ofthe solutions were read at 620 nm using the Biotrak II PlateReader (Amersham Biosciences).

Purification of IgM from human serum

Purification of human IgM from human serum was accomplishedby affinity chromatography on goat-anti-human IgM (m-chainspecific) agarose (Sigma-Aldrich A9935). The affinity matrix wasloaded onto a glass column (0.5 cm ID, 2.0ml bed volume) andpacked at a flow rate of 1.5ml/min using PBS as mobile phase.The column was conditioned at a flow rate of 0.5ml/min withPBS pH 7.2 using at least five column volumes of buffer prior tosample loading. A sample of diluted human serum obtaineddiluting 3.0ml of serum to 12ml with PBS pH 7.2 were loadedonto the column at a flow rate of 0.5ml/min; the chromato-graphic run was monitored at 220 and 280 nm. The unretainedmaterial (unbound fraction) was collected and the column waswashed with reagent grade water at 1.5ml/min flow rate until theconductivity of the flowtrought dropped to zero. The boundfraction was then eluted using glycine/HCl buffer 0.1M at pH 2.5,the flow rate used during elution was 1.5ml/min. The collectedsolution was neutralized adding a saturated solution of NaHCO3,100ml of saturated bicarbonate solution were used for eachmillilitre of collected solution. In a typical experiment, the boundfractions from three different chromatographic runs were pooledtogether and the buffer was exchanged by centrifugal filtrationby using the Amicon1 Ultra 10 000 MWCO centrifugal filtrationdevices (Millipore). To completely remove any traces of glycine,the concentrate was treated with PBS (10� 5ml) and eventuallyreduced to about 0.5ml; the protein concentration wasdetermined by the Bradford method. The solution was dividedinto 200ml aliquots and stored at �208C until further use.

Human IgM biotinylation

The concentration of the IgM solution resulting from thepurification process was adjusted to 1260mg/ml adding PBS; asolution of bicarbonate buffer 0.1M pH 8.5 containing NaCl 0.5Mwas added to the solution in a 1:1.5 v/v ratio. The mixture wastreated with a freshly prepared solution of N-hydroxysuccinimideactivated biotin (SIGMA B3295-50MG) 1.0mg/ml in DMSO in a1:12 v/v ratio, the solution was stirred for 4 h and afterwardsdialysed overnight against 1 L of PBS. The dialysis process wasrepeated two more times against 1 L of PBS. The protein solutionwas collected and assayed for total protein content according tothe Bradford method.

Biotinylation of SCCA

Squamous Cell Carcinoma Antigen was obtained from cell lysate aspreviously described. (De Falco et al., 2001) 500ml of SCCA solutionin morpholinoethansulphonic acid (MES) was subjected tocentrifugal filtration in order to change the milieu to phosphatebuffered saline (PBS) using the microcon centrifugal filtrationdevices as recommended by the manufacturer. The resultingsolution was diluted to 600ml with PBS. An equal volume of

2010 John Wiley & Sons, Ltd. J. Mol. Recognit. 2010; 23: 551–558

SOLID-PHASE PREPARATION OF PROTEIN COMPLEXES

5

bicarbonate buffer 0.1M pH 8.5 containing NaCl 0.5M was added,to the solution were added 60ml of a solution of NHS activatedbiotin (SIGMA B3295-50MG) 1.0mg/ml in DMSO, and then thesolution was stirred for 4 h and afterwards dialysed overnightagainst 1 L of PBS. The protein solution was collected and assayedfor total protein content according to the Bradford method.

Biotinylation of alphafetoprotein (AFP)

500ml of AFP solution in PBS was subjected to centrifugalfiltration in order to remove any free amines present in themedium using the microcon centrifugal filtration devices asrecommended by the manufacturer. The resulting solution wasdiluted to 600ml with PBS. An equal volume of bicarbonate buffer0.1M pH 8.5 containing NaCl 0.5M was added, to the solutionwere added 60ml of a solution of N-hydroxysuccinimide activatedbiotin (SIGMA B3295-50MG) 1.0mg/ml in DMSO, and then thesolution was stirred for 4 h and afterwards dialysed overnightagainst 1 L of PBS pH 7.2. The protein solution was collectedand assayed for total protein content according to the Bradfordmethod.

