Biosensor-Based Micro-Affinity Purification for the Proteomic Analysis

May 8, 2005 - ways.2,3 Affinity purification-mass spectrometry (MS), combin- ... sensor surfaces at high concentration in small volumes (1-10. µL) ha...
1 downloads 0 Views 325KB Size
Biosensor-Based Micro-Affinity Purification for the Proteomic Analysis of Protein Complexes B. Catimel,† J. Rothacker,† J. Catimel,† M. Faux,† J. Ross,† L. Connolly,‡ A. Clippingdale,‡ A. W. Burgess,† and E. Nice*,† The Ludwig Institute for Cancer Research, Melbourne Tumor Biology Branch, PO Box 2008 Royal Melbourne Hospital Victoria 3050 Australia, and Joint Proteomics Services Facility, Melbourne, Australia Received May 8, 2005

A biosensor-based micro-affinity purification method to recover protein binding partners and their complexes for down stream proteomics analysis has been developed using the BIAcore 3000 fitted with a prototype Surface Prep Unit (SPU). The recombinant GST-intracellular domain of E-cadherin or the recombinant GST-β-catenin binding domain of Adenomatous Polyposis Coli (APC) were immobilized onto the SPU and used to affinity purify binding partners from chromatographically enriched SW480 colon cancer cell lysates. A GST- immobilized surface was used as a control. Samples recovered from the SPU were subjected to SDS-PAGE with sensitive Coomassie staining followed by automated in-gel digestion and LC-MS/MS. The results obtained using the SPU were compared with similar experiments performed using Sepharose beads. Keywords: biosensor micro-affinity purification • proteomics • APC • E-cadherin • β-catenin

Introduction Cell functionality is controlled by multi-protein complexes that interact in a dynamic and cooperative manner in time and space.1 Functional proteomics is rapidly becoming one of the most promising approaches for deciphering the intricacies of cellular biochemistry and provides valuable insight into the organization and regulation of protein networks and pathways.2,3 Affinity purification-mass spectrometry (MS), combining biochemical selectivity with MS analysis sensitivity, is an established technique for the isolation and characterization of protein complexes in this field.2-4 Immunoprecipitation, epitope tagging, GST pull down, and tandem affinity purification (TAP) are commonly used affinity techniques.2-4 Optical SPR-biosensors (e.g., BIAcore, IAsys) also provide an excellent platform for microaffinity purification that can be readily used in the field of affinity-MS. Biosensor technology combined with micropreparative HPLC was originally used for the sensitive and specific screening of chromatographic fractions5 and the identification and purification of orphan biomolecules.6-14 Further development of this approach has led to the utilization of biosensor surfaces as micro-affinity purification platforms to recover sufficient material for ligand identification using specific and sensitive downstream analytical techniques (e.g., mass spectrometry, microsequence analysis, bioassays).15,16 * To whom correspondence should be addressed. Ludwig Institute for Cancer Research, Melbourne Tumor Biology Branch, Post Office Royal Melbourne Hospital, Melbourne, Australia. Tel: + 61 3 9341 3135. Fax: +61 3 9341 3104. E-mail: [email protected]. † The Ludwig Institute for Cancer Research, Melbourne Tumor Biology Branch. ‡ Joint Proteomics Services Facility.

1646

Journal of Proteome Research 2005, 4, 1646-1656

Published on Web 08/09/2005

The multiplexed analytical technique involving the use of BIAcore sensor chips as preparative surfaces with subsequent direct characterization of bound proteins by matrix-assisted laser desorption-ionization time-of-flight (MALDI-TOF) has been named biomolecular interaction analysis mass spectrometry (BIA/MS).17 In a two-dimensional approach, the biosensor analysis is used to both affinity purify the binding partners and simultaneously analyze the biomolecular interaction and determine its binding parameters, while the MALDI-TOF reveals mass information on compounds bound to the sensor surface. BIA/MS studies use the sensor chip as a MALDI-TOF target and therefore require the sensor chip to be introduced directly into the mass spectrometer.18-33 This requires careful and expert experimental manipulations (MALDI-TOF optimization, choice of matrix and matrix application, chip cutting and insertion into the mass spectrometer) and has to date been performed using in-house fabricated devices.21 Moving toward multiplexed SPR/MS platforms, multi-affinity ligand surfaces were also studied using BIA/MS by immobilizing up to 6 antibodies onto a single flow cell.34,35 MALDI-TOF MS analysis of affinity captured proteins from human plasma has allowed the identification of all targeted proteins.34,35 A more flexible approach to using biosensor surfaces for micropurification and identification is the elution and recovery of specifically bound proteins from the sensor surface prior to downstream MS analysis. We have previously shown the potential of cuvette based biosensors (e.g., IAsys Auto+) for automated micropurification and downstream analysis including MS, N-terminal sequence analysis, SDS-PAGE and Western Blot analysis.36,37 The surface area of IAsys micropreparative cuvettes (16 mm2) increased the binding capacity, enabling increased ligand recovery compared with standard BIAcore 10.1021/pr050132x CCC: $30.25

 2005 American Chemical Society

research articles

Biosensor-Based Micro-Affinity Purification

surfaces. While the smaller surface area of the BIAcore flow cells (1.6 mm2) limits the capacity and hence the amount of material (femtomole levels) that can be recovered at each cycle, the use of the manual BIAcore PROBE, which has a surface area of 12-20 mm2, has allowed this limitation to be overcome.38,39 In an alternative approach, using the BIAcore X (manual instrument) proteins bound to the sensor chip were fragmented by delivering proteolytic enzymes directly to the sensing flow cells. The peptides that were generated were eluted and trapped onto a microcapillary RP precolumn connected to the BIAcore X using an in-house fabricated “recovery port”.40,41 The trapped peptide mixture was then analyzed using HPLC-ESI-MS/MS. In a recent study, BIAcore X captured proteins from dedrimer-based immunosensing surfaces were characterized using MALDI-TOF fingerprinting following on chip-digestion in an in-house fabricated tight-sealed chamber.42 The potential for direct automated elution and recovery from sensor surfaces at high concentration in small volumes (1-10 µL) has now been implemented into some flow-based instruments.38,43-45 The integrated flow-cell of the BIAcore 2000 was designed for sample recovery after the sample had passed through the integrated fluidic cartridge, although sample diffusion hampered its practical use. This problem was addressed in the design of the BIAcore 3000, where dilution effects during recovery were minimized by air segmentation and reversal of the buffer flow to return the sample back through the flow cell with collection via the autosampler.46-48 Following automated recovery onto a MALDI target and on-target digestion, the recovered proteins were identified by peptide mass fingerprinting using MALDI-TOF MS analysis. Electrospray tandem mass spectrometry (ESI-MS/MS) has been combined with microrecovery from the conventional IFC of the BIAcore 3000 to identify putative p53-interacting proteins (haemoglobin, protein phosphatase PP1C and a cyclin-dependent kinase inhibitor p57/Kip2).49 The BIAcore 3000 microelution procedure was also used to study G-protein coupled receptors using cell membranes immobilized onto a BIAcore LI hydrophobic surface. This surface was then used to trap the binding partner for the receptor which was then eluted and characterized by LC-MS/MS.50 While the above studies have clearly demonstrated the potential of this approach, in the large majority of examples the experimental systems have used well characterized antibodyligand interactions of high affinity or recombinant proteins that are expressed at significantly higher levels than the proteins of interest in native biological samples. In most cases, only single interacting species were identified. Furthermore, most of the protein identification was based on direct mass determination or mass fingerprinting without confirmation of peptide identity by MS/MS. Protein mass fingerprinting fails to identify low molecular mass proteins or the proteins in complex mixtures reliably, especially at low protein levels.51 There is also the potential for false positive identification when using tandem mass spectral data and matching of experimentally generated MS/MS data to the theoretical best match from protein databases.52 Sequence generating methods such as ESI-MS/ MS with careful manual inspection of the generated spectra are preferable for unequivocal protein identification.51 In this study we have evaluated the potential of a prototype BIAcore Surface Prep Unit (SPU) for the micropurification of proteins and protein complexes from biological samples. The SPU (Figure 1), which can be installed in one of the sample racks in the BIAcore 3000 biosensor workstation, increases the

Figure 1. (A) Schematic showing the surface area of the SPU and conventional IFC. Left hand panel: normal flow cell profile giving 4 flow cells each with a surface area of 1.2 mm2, righthand panel: SPU flow cell profile giving a total surface area of 16 mm2. (B) Photo showing the SPU installed in the right-hand sample rack in the instrument platform

available surface area to 16 mm2 (0.8 µL cell volume). The BIAcore SPU has short injection and recovery channels that minimize carry-over from material in the injected sample to the recovered sample. However, since the prototype unit is not coupled to the detector, real-time detection is not available to monitor the binding events occurring onto the SPU surface. Therefore, conditions for optimal immobilization, binding, and recovery have to be first established using a sensor chip docked in the standard BIAcore IFC before using on the SPU. Sample injections, flow rate, washing steps and analyte recovery are handled automatically by the analyte recovery wizard program using SPU-specific commands. Bound analyte is recovered in 2µL of regeneration solution or can be automatically deposited onto a MALDI target. In our experiment, sample recovered from multiple cycles of binding, elution, and regeneration were pooled in a single tube. The recombinant GST-intracellular domain of E-cadherin (GST-E-cadherin-ICD) and the recombinant GST-β-catenin binding domain of Adenomatous Polyposis Coli (APC) (GSTAPC2, which contains residues 1011-1470 of the APC sequence53) were immobilized onto the SPU and used to affinity purify binding partners from chromatographically enriched SW480 cell lysates. Recombinant GST was immobilized onto the SPU as a control surface. These SPU-affinity experiments were compared with classical immunopurification techniques using GST-E-cadherin-ICD and GST-APC2 immobilized onto NHS-activated Sepharose activated beads. Proteins recovered from both the SPU and the Sepharose beads were subjected to SDS-PAGE followed by automated in-gel digestion and LCMS/MS for the sensitive and specific identification of the microaffinity purified components.

Materials and Methods Preparation and Purification of Recombinant GST-Ecadherin-ICD, GST-APC2, and GST. The cDNA of E-cadherinICD (residues 736-882, accession number P12830) or β-catenin binding domain of APC (APC2, residues 1011-1470, accession no. P25054) were cloned into the GST gene fusion vector, pGEX4T-1 (Pharmacia Biosciences, Uppsala, Sweden). The plasmids were then transformed into BL21 (DE3) bacterial strain and grown to log phase, where protein expression was induced with 0.2 mM IPTG for 4 h at 37 °C. GST protein was produced by transforming the empty pGEX-4T-1 vector into the BL21 (DE3) cells and then inducing protein expression in the manner described above. The proteins were affinity purified from bacterial lysates using glutathione-Sepharose (Sigma, St Louis, MO) affinity chromatography followed by size exclusion chroJournal of Proteome Research • Vol. 4, No. 5, 2005 1647

research articles matography using Superdex 75. Purified proteins were analyzed by SDS-PAGE and their identities confirmed by MS/MS analysis. Preparation of SW480 Colonic Carcinoma Cell Extracts. SW480 colonic carcinoma cells (1 × 109 cells) were washed 3 times in PBS and solubilized (108 cells/mL) for 30 min at 4 °C using 2.5% (v/v) CHAPS in 15 mM Hepes pH 7.4, 2 mM mercaptoethanol, 1mM EGTA and a cocktail of protease inhibitors (Complete, Roche Applied Science, Mannheim, Germany). After centrifugation at 15 000 × g for 20 min, the lysate was collected for anion exchange chromatographic purification. The SW480 cell extract (10 mL, 32 mg of protein) was loaded at 4 °C onto a Mono Q HR 10/10 column, connected to an AKTA HPLC system (Pharmacia Biosciences, Uppsala, Sweden), equilibrated in 10 mM Tris-HCl (pH 7.4) containing 0.05% (w/v) CHAPS. After washing with this buffer, bound proteins were eluted using a 45 min linear 0-1 M NaCl gradient at a flow rate of 1 mL/min. Fractions (2 mL) were collected and aliquots (40 µL) analyzed by SDS-PAGE and analytical biosensor analysis13,54,55 using immobilized recombinant GSTintracellular E-cadherin domain. Active fractions (14-38 min) were then pooled (24 mL) and aliquots buffer-exchanged into HBS (10 mM Hepes pH 7.4, 3.2 mM EDTA, 150 mM NaCl, 0.005% Tween 20) using a PD10 column (Amersham Biosciences, Uppsala, Sweden) prior to injection onto the biosensor for microaffinity purification. Sensor Chip Preparation. All experiments were carried out using a BIAcore 3000 (BIAcore, Uppsala, Sweden) and Sensor Chip CM5 at 25 °C. The GST-E-cadherin-ICD was immobilized using NHS/EDC amine coupling chemistry with HBS as the running buffer as described previously.55 For immobilization using the Surface Preparation Unit (SPU) an increased volume of NHS/EDC (150 µL) was used to ensure complete coverage of the chip surface. GST-E-cadherin-ICD was then injected over the sensor surface at a concentration of 80 µg/mL in acetate buffer (pH 3.7) for 30 min at a flow rate of 10 µL/min. The surface was washed with 10 mM HCl to remove any noncovalently bound material and then blocked with 100 µL ethanolamine (1M, pH 8.0). Recombinant GSTAPC2 and a GST control surface were immobilized using identical conditions. Surface Characterization. To assess immobilization level and surface reactivity, the sensor chip was docked in the standard IFC after immobilization. The level immobilized was estimated by subtracting the relative response (RU) obtained after immobilization with the response level prior to immobilization for each of the 4 channels. The immobilization level was estimated by averaging the responses obtained on the 4 channels. The surface reactivity was then checked by flowing the SW480 MonoQ pool over the chip for 5 min at a flow rate of 10 µL/min. The regeneration conditions were also optimized (buffer, contact time) and the sample was re-injected after regeneration to confirm retained surface reactivity after desorption. Preparative Analyte Binding and Recovery. The following automated method was set up using the BIAcore Analyte Recovery Wizard supplied with the instrument to control sample application, elution, and recovery: MAIN RACK 1 thermo_a RACK r reag_a RACK 2 sp_2 APROG MSPrep (50 cycles were performed) 1648