ELISA assays for the estimation of the complex reactivity

SCCA-IgM complex

High binding 96-wells microplates were sensitized with poly-clonal rabbit anti-SCCA at the concentration of 10mg/ml, 100ml/well. The coating procedure was performed overnight at 48C in ahumid closed chamber. The solution was removed and the plateswashed with PBS containing 0.05% Tween20 (3� 300ml/well).The wells were filled with 300ml of PBS containing 3% bovineserum albumin to saturate the residual reactive sites of the plasticplate; the plate was allowed to stand at room temperature for 2 h.The samples were loaded on the plate by performing a two-foldserial dilution starting from the concentration of 100mg/ml,PBS containing 1% BSA and 0.05% Tween20 was used asdilution buffer. The final volume was 100ml/well, the sampleswere incubated for 1 h at room temperature. The solutions wereremoved by suction and the plate was washed with PBScontaining 0.05% Tween20 (6� 300ml/well). A solution ofgoat-anti-human IgM complexed to HRP at the 1:1000 dilution(100ml/well) was added to the wells and incubated for 1 h atroom temperature. The solution of secondary antibody wasremoved by suction and the plate washed with a PBS containing0.05% Tween20 (6� 300ml). A solution containing ABTS 0.2mg/ml and 5mM hydrogen peroxide in phosphate-citrate buffer pH5.0 was added to each well (150ml/well), the enzymatic reactionwas allowed to proceed for 20min at 378C in the dark.Absorbance was read at 405 nm.

AFP-IgM complex

High binding 96-wells microplates were sensitized with poly-clonal rabbit anti-AFP (DAKOA008) at the concentration of 10mg/ml, 100ml/well. The coating procedure was performed overnightat 48C in a humid closed chamber. The solution was removedand the plates washed with PBS containing 0.05% Tween20(3� 300ml/well). The wells were filled with 300ml of PBScontaining 3% bovine serum albumin to saturate the residualreactive sites of the plastic plate, the plate was allowed to stand atroom temperature for 2 h. The samples were loaded on the plateby performing a two-fold serial dilution starting from the

J. Mol. Recognit. 2010; 23: 551–558 Copyright � 2010 John Wiley &

concentration of 100mg/ml, PBS containing 1% BSA and 0.05%Tween20 was used as dilution buffer. The final volume was100ml/well, the samples were incubated for 1 h at roomtemperature. The solutions were removed by suction and theplate was washed with PBS containing 0.05% Tween20(6� 300ml/well). A solution of goat-anti-human IgM complexedto HRP at the 1:1000 (100ml/well) was added to the wells andincubated for 1 h at room temperature. The solution of secondaryantibody was removed by suction and the plate washed with aPBS containing 0.05% Tween20 (6� 300ml). A solution contain-ing ABTS 0.2mg/ml and 5mM hydrogen peroxide in phospha-te-citrate buffer pH 5.0 was added to each well (150ml/well), theenzymatic reaction was allowed to proceed for 20min at 378C inthe dark. Absorbance was read at 405 nm.

Solid-phase assembly of the SCCA-IgM complex

The affinity matrix featuring covalently linked goat antibodiesraised against human IgM was packed in a glass column (0.5 cmID, 5 cm height) and conditioned at 0.5ml/min flow rate usingPBS asmobile phase. The solution of biotinylated IgM (250mg/ml)was injected in 250ml aliquots (6� 250ml), the process wasmonitored at 280 nm and the unbound material was collected.After the addition of the IgM solution was complete, the columnwas briefly washed with PBS. A solution of avidin (600mg/ml) wasinjected in small aliquots (4� 250ml), monitoring the process at280 nm. After the addition of avidin, the columnwas washed withPBS. A solution of biotinylated SCCA (200mg/ml) was added insmall portions (4� 250ml) monitoring the injection processat 280 nm, at the end of the addition, the column wasbriefly washed with PBS and afterwards a solution ofN-hydroxysuccinimide activated biotin previously reacted withlysine (biotin concentration 100mg/ml) was added (3� 250ml),the immobilized complex was washed with PBS pH 7.2. Thecomplex was recovered by changing the mobile phase to0.1 glycine buffer pH 2.5. The complex eluted as a relativelynarrow peak, the eluted solution was neutralized by the additionof a saturated sodium bicarbonate solution.