Journal of Proteome Research • Vol. 4, No. 5, 2005

Catimel et al.

END DEFINE APROG MSPrep FLOW 10 MSP_PRIME MSP_INJECT rr2 50 10 MSP_RECOVER rr4 rr8 120 r1d1 END rr2 ) sample position rr4 ) wash buffer position rr8 ) elution buffer position Fifty consecutive cycles of sample injection, wash, recovery, and regeneration were performed with the E-cadherin and APC surfaces. Pooled SW480 MonoQ fractions were injected over the sensor surface for 5 min at a flow rate of 10 µL/min. The initial binding and washing steps were with HBS buffer (rr4). Recovery and regeneration was performed using 2 µL of 10 mM HCl (rr8) for 120 s and the recovered fractions were collected automatically into a single tube (r1d1) containing 10 µL of SDSPAGE sample loading buffer (Invitrogen, Carlsbad, Ca, USA). Immuno-Purification Using Sepharose Affinity Beads. Purified GST-E-cadherin-ICD and GST-APC2 were conjugated onto NHS-activated Sepharose (Amersham Biosciences, Uppsala, Sweden) at a ligand concentration of 0.5 mg protein per ml of beads. Coupling efficiency was monitored using SDSPAGE as well as size-exclusion chromatography analysis of a sample aliquot (100 µL) before, and after, conjugation. Remaining active NHS groups were blocked using 1 M ethanolamine (pH 8.0). Two mL of antigen conjugated beads were used for each immuno-purification experiment. Affinity beads were incubated overnight with the SW480 Mono Q pool (24 mL), the slurry poured into a 1 × 10 cm EconoColumn (Bio-Rad, Hercules, Ca, USA), washed extensively with HBS buffer and eluted with 2 mL of 10 mM HCl for SDS-PAGE and proteomics analysis. SDS-PAGE Analysis. The pooled proteins recovered from the sensor surface or the affinity columns were concentrated to 30 µL using a Christ RVC 2-25 rotational vacuum concentrator (Quantum Scientific, Australia) and loaded under reducing conditions onto a 4-12% SDS-PAGE gel (Invitrogen, Carlsbad, Ca, USA). Proteins were stained using sensitive Coomassie stain56 and the protein bands excised and subjected to in-situ trypsin digestion using a robotic digestion station (MassPREP, Micromass, Altrincham, UK). MS/MS Analysis. An electrospray ion trap mass spectrometer (model LCQ DECA, Finnigan, San Jose, CA), equipped with a Picoview source (New Objective, Woburn, MA) and coupled on line to an Agilent model 1100 capillary HPLC (Agilent, Waldbronn, Germany), was used for MS/MS analysis. A 100 × 0.15 mm inner diameter C4 capillary column (3 µm; 300 Å, model ProteCol-C4, SGE, Melbourne, Australia) was used with a linear 80-min gradient (flow rate 0.8 µL/min) from 0 to 100% B, where solvent A was 0.1% (v/v) aqueous formic acid and solvent B was 0.1% aqueous formic acid in 60% (v/v) acetonitrile. The electrospray ionization (ESI) parameters were as follows: PicoTip emitter (model FS360-50-30-D20, New Objective, Woburn, MA), spray voltage, 1.9 kV; capillary temperature, 175 °C; capillary voltage, 15.95 V. All data were collected using “Triple play” experiments, consisting of MS/zoom scan/ and MS/MS. MS/MS data was searched using the Mascot algorithm (Matrix Science, London, UK) and resultant protein hits were validated by manual inspection of the MS/MS spectra.

Biosensor-Based Micro-Affinity Purification

research articles

Figure 2. (A) Anion exchange chromatography of SW480 cell lysate. Fractions eluting between 14 and 38 min were pooled for biosensor recovery. (B) Analytical biosensor analysis of SW480 anion exchange fractions using immobilized E-cadherin intracellular domain. The curves shown are blank subtracted using a derivatized blank channel. (C) Coomassie-stained SDS-PAGE gel of the anion exchange fractions: lanes 1 to 7 represent the fractions eluting between 14 and 18, 18-22, 22-26, 26-30, 30-34, 34-38, and 38-42 min, respectively.

Figure 3. Analytical biosensor analysis of the starting material for the biosensor microrecovery experiment using the SPU. (A) Injection of the SW480 Mono Q pool SPU immobilized GST-E-cadherin ICD using the 4 channels of the IFC. (B) Injection of the SW480 Mono Q pool SPU immobilized GST-APC2 using the 4 channels of the IFC. (C) Injection of the SW480 Mono Q pool SPU immobilized GST using the 4 channels of the IFC. The start of sample injection is indicated by an arrow.

Results Preparation of the SPU Surface. For the BIAcore recovery studies, either GST-E-cadherin-ICD, GST-APC2, or a GST control were immobilized in situ onto a CM5 sensor chip docked in the SPU. Following immobilization in the SPU, the biosensor chip was re-installed in the standard BIAcore IFC to check for immobilization levels. From the increase in signal obtained (average of 20 123 RU across the 4 channels), the level of E-cadherin immobilization was estimated at 20 ng/mm2 (1000 RU ) 1 ng/mm2)57 equivalent to 320 ng immobilized on the working area of the SPU sensor chip. Similar molar levels were immobilized on the GST-APC2 and GST surfaces (19 800 and 11 500 RU, respectively). Chromatographic Concentration and Pre-fractionation Using Anion Exchange Chromatography on MonoQ HR10/ 10. SW480 colonic carcinoma cells58 were used as a source of binding proteins. Cells (109) were lysed using 2.5% CHAPS buffer (10 mL, 32 mg protein after cell lysis) before an initial anion-exchange chromatographic pre-fractionation and concentration step using a Mono Q (HR10/10 column) (Figure 2A). To select fractions containing potential E-cadherin binding partners for microaffinity purification on the SPU, aliquots (30 µL) of the anion-exchange fractions (diluted 1:3 in HBS buffer) were screened analytically using the BIAcore3000 with GSTE-cadherin-ICD immobilized onto a CM5 biosensor surface (Figure 2B). Reactivity was detected across multiple fractions (14-36 min) corresponding to a number of UV-adsorbing peaks. The binding signals observed on the BIAcore displayed different characteristics: in particular, the dissociation rate constant (kd), which is independent of protein concentration,59

varied from 0.013 s-1 (14-18 min fraction) to 5.6 × 10-4 s-1 (34-38 min) depending on the fraction analyzed (Figure 2B). This heterogeneity suggested the presence of more than one binding partner, possible protein complexes or the presence of post-translational modifications affecting the affinity of the binding partner. The active Mono Q fractions were pooled for subsequent micro-affinity purification on the BIA SPU system. Each affinity purification experiment was performed using similar fractions from a fresh SW480 Mono Q run, which were shown to be active by analytical BIAcore analysis.13,54 The protein complexity of these fractions on SDS-PAGE analysis, from a representative sample, is shown in Figure 2C. Monitoring of SPU Surface Reactivity. Since the prototype SPU is not coupled to the detector, the surface reactivity and recovery conditions were established prior to performing the preparative runs by reinserting the chip, which had been derivatized in the SPU, back into the standard IFC. Surface reactivity was assessed by injecting 50 µL of the pooled SW480 Mono Q fractions over the surface at a flow rate of 10 µL/min. Using the GST-E-cadherin-ICD surface the binding signal observed after injection across the 4 channels of the IFC was approximately 1600 RU (Figure 3A) corresponding to 1.6 ng/ mm2. Following 2 min dissociation when HBS buffer alone was flowing over the surface, only a small decrease in signal was observed, leaving a binding signal of approximately 1000 RU equivalent to approximately 1 ng/mm2. These values would suggest that, using the SPU in the preparative mode, we could recover approximately 16 ng of protein/cycle if the surface reactivity was stable. The sensor surface could be easily Journal of Proteome Research • Vol. 4, No. 5, 2005 1649

research articles

Catimel et al. Table 1. E-Cadherin SPU Recoverya protein identified

no. of peptides identified

P35221 4 R1-catenin 5% coverage Q43707 2 R-actinin 4 3% coverage

position in sequence b

166-178

NAGNEQDLGIQYK

654-670b 671-681 684-695b 115-122

TSVQTEDDQLIAGQSAR AIMAQLPQEQK Oxidation (M) IAEQVASFQEEK ALDFIASK

883-899

MAPYQGPDAVPGALDYK Oxidation (M) ATQADLMELDMAMEPDRK Acetyl (N-term); 3 Oxidation (M) LAEPSQMLK Oxidation (M) LLNDEDQVVVN AAVMVHQLSK Oxidation (M) SPQMVSAIVR SPQMVSAIVR Oxidation (M) TMQNTNDVETAR Oxidation (M) EGLLAIFK SGGIPALVK LVQNCLWTLR AGDREDITEPAICALR LHYGLPVVVK NLALCPANHAPLR LVQLLVR TSMGGTQQQFVEGVR VAAGVLCELAQDK NEGVATYAAAVLFR MSEDKPQDYK

2-19 P35222 β-catenin

18 26% coverage

Figure 4. Analysis of recovered fractions using the Surface Prep Unit. Lane 1: Western blot analysis of GST-E-cadherin SPU fraction, lane 2: Coomassie stained SDS-PAGE of GST-Ecadherin, lane 3: Western blot analysis of GST-APC2 SPU fraction, lane 4: Coomassie stained SDS-PAGE of GST-APC2 SPU fraction, lane 5: Western blot analysis of GST-SPU fraction, lane 6: Coomassie stained SDS-PAGE of GST-SPU fraction. Western blot analyses (lanes 1, 3, and 5) were performed using anti-β-catenin IgG.

regenerated back to baseline using 10 mM HCl and identical reactivity was obtained upon re-injection of the sample after regeneration (data not shown), suggesting that the surface could be used for a large number of repetitive capture cycles. Using the GST-APC surface, the binding signal observed after injection across the 4 channels of the IFC was approximately 375 RU corresponding to approximately 0.38 ng/ mm2 (Figure 3B). The binding was very stable, and after 2 min dissociation a residual binding signal of 340 RU was observed, demonstrating that approximately 5.5 ng of protein/cycle could be recovered using the SPU. Injection of 50 µL of the pooled SW480 Mono Q fractions over the GST control surface resulted in only very low level binding (50RU) (Figure 3C), confirming the specific nature of the binding to the GST-E-cadherin-ICD and GST-APC2 channels (Figure 3A,B). Preparative Micro-Purification Using the SPU Recovery Unit. (1) GST-E-cadherin ICD Immobilized onto the SPU. On completion of the surface characterization using the conventional IFC, the chip was transferred back to the SPU for preparative micro-affinity purification of proteins binding to the GST-E-cadherin-ICD fusion protein: 50 consecutive cycles of automated injection/recovery were performed. Recovery was achieved after a flow cell wash with HBS by incubating the sensor surface with 2 µL of 10 mM HCl for 120 s. Air segmentation was used to prevent buffer contamination. The sensor chip reactivity was checked after 15, 30, and 50 cycles by re-docking the chip into the conventional IFC. Compared to the first injection, 82% of reactivity was retained after 15 cycles, 62% after 30 cycles and 35% after 50 cycles. The recovered fractions (2 µL/cycle, 100 µL total) were collected in a single vial containing 10 µL of SDS-PAGE sample buffer and then concentrated to a final volume of 30 µL and subjected to SDS-PAGE (25 µL) and Western blot analyses (5 µL). The presence of β-catenin, a known binding partner of the intracellular domain of E-cadherin,60 was detected by Western blot using an anti-β-catenin IgG (BD Biosciences Pharmigen, IL, USA: Figure 4, lane 1). Sensitive Coomassie staining56 revealed a number of protein bands which were manually excised (Figure 4, lane 2). In-gel digestion was performed on these bands followed by protein identification using capillary LC-MS/MS together with data1650