Solid-phase assembly of the AFP-IgM complex

The affinity matrix featuring covalently linked goat antibodiesraised against human IgM was packed in a glass column (0.5 cmID, 5 cm height) and conditioned at 0.5ml/min flow rate usingPBS as mobile phase. The IgM solution (250mg/ml) was injectedin 250ml aliquots (5� 250ml), the process was monitored at280 nm and the unbound material was collected. After theaddition of the IgM solution was complete, the column wasbriefly washed with PBS. A solution of avidin (600mg/ml) wasinjected in small aliquots (4� 250ml), monitoring the process at280 nm. After the addition of avidin, the columnwas washed withPBS. A solution of biotinylated AFP (100mg/ml) was added insmall portions (2� 250ml) monitoring the effluent at 280 nm, atthe end of the addition, the column was briefly washed with PBSand afterwards a solution of N-hydroxysuccinimide activatedbiotin previously reacted with lysine (biotin concentration100mg/ml) was added (3� 250ml) and the immobilized complexwas washed with PBS. The complex was recovered by changingthe mobile phase to 0.1 citrate buffer pH 2.5. The complex elutedas a relatively narrow peak and the eluted solution wasneutralized by the addition of a saturated sodium bicarbonatesolution.

Sons, Ltd. View this article online at wileyonlinelibrary.com

53

P. PENGO ET AL.

554

SDS–PAGE and immunoblot of SCCA-IgM complex

SDS–PAGE of SCCA-IgM complex was performed according to theLaemmli procedure on a discontinuous gel system consisting of a10% resolving gel and 4% stacking gel. The protein samples(5mg) were treated with 1:6 (v/v) of sample buffer containing0.5M Tris-Cl pH 6.8, 0.35M SDS, glycerol 30% (v/v), 0.175mMBromophenol Blue and 6mMDTT. 16ml of a solution of SCCA-IgMcomplex were treated with 1:3.2 (v/v) of a modified sample buffercontaining 0.5M Tris-Cl pH 6.8, 0.83M SDS, glycerol 30% (v/v),0.29mM Bromophenol Blue and 10mM DTT. Prior to the loadingin the gel, samples were heated for 5min at 1008C and brieflycentrifuged. The separation was carried out at the constantvoltage of 250 V until the tracking dye exited the gel bottom.The total run time was 45min. After electrophoresis stacking gelwas discarded and the transfer of polyacrylamide gel resolvedproteins to the nitrocellulose membrane was carried out asdescribed by Towbin et al. (1979) at the constant voltage of 100 V;the total running time was 1 h. After the transfer, the nitro-cellulose membrane was dried overnight at the room tempera-ture. The dried membrane was immersed for 1 h in 10ml ofblocking buffer (PBS containing 0.05% Tween20 and 5% skimmedmilk). The membrane was incubated with a solution 10mg/ml ofoligoclonal rabbit IgG anti-SCCA (Xeptagen SpA) for 1 h at theroom temperature and then washed for 15min four times withPBS containing 0.05% Tween20. A solution of goat-anti-rabbit IgGcomplexed to the HRP at the 1:1000 (10ml) was added to themembrane as secondary antibody and incubated for 1 h at theroom temperature. The membrane was washed for 15min fourtimes with PBS containing 0.05% Tween20 and then immersed in6ml of a chemiluminescent solution (Immobilon Western substrateHRP, Millipore) for 1min. After the detection of the reactive bands,the membrane was washed, blocked as described above and wasincubated for 1 h at the room temperature with a solution ofgoat-anti-human IgM complexed to the HRP at the 1:1000 (10ml).After the antibody incubation, the membrane was washed for15min four times with PBS containing 0.05% Tween20 and thentreated with the chemiluminescent solution as described above.The reactive bands were scanned using the Image Station 440 CF(Kodak) and analysed with the Kodak 1D Software.

Gel filtration chromatography

A preliminary mass analysis of the oligomeric complex wascarried out by gel-filtration, using a BioSep SEC-s4000 column(Phenomenex), the column was calibrated by using a set ofproteins of known molecular weight, spanning the range900–17 KDa (human IgM, human IgG, BSA, horse myoglobin).A total volume of 200ml of complex solution (50mg/ml) wasinjected at 1.0ml/min flow rate using PBS pH 7.2 as mobile phase.The obtained chromatogram (Figure 5) displays a single peakeluting at 6.6min. The column effluent was collected in 500mlfractions, and their immunoreactivity was assayed by dedicatedELISA. A significant immune reactivity was detected in only thosefractions eluting at 6.6min (Kav 0.08), thus confirming the natureof the peak.