Journal of Proteome Research • Vol. 4, No. 5, 2005

peptide sequence

125-133 159-170b 171-180b 191-200 191-200 210-212b 226-233 234-242 377-386 454-469 487-496 516-528b 536-542 551-565 613-625b 648-661b 662-671b

a Proteins were identified by µHPLC-ESI-MS/MS analysis of tryptic digests. Results from 2 experiments are given. b These peptides were identified from the second experiment.

base searching (Table 1). To ensure the highest possible confidence in the protein identifications, the MS/MS spectra of the peptide matches were inspected manually to check the validity of the sequence assigned by the search algorithm. Representative MS/MS spectra of β-catenin, R1-catenin and R-actinin are shown in Figure 5. In this experiment, eleven β-catenin peptides were identified and sequenced following 50 cycles of binding and recovery, providing 14% protein coverage. R1-catenin, a β-catenin binding protein was identified from the unambiguous sequencing of a single peptide. R-actinin, a protein that acts as a bridge between R-catenin and actin, was also identified by 2 peptide sequences giving 3% protein coverage. The two lower molecular weight protein bands (Figure 4, lane 2) resulted in the identification of bovine serum albumin precursor (6 peptides), human cytokeratin 1 (4 peptides), cytokeratin 10 (2 peptides), β-actin (1 peptide), and heat shock proteins HS90A (Heat shock protein HSP 90R) (3 peptides), HSP71 (Heat shock 70 kDa protein 1) (1 peptide), and Hsp60 (CH60) (2 peptides). Several very faint bands of low molecular weight (below 36 kDa) were also observed but, because of the low levels, were not subjected to MS/MS analysis in this experiment. This experiment was repeated with E-cadherin-ICD immobilized at a level of 18 000 RU (18 ng/mm2) giving a binding signal of approximately 1650 RU upon injection of another SW480 Mono Q pool. This experiment lead to the identification of 4 R1-catenin peptides as well as 16 β-catenin peptides, resulting in a further identification of 3 additional R1-catenin and 7 β-catenin peptides to those observed in the initial experiment. Taken together, this resulted in overall protein coverage of 26% for β-catenin and 5% for R1-catenin. A summary of the total peptides sequences obtained for β-catenin, R1-catenin, and R-actinin from the two micro-affinity experiments with the E-cadherin-SPU is given in Table 1.

research articles

Biosensor-Based Micro-Affinity Purification

Table 2. Summary of the Proteins Identified Using the SPU Surfaces

GST-E-cadherin-ICD P35221 O43707 P35222 P07900 P10809 P08107 P79371 P19388 P13010 P02769 P04264 Z29074 P02570 P02545 P49748 P01876 P21980

R1-catenina R-actinin 4a β-catenin* HSP90A HSP60 HSP71 TCPB-1 nucleolin KU86A serum albumin precursor cytokeratin cytokeratin β-actin lamin ACADV IGHA1 TGM2

GST

P02570

GST-APC2 (β-catenin binding domain)

P35222

β-catenina

P02769 P04264 Z29074

serum albumin precursor cytokeratin cytokeratin

P01876 P21980

IGHA1 TGM2

β-actin

a These proteins are known to act specifically with the immobilized target protein.61-79 The protein accession numbers from the Swiss-Prot Protein knowledgebase (http://au.expasy.org/sprot/) are indicated before the protein names. Results of 2 experiments for GST-E-cadherin-SPU, GST-APC2SPU, and GST-SPU are shown.

Figure 5. ESI-MS/MS spectrum of β-catenin, R-1 catenin and R-actinin 4 recovered from the GST-E-cadherin ICD-SPU. (A) β-catenin residues 191-200, (B) R1-catenin residues 671-681, (C) R-actinin-4 residues 115-122.

HSP90, HSP71, HSP60, cytokeratin, and β-actin were again identified. In this experiment, where the levels recovered were slightly higher than those observed in the initial experiment (1650 RU per cycle compared with 1000 RU), all visible Coomassie bands were analyzed. This resulted in the identification of TGM2 protein glutamine δ glutamyltransferase (2 peptides), Acetyl CoA dehydrogenase (ACADV) (2 peptides), KU86 dependent DNA helicaseII (1 peptide), nucleolin (1 peptide), Ig R-1 chain C region (IGHA1) (1 peptide), lamin (nuclear lamina component) (3 peptides) and the molecular chaperone TCPB-1 (T-complex protein 1) (3 peptides) (Table 2). (2) APC2 (β-Catenin Binding Domain of APC) Immobilized onto the SPU. The β-catenin binding domain of APC (APC2 1011-1472) was immobilized onto the SPU sensor surface as a GST fusion protein (GST-APC2), using the same chemistry as described above for E-cadherin, and was used to capture potential binding partners from a SW480 Mono Q pool. A similar immobilization level was obtained for APC2 (average of 19 880 RU across the 4 channels, approximately 20 ng/mm2) compared with the E-cadherin-ICD channel (18 000 RU). However, the binding per cycle observed with this surface was low (340 RU), and therefore, 2 × 50 consecutive cycles of injection and recovery were performed to recover sufficient protein for SDS-PAGE (Figure 4, lanes 3 and 4) and proteomics