RESULTS AND DISCUSSION

Principle and rationale of the method

A solid-phase approach to the preparation of protein complexesby exploiting solely non-covalent interactions is conceivable but

View this article online at wileyonlinelibrary.com Copyright �

it would be effective only if the non-covalent interactions holdingtogether the assembled complex are stronger than thoseexploited to immobilize the first building block to the solidsupport. In fact, only in these conditions a certain degree ofselectivity allowing the recovery of the intact complex could beattained. From this starting point, we selected the antige-n–antibody interaction to immobilize the first component of thecomplex on a solid support and the avidin–biotin interaction toassemble the complex; the differences in the associationconstants for the two processes should be sufficient to achievea reasonable degree of selectivity. Of course, this approachrequires the use of biotinylated protein building blocks and theuse of avidin as ‘linker’. The tremendous advantage of usingavidin as ‘linker’ is its high specificity and its high bindingconstant towards biotinylated molecules which in turn allows theuse of diluted solutions of biotinylated species. The outlinedmethod requires to use a solid phase consisting of animmunoaffinity matrix featuring covalently linked antibodiesagainst the first of the complex components. The firstbiotinylated protein building block, once immobilized on thesolid support by the antigen–antibody interactions, would serveas a handle to build the assembled species by sequential additionof avidin and other different biotinylated molecules. Of coursebecause of the heterogeneity in protein biotinylation, a certaindegree of heterogeneity could also be expected in theassembling process, leading to different reactivities of differentprotein complexes. This is indeed the case for all the conjugationprocesses not relying on protein engineering which, instead,ensure the precise identification of the reaction sites and anunequivocal definition of the complex architecture.The working principle of the method for the assembling of just

two different biotinylated protein building blocks (BB1; BB2) isgraphically depicted in Figure 1. In Step 1 the first biotinylatedbuilding block, BB1, is captured by the antibody specific to itcovalently bound to the solid support; the species not recognizedby the antibody are washed away. In Step 2, an excess of avidin isadded to the resin-bound BB1 forming a complex with theavailable biotin moieties on its surface. The other binding sites ofavidin far apart from the BB1 surface will remain free and will bethus available for further binding with other biotinylated species.During this process, some of the available biotin moieties on BB1will remain unreacted either because of steric hindrance due toavidin molecules bound to nearby biotins, or because of theirlimited accessibility on the side of the BB1 molecule facing theantibody bound to the bead surface. The addition of a differentbiotinylated building block, BB2, in Step 3 allows the formation ofthe immobilized complex by reaction with the free biotin bindingsites on the avidin molecules bound to the immobilized BB1. InStep 4, the non-reacted biotin binding sites on avidin stillavailable for binding but not reacted to BB2 must be saturated inorder to prevent oligomerization after cleavage of the complexfrom the solid support; this can easily be achieved by adding anexcess of free biotin providing an efficient capping methodology.At the end of the assembling process, the complex can beremoved from the solid support by changing the pH, thusbreaking the antibody–protein interaction responsible for itsimmobilization (Step 5). The pH change does not affect thebiotin–avidin interaction allowing the recovery of the intactcomplex. After cleavage the column could be re-used withoutloss of binding capacity. From a general point of view, whatmolecule should be used as first building block and what shouldbe used as the second is largely a matter of choice and may vary

2010 John Wiley & Sons, Ltd. J. Mol. Recognit. 2010; 23: 551–558

Figure 1. Schematic representation of the continuous-flow synthetic process. In Step 1, the first biotinylated protein building block is immobilized on

the beads carrying its specific antibody, all the other protein species present are washed away. In Step 2, an excess of avidin is added in order to react withthe accessible biotin moieties. Some of the biotins will not react because of steric hindrance or limited accessibility. In Step 3, the second biotinylated

building block is added reacting with the free biotin binding sites on avidin. To provide an efficient capping methodology, in Step 4 an excess of free

biotin is added to saturate any remaining binding sites on the avidin molecules. In Step 5, a pH change of the medium favours the ‘cleavage’ of the

complex from the resin weakening the interaction between the BB1 and the antibody specific to it. The resin can be washed and reused for anothersynthesis.

SOLID-PHASE PREPARATION OF PROTEIN COMPLEXES

5

according to the intended use of the complex. As mentionedabove, the mandatory requisites for the method were to be costeffective and require no purification of the final product. Clearlythese considerations strongly restrict the number of possibleapproaches to those optimizing the use of the most expensivespecies. Therefore, aiming to obtain molecular complexes with animmune reactivity similar to that of natural immune complexesformed by tumour antigens and IgM, the use of the antigen wasrequired to be as limited as possible. These considerationsprompted us to devise the presentedmethod, which required theimmobilization of the IgM as the first building block andproceeding to the formation of the complex by addition of avidinand antigen. This approach has the advantage of using as the firstbuilding block a cheap species, which does not interfere in theassay in case of incomplete assembling of the complex.On the contrary, in case of incomplete reaction, the

immobilization of the antigen as the first building block wouldhave led to some free antigen in the same solution containing thecomplex, which could compete for the latter in the immunoassay,thus lowering the observed reactivity. In this case a post-syntheticpurification step would have been necessary. As a concludingremark, it is worth stressing that the solid-phase complexpreparation described above allows a certain degree of controlon the topology of the complex, maintaining a high surface tovolume ratio. An alternative solution phase method using thesame building blocks is conceivable but in this case the topologyof the complex would be much difficult to control. In closerdetails, a solution phase method would produce a polymericspecies with many molecules buried inside the complex and thusinaccessible for recognition by the antibodies used in ELISAassays. In this case only those molecules on the surface would beavailable for recognition, while the rest could just serve as

J. Mol. Recognit. 2010; 23: 551–558 Copyright � 2010 John Wiley &

scaffold. This approach is clearly less efficient than thesolid-phase method in terms of use of materials.

Solid-phase assembly of the SCCA/IgM andAFP/IgM complexes

The proof of principle of the method outlined above wasobtained by preparing two different complexes containing IgMand tumour associated antigens, the latter represents an exampleof relatively expensive proteins usually available in limitedamounts, while the former represents an example of a proteinprone to form aggregates. In closed details, the first complexprepared comprised human IgM (BB1) and the Squamous CellCarcinoma Antigen (SCCA) (BB2), while the second complexcomprised human IgM (BB1) and alpha-fetoprotein (AFP) (BB2).The reason for the preparation of these complexes resides in theimportant role played by SCCA-IgM and AFP-IgM immunecomplexes in the diagnosis of hepatocellular carcinoma and inthe urgent industrial need of synthetic complexes displayingimmunoreactivity mimicking that of the natural cognate immunecomplexes. The choice of using IgM instead of the tumourantigen as the first building block arose from a practicalconsideration. In case of incomplete assembly of the complex,the free antigen released upon cleavage would have interfered inthe ELISA assay. In this case, a post-synthetic purification stepwould have been necessary. The experimental design requiredthe use of an anti-IgM antibody covalently linked to a solid phaseto temporarily immobilize the biotinylated IgM; the resin waspacked in a standard affinity column mounted on a standardFPLC system. The column was conditioned in PBS at a flow rate of0.5ml/min, the addition of the biotinylated protein was achievedby repeatedly injecting 250ml of diluted protein solution, the

Sons, Ltd. View this article online at wileyonlinelibrary.com

55

Figure 2. Chromatographic traces recorded during the continuous-flow synthesis of the SCCA-IgM (Panel A) and AFP-IgM complex (Panel B).The chromatograms were recorded at 280 nm, dashed lines represents the injections of the various species.

Figure 3. Immunoreactivity of the SCCA-IgM (squares, solid squares

first synthesis, open squares second synthesis, the concentration of the

complex was adjusted to 100mg/ml in both cases) and AFP-IgM (circles,open circles first synthesis, solid circles second synthesis, the concen-

tration of the complexes was adjusted at 100mg/ml in both cases)

complexes. Two mixtures comprising an equimolar amounts of biotiny-

lated IgM and SCCA (diamonds) and equimolar amounts of biotinylatedIgM and AFP with a total protein concentration of 100mg/ml but in

absence of avidin (cross) were used as negative control.