analysis (Table 2). Again, regeneration and recovery could be achieved using 10 mM HCl without significant denaturation of the sensor surface, with 67% of the initial binding activity remaining after 50 cycles. β-catenin was identified using Western blot analysis (Figure 4, lane 3) as well as MS/MS sequencing (2 peptides (residues 125-132 and 201-212, resulting in 2% protein coverage, Table 2). TGM2 protein glutamine δ glutamyltransferase (2 peptides), Ig R1 chain C region (IGHA1) (1 peptide), serum albumin precursor (2 peptides) β-actin (5 peptides) and cytokeratin (7 peptides) were also identified (Table 2). (3) GST Immobilized onto the SPU. GST-fusion E-cadherin ICD and APC2 were used in our microaffinty purification since the GST-tag enabled us to efficiently purify these recombinant proteins from bacterial cell lysates. A control fishing experiment was therefore performed with GST which was immobilized onto the sensor chip using the SPU unit (approximately 11 500 RU immobilized corresponding to approximately 11.5 ng/mm2). Injection of 50 µL of the pooled SW480 Mono Q fractions over the GST control surface resulted in only very low level binding (50 RU) (Figure 3C). The control recovery experiment therefore consisted of two fifty consecutive cycles of injection/recovery using the GST-chip docked in the SPU unit. The fractions recovered from the SPU were analyzed using SDS-PAGE and Western blot analysis (Figure 4, lanes 5, 6) as described above. Two low molecular bands were observed on the Coomassiestained gel: one band was identified as β-actin (1 peptide, Table 2) using MS/MS sequencing while the other band was too low level to yield any sequence data. No β-catenin was detected by either Coomasie staining or Western blot analysis (Figure 4, lane 5) using anti-β-catenin IgG. Immunopurification Using Sepharose Beads. GST-E-Cadherin-ICD-Sepharose. Three separate affinity experiments were performed using a 2 mL column of E-cadherin-ICD immobilized on Sepharose beads at a ligand density of 0.5 mg/ mL (1 mg total compared to approximately 300 ng immobilized onto the biosensor SPU surface). Affinity beads were incubated overnight with the SW480 Mono Q pool, washed extensively with HBS buffer and eluted with 4 mL of 10mM HCl. The eluted Journal of Proteome Research • Vol. 4, No. 5, 2005 1651

research articles

Catimel et al.

Figure 6. SDS-PAGE analysis of fractions from Sepharose affinity beads. Lane 1: GST-E-cadherin, lane 2: GST-APC2, lane 3: GST control. Proteins identified which are a result of bleeding of immobilized target from the column (GST-APC, GST-E-cadherin-ICD) are indicated by an asterix. Table 3. E-Cadherin-Sepharose Recoverya

protein identified P35221 Q12795

R1-catenin R2-catenin

Q42707

R-actinin 4

no. of peptides identified

position in sequence

peptide sequence

1 2 3% coverage 1 1% coverage

288-300 288-300 655-671

QIIVDPLSFSEER QIIVDPLSFSEER TSVQTEDDQLIAGQSAR

734-745

VGWEQLLTTIAR

2-19 P35222

β-catenin

peptides respectively) were also found, indicating leakage from the affinity column (Figure 6).

13 23% coverage

125-133 134-151 159-170 234-242 377-386 454-469 475-486 497-508 551-565 626-647 648-661 674-684

ATQADLMELDMAMEPDRK Acetyl (N-term); 3 Oxidation (M) LAEPSQMLK Oxidation (M) HAVVNLINYQDDAELATR (x2) LLNDEDQVVVNK SGGIPALVK LVQNCLWTLR AGDREDITEPAICALR HQEAEMAQNAVR Oxidation (M) LLHPPSHWPLIK TSMGGTQQQFVEGVR Oxidation (M) EAAEAIEAEGATAPLTELLHSR NEGVATYAAAVLFR LSVELTSSLFR

a

Proteins were identified by µHPLC-ESI-MS/MS analysis of tryptic digests. Results from 3 experiments are given.

sample was concentrated to 30 µL and analyzed using SDSPAGE with sensitive Coomassie staining. Protein bands were excised and subjected to MS/MS analysis as described (Figure 6). These experiments resulted in the identification of β-catenin (13 peptides), R1-catenin (1 peptide), R2-catenin (2 peptides), and R-actinin 4 (1 peptide) (Table 3). R actin (1 peptide), β-actin (3 peptides), and δ-actin (3 peptides) were also identified in these experiments. Of these, only R2-catenin was not detected in the SPU micro-affinity experiment. Others proteins identified were heat shock proteins GRP78 (2 peptides) and GR75 (1 peptide), A32B acidic-rich nuclear phosphoprotein 32 (1 peptide), NIPS1, and NPS2 proteins (NipSnap protein 1 and 2) (2 peptides), MA32 complement component 1 (1 peptide) and SET protein, an inhibitor-2 of protein phosphatase-2A (1 peptide) and serum albumin precursor (4 peptides). E-cadherin and GST peptides (3 and 20 1652

Journal of Proteome Research • Vol. 4, No. 5, 2005

APC2-Sepharose. Two separate affinity experiments were performed using the GST-APC2-Sepharose beads under the same experimental conditions as GST-E-cadherin-ICDSepharose. β-catenin was immunopurified from the SW480 Mono Q pool using APC2-Sepharose beads as demonstrated by Western blot analysis (result not shown) and MS/MS sequencing (2 peptides) (Figure 6, Table 4). R1-catenin, and R2-catenin, identified in E-cadherin-Sepharose affinity experiments were not present in the APC2-Sepharose affinity purified sample. DDB1 DNA damage binding protein 1 (1 peptide), TF1B transcription factor (1 peptide), MCM5 DNA replication factor (CDC46 homolog) (5 peptides), S23A protein component of ERderived vesicles involved in transport (5 peptides), GRP78 (4 peptides), CAZ1 and CAZ2 F-acting capping protein (8 peptides), RSP4 40s ribosomal protein (2 peptides), serum albumin precursor (1 peptide), cytokeratin (6 peptides) β-actin (1 peptide) were also identified. Again, APC2 and GST sequences were detected due to column bleeding during elution (Figure 6). GST-Sepharose. A control experiment was performed to assess the level of protein binding when GST alone is derivatized onto the Sepharose beads (Figure 6). This experiment was conducted using the same incubation time, washing and elution conditions as with the experiments with GST-Ecadherin ICD and GST-APC2 beads using the SW480 Mono Q. No β-catenin, R-catenin, or R-actinin were detected in the affinity purified proteins using either Western blot analysis with anti β-catenin IgG or MS/MS sequencing (Figure 6, Table 4). This suggested that the interaction of these proteins with the immobilized targets was specific. Cytokeratin (2 peptides), β-actin (15peptides), GR78 (17 peptides), GR75 (9 peptides), CAZ1-F-actin capping protein (2 peptides), and RSP4 ribosomal protein (7 peptides) were detected in the GST-eluate (Figure 6). These proteins were also identified in the GST-E-cadherin ICD and GST-APC2Sepharose experiments (Table 4). TGM2 Protein-glutamine δ-glutamyltransferase (19 peptides) identified on GST-Sepharose was also found on GST-E-cadherin ICD and GST-APC2 immobilized onto the surface Prep Unit (Table 2). These proteins can clearly interact with the GST domain of the fusion proteins.

research articles

Biosensor-Based Micro-Affinity Purification Table 4. Summary of the Proteins Identified Using the Sepharose Affinity Beadsa,c

GST-E-cadherin ICD P35221 Q12795 O43707 P35222

R1-cateninb R2-cateninb R-actinin 4b β-cateninb

P11021 P38646

GR78 GR75

P02769 P05783 P05787 P04264 P12718 P02570 P02571

Q01105 Q92688 Q07021 Q9BPW8 O75323

serum albumin precursor cytokeratin cytokeratin cytokeratin R-actin β-actin γ-actin

GST

P11021 P38646 P11142 P04792 P21980 P04406 P08865 Q99536 Q9Y230 Q99536 P35613 Q9Y266 P05198 P41091

GR78 GR75 HS7C HSB1 TGM2 G3P2 RSP4 ATBP RUV2 VAT1 CD147 MNUDC IF2A IF2G

GST-APC2 (β-catenin binding domain)

O43707 P35222 Q16531 Q13263 P33992 Q15436 P11021

R -actinin 4 β-cateninb DDB1 TF1B MCM5 S23A GR78

P08865

RSP4

P02769 P04264

cytokeratin

P35527

serum albumin precursor cytokeratin

P02570

β-actin

P02570

β-actin

P42025 P52907

β centractin CAZ1

P52907 P47755

CAZ1 CAZ2

P35656

OST4

SET A32B MA32 NISP1 NISP2

a The protein accession numbers from the Swiss-Prot Protein knowledgebase (http://au.expasy.org/sprot/) are indicated before the protein names. b These proteins are known to act specifically with the immobilized target protein.61-79 c Results from 3 experiments for GST-E-cadherin-ICDSepharose and 2 experiments for GST-APC-Sepharose and GST-Sepharose are shown.

Other proteins detected with GST-Sepharose were the heat shock proteins HS7C (8 peptides) and HSB1(2 peptides), β-centractin (Actin related protein 1B) (1 peptide), RUV2 (TATA box-binding protein) (10 peptides), ATPB (ATP synthase β chain) (7 peptides), CD147 antigen (Basigin precursor) (1 peptide), IF2G and IF2A translation initiation factor subunits (2 peptides), MNUDC (nuclear distribution protein C), G3P2 (Glyceraldehyde-3-phosphate Dehydrogenase) (4 peptides), VAT1 (Synaptic vesicle membrane protein homolog) (4 peptides), OST4 (Dolichyl-diphosphooligosaccharidesprotein glycosyltransferase) (1 peptide).

Discussion In our previous studies,36 we demonstrated the recovery potential of cuvette-based instruments for the micro-affinity purification of the A33 epithelial antigen present on colonic carcinoma cell lines using an immobilized anti-A33 IgG on the sensor surface. Following 60 automated cycles of binding and recovery, microgram quantities of essentially homogeneous A33 antigen were recovered and analyzed using SDS-PAGE, Western blotting, micropreparative RP-HPLC and N-terminal microsequence analysis. In an extension of this study, the cuvette-based

microaffinity purification of the A33 antigen was used in combination with enzymatic digestion of recovered proteins and downstream MALDI-TOF analysis,37 enabling identification of ten A33 antigen peptides, resulting in 54% coverage of the protein. We now describe the use of a prototype Surface Prep Unit (SPU) installed in the flow-based BIAcore 3000 surface plasmon resonance (SPR) biosensor, for the micro-affinity purification of protein binding partners and complexes from complex biological samples. The BIAcore is the most widely used SPR instrument,60 and the recovery methodology allows recovery in very small sample volumes (2 µL/cycle). The SPU increases the available surface area from 1.2 mm2 per channel to 16 mm2 (Figure 1), with a concomitant increase in the capacity of the chip available for micro-affinity capture. Multiple automated cycles of injection, washes and recovery can be performed using the SPU surface. In this study, samples recovered from the SPU were subjected to SDS-PAGE followed by in-gel digestion and LC-MS/MS for the sensitive and specific identification of the micro-affinity purified components. We have investigated the use of two β-catenin binding proteins as targets for the micropurification of proteins in chromatographically enriched fractions from SW480 colonic carcinoma cell extracts: the intracellular domain of E-cadherin (GST-E-cadherin-ICD) and the β-catenin binding domain of APC (GSTAPC2). Since we had used GST-fusion proteins to facilitate purification, a GST control surface was also generated in an attempt to identify any nonspecific binding to this region of the construct. E-cadherin is a transmembrane protein that mediates cellcell adhesion.61-63 The extracellular domain of E-cadherin homodimerizes in a calcium dependent manner, while the intracellular domain interacts indirectly with the actin cytoskeleton via proteins such as β-catenin, R-catenin, R-actinin and vinculin.64-67 Loss of E-cadherin function or expression has been associated with human epithelial cancer.68 The cytoplasmic domain of E-cadherin has been shown to display metastasis/tumor suppressor activity, probably by sequestering β-catenin.68 As well as being involved in cell-cell adherens junctions, β-catenin is a protein component of the Wnt signal transduction pathway that determines the transcription of the T-cell factor (TCF) responsive target genes.69 The APC tumor suppressor protein, in combination with axin and glycogen synthase kinase (GSK3β), is part of a Wnt signaling complex that mediates phosphorylation-dependent degradation of β-catenin.69 APC is a high molecular weight polypeptide (2843 aa, ≈ 312kD) of low abundance which is present in a wide variety of epithelial tissues.69 APC binds to β-catenin through a series of homologous 15 and 20 amino acids repeats.69 Dysregulation of the Wnt pathway in adult tissues results in the aberrant accumulation of β-catenin in the nucleus and is often associated with cancer. The cytosolic and membrane bound pools of β-catenin have different functions and have been shown to compete for binding partners such as APC, E-cadherin, and TCF.70,71 The central armadillo repeat region of β-catenin binds multiple partners including E-cadherin, APC, axin, TcF/LeF, resulting in the binding of APC or E-cadherin to β-catenin being mutually exclusive.72-76 The APC2 (1011-1470) domain used in our study encompasses the three 15 amino acid β-catenin binding repeats as well as two of the 20 amino acid β-catenin binding repeats present on the protein. The 15 aa β-catenin repeats of APC interact with armadillo repeats 5-8 of β-catenin in a phosJournal of Proteome Research • Vol. 4, No. 5, 2005 1653