P. PENGO ET AL.

556

injection process and the flow-through of unretained materialwere monitored at 280 nm, the resulting ‘chromatographic’ traceis displayed in Figure 2 (preparation of SCCA-IgM complexdisplayed in panel A, preparation of AFP-IgM complex displayedin panel B).The immobilized IgM were then reacted with an excess of

avidin thus saturating the accessible biotin moieties on the IgMsurface (Figure 1, Step 2). The reaction occurred by injecting intothe column a solution of avidin in PBS; at this stage, the avidinmolecules display some of their free binding sites available forbinding with other biotinylated species. The addition of thebiotinylated SCCA or biotinylated AFP results in the formation ofthe complex. The saturation of the remaining binding sites ofavidin eventually present could be achieved by injecting into thecolumn an excess of biotin. This step is necessary in order to avoidthe polymerization of the complexes once they are removed fromthe solid phase. By using this procedure, multiple conjugationcannot be ruled out. The release of the complexes was achievedby changing the pH of the mobile phase to 2.5, thus weakeningthe antibody–IgM interaction without affecting the avidin–biotinbinding. In a typical experiment, the recovery of the complexin terms of total protein content was about 70% of the totalprotein loaded on the column. The procedures used for thebiotinylation of the building blocks were optimized in order toreduce the number of biotinmoieties introduced in each buildingblock. This was necessary in order to limit the number of avidinmolecules involved in the formation of the complex. A harsherbiotinylation procedure generally resulted in a diminishedreactivity of the complex, probably because of the sterichindrance caused by the presence of a large number of avidinmolecules.

Characterization of the complexes

The collected protein complexes were analysed to ascertain theexistence of connectivity between IgM and SCCA or IgM and AFPwithin the same molecular species as a direct indication ofthe successful assembling. To this aim dedicated ELISA assayswere developed by using anti-SCCA or anti-AFP antibodies ascapturing phases and the use of an anti-IgM antibody complexedto HRP as tracing species; in these experiments ABTS was used as

View this article online at wileyonlinelibrary.com Copyright �

chromogenic substrate in the presence of hydrogen peroxide.The results of the ELISA analysis, reported in Figure 3, indicatethat SCCA and IgM were present within the first complex and AFPand IgM were present in the second. In both cases a mixture ofuncomplexed SCCA and IgM or AFP and IgM did not displaysignificant reactivity (Figure 3).The repeatability of the synthetic method was tested for the

preparation of both the SCCA-IgM and AFP-IgM complexes. Theobserved immune reactivity of the different preparations wasvery similar (Figure 3).In the case of SCCA-IgM mimetic complex, the presence of

both SCCA and IgM within the same species was confirmed byWestern blot (Figure 4). About 1mg of the complex was subjectedto SDS–PAGE analysis, under reducing conditions, on a 0.75mm

2010 John Wiley & Sons, Ltd. J. Mol. Recognit. 2010; 23: 551–558

Figure 5. Gel filtration and immune reactivity analyses of the SCCA-IgM complex. Panel A reports the gel filtration profile, panel B reports themeasured

immune reactivity in the collected fractions. Insert displays the calibration curve of the column.

Figure 4. Negative image of the Western blot analysis of the SCCA-IgM complex. Panel A reports the detection of SCCA by incubation of the

nitrocellulose membrane with rabbit anti-SCCA polyclonal antibody and subsequent reaction with goat-anti-rabbit IgG complexed to horseradish

peroxidase. Paned B reports the detection of IgM heavy chains after incubation of the nitrocellulose membrane with goat-anti-human IgM (m-chainspecific) antibody complexed to horseradish peroxidase. Prior to incubation with the anti-IgM antibody, the membrane was soaked in 0.5M NaOH

for 5min to strip the antibodies used for the SCCA detection, little SCCA reactivity from the authentic sample used as reference is however still

detectable.

SOLID-PHASE PREPARATION OF PROTEIN COMPLEXES

5

(10%) polyacrylamide gel slab, the bands were transferred onnitrocellulose membranes and reacted with anti-IgM andanti-SCCA polyclonal antibodies. In these conditions, the complexfalls apart, and the identity of the bands was confirmed bycomparison with original samples. Due to the relatively largeexpected molar mass of the complex, SDS–PAGE analysis couldonly give a confirmation of the simultaneous presence of bothmolecules in the complex rather than allowing a preciseidentification of the complex mass distribution.

Gel filtration analysis of the complexes

A sample of the artificial SCCA-IgM complex was analysed bygel-filtration chromatography (Figure 5), the complex eluted witha retention time of 6.6min. The effluent was collected in 500mlfractions and each fraction was analysed by ELISA in order toconfirm the nature of the compound. The immune reactivity ofthe complex was detected in the fractions eluting at a retention

J. Mol. Recognit. 2010; 23: 551–558 Copyright � 2010 John Wiley &

volume close of about 6.6 ml, the calculated Kav for the elutedspecies was 0.08, no reactivity was detected in other fractions.From the calibration curve of the column, the molecular mass ofthe complex could be estimated to be in the range of 1–2MDa.However, since the resolution of the ordinary stationary phasesfor gel filtration chromatography is low in this range of molecularmasses, these values should be considered as lower bounds.