research articles phorylation independent manner.77 By contrast, the binding of the 20 aa β-catenin binding repeats of APC to β-catenin is regulated by phosphorylation. A phosphorylated form of the 20 aa β-catenin binding repeats interact with armadillo repeats 1-5 of β-catenin and enhance the interaction between the two proteins by 300 to 500-fold.75 The intracellular domain of E-cadherin is known to interact with β-catenin and p120ctn 72,73. E-cadherin ICD interaction with β-catenin spans all the 12 armadillo repeats of β-catenin. In addition, a fragment of the phosphorylated E-cadherin composed of Ser690, Leu691, phosphorylated Ser692, and Leu 694 (mature E-cadherin sequence) binds to β-catenin armadillo repeats 3-5 in a similar way to the phosphorylated 20 aa β-catenin binding repeats of APC.72 Phosphorylation of Ecadherin enhances its binding to β-catenin78 while β-catenin phosphorylation decreases this interaction.79 p120ctn binds to the membrane proximal region of E-cadherin intracellular domain.73 We used SW480 carcinoma cell line extracts as a source of binding partners in these studies. SW480 cells have a truncated form of APC, resulting in high levels of cytoplasmic and nuclear unphosphorylated β-catenin. In this cell line, E-cadherin displays cytoplasmic puncta localization.80 Re-expression of full length APC protein in these cells results in redistribution of β-catenin and E-cadherin to the membrane.80 It is important to note that some native E-cadherin was also detected in the anion-exchange fractions by Western blot analysis using an anti-E-cadherin monoclonal antibody (data not shown). This native protein may have already formed complexes which would not be available for our binding studies. However, binding signals to immobilized GST-E-cadherin were specifically detected as compared to a blank derivatized channel or a GST control channel. Using a sensor surface which had been functionalized with β-catenin binding proteins (either GST-E-cadherin-ICD or GST-APC2) installed in the SPU in the BIAcore 3000, we were able to recover nanogram quantities of interacting proteins (βcatenin, R-catenin and R-actinin) present in SW480 cell lysate preparations in small volumes (2 µL/cycle) following automated repetitive injection and recovery. This enabled post-recovery analysis using sensitive Coomassie-stained SDS-PAGE followed by excision of the bands, enzymatic digestion and protein identification by LC-MS/MS analysis. The number of peptides detected from samples recovered from the SPU surface was similar to those obtained when larger amounts of target (1 mg) were immobilized onto Sepharose beads. β-catenin was specifically immunopurified using both GSTE-cadherin ICD and GST-APC2 immobilized on SPU surfaces. However, the number of peptides sequenced differed between surfaces (18 peptides, 26% surface coverage for GST-Ecadherin ICD; 2 peptides, 2% coverage for APC2 SPU). This is probably due to the difference in affinity between these two interactions: the binding between unphosphorylated APC and β-catenin is known to be of low affinity (KD ) 0.6 µM75) while we have shown, using the BIAcore that the KD for the interaction between unphosphorylated E-cadherin/ β-catenin is approximately 100 nM (Catimel et al., unpublished observation): both proteins were nonphosphorylated when immobilsed onto the SPU. This difference in affinity also possibly explains the difference in reactivity between APC2-SPU (binding signal of approximately 350 RU) and E-cadherin SPU (signal approximately 1600 RU) upon injection of the Mono Q starting material. 1654

Journal of Proteome Research • Vol. 4, No. 5, 2005

Catimel et al.

R1-catenin and R-actinin, that were immunopurified from the SPU using GST-E-cadherin ICD, were not identified using GST-APC2. This may also be due to the low level of β-catenin immunopurified. Alternatively, this result may suggest that the APC2 surface is interacting with a different pool of β-catenin to the one interacting with E-cadherin ICD. The interaction between β-catenin and R1-catenin is reported to have a KD of 100 nM.66 E-cadherin binding protein p120ctn and the R-catenin binding protein, vinculin67,73 were also not detected under our experimental conditions. Human cytokeratins and β-actin, which are frequently observed contaminants in proteomics studies, and bovine serum albumin precursor which is probably derived from the media used during cell culture, were eluted from both SPU surfaces. Heat shock proteins HSP90A, HSP71, and HSP60 were only identified on the E-cadherin SPU. These are all present at high levels in cells and interact with multiple key components in the cell to perform a diverse array of functions including protein folding, stress response, signal transduction and immunoregulatory activities.81 A control GST surface was used to further assess binding specificity. Very low levels of binding (50 RU/cycle) were observed on this surface, and β-actin was the only protein which could be identified. This suggests that some of the protein binding observed on the E-cadherin and APC2 SPU surfaces, which cannot be readily linked to known biological pathways, correspond to “irrelevant binding” i.e., binding to regions of the immobilized protein, but not part of the functional pathways. This is the case for the heat shock proteins, Acetyl CoA dehydrogenase (ACADV), TCBP-1, and the nuclear proteins KU86. Nucleolin and lamin were detected using the GST-E-cadherin-ICD but not with the GST-SPU control, and may also represent irrelevant binding to the GSTcadherin surface. IGHA1 and TMG2 were detected on both GST-E-cadherin-ICD and GST-APC2 SPU surfaces. TMG2 was later identified as a potential nonspecific component using GST-Sepharose affinity columns (see below). Biosensor surfaces have been suggested to have lower nonspecific binding compared to the corresponding chromatographic supports.21,36,82 This may be due, in part, to (1) the hydrophilic nature of the sensor surfaces which are coated with carboxymethyldextran; (2) the nonporous nature of the sensor surface compared with most comonly used chromatographic supports which are porous and were originally designed to have high capacity; and (3) the relatively high level of functional groups attached to the sensor surfaces. Low nonspecific binding can be vital when analyzing highly complex and heterogeneous biological solutions such as serum and cell lysates. The performance and specificity of corresponding surfaces on the SPU biosensor experiments and NHS Sepharose beads were therefore compared. A number of proteins were present in both the eluates from Sepharose