CONCLUSION

In this report, we preliminarily described a novel methodology forthe continuous-flow solid-phase assembling of two differentbiotinylated molecules. As a proof of principle, the method hasbeen validated by preparing complexes with immune reactivitymimicking that of the natural SCCA-IgM and AFP-IgM immunecomplexes, but other applications of the procedure can beenvisaged. The design of the approach was led by practical

Sons, Ltd. View this article online at wileyonlinelibrary.com

57

P. PENGO ET AL.

558

considerations aiming to limit the costs associated to the process.Themethod allowed the preparation of robust protein complexesby using a limited amount of proteins in mild conditions. Theimmobilization of the first building block on the resin bymeans ofa specific antibody blocks, from the steric point of view, part ofthe protein surface from any further reaction, ensuring at thesame time the recovery of a significant immune reactivity afterthe synthesis. The crude preparations of the protein complexdisplayed immune reactivity mimicking that of the naturalimmune complexes, and no purification steps were required. Inboth cases, the chemical connectivity between SCCA and IgM or

View this article online at wileyonlinelibrary.com Copyright �

between AFP and IgM within the same complex was firmlyestablished by dedicated sandwich ELISA; the signals obtained inthe same ELISA by using unassembled proteins were negligible.The presence of both species within the same complex was alsoconfirmed by immunoblot analysis with specific antibodies. Thesynthetic strategy devised exploits ‘orthogonal’ non-covalentinteractions for protein immobilization and functionalization andallows the repeated use of the same solid phase.The implementation of this process on a larger scale suitable

for industrial production is an ongoing project and will bereported in due course.

REFERENCES

Best MD. 2009. Click chemistry and bioorthogonal reactions: unprece-dented selectivity in the labeling of biological molecules. Biochem-istry 48: 6571–6584.

Brennan JL, Hatzakis NS, Tshikhudo TR, Dirvianskyte N, Razumas V, PatkarS, Vind J, Svendsen A, Nolte RJM, Rowan AE, Brust M. 2006. Biona-noconjugation via click chemistry: the creation of functional hybridsof lipases and gold nanoparticles. Biocomplex Chem. 17: 1373–1375.

De Falco S, Ruvoletto MG, Verdoliva A, Ruvo M, Raucci A, Marino M,Senatore S, Cassani G, Alberti A, Pontisso P, Fassina G. 2001. Cloningand expression of a novel hepatitis B virus-binding protein fromHepG2 cells. J. Biol. Chem. 276: 36613–36623.

Gauthier MA, Klok HA. 2008. Peptide/protein-polymer complexes:synthetic strategies and design concepts. Chem. Commun.2591–2611.

Hu T, Li D, Su Z. 2003. Preparation and characterization of dimeric bovinehemoglobin tetramers. J. Protein Chem. 22: 411–416.

Hu T, Su Z. 2003. A solid phase adsorption method for preparation ofbovine serum albumin-bovine hemoglobin complex. J. Biotechnol.100: 267–275.

Russell J, Colpitts T, Holets-McCormack S, Spring T, Stroupe S. 2004.Defined protein complexes as signaling agents in immunoassays.Clin. Chem. 50: 1921–1929 .

Russell LC, Colpitts TL, Holets-McCormack SR, Spring TG, Stroupe SD, VogtAD, Wong ST. 2002. Solid phase assembly of defined protein com-plexes. Bioconjug. Chem. 13: 958–965.

Suo X, Zheng C, Yu P, Lu X, Ma G, Su Z. 2009. Solid phase pegylationof hemoglobin. Artif. Cells Blood Substit. Immobil. Biotechnol. 37:147–155.

Towbin H, Staehelin T, Gordon J. 1979. Electrophoretic transfer of proteinsfrom polyacrylamide gels to nitrocellulose sheets: procedure andsome applications. Proc. Natl. Acad. Sci. USA 76: 4350–4354.

Wong SS. 1991. Chemistry of Protein Conjugation and Cross-linking. CRCpress: USA. (and references cited therein).

2010 John Wiley & Sons, Ltd. J. Mol. Recognit. 2010; 23: 551–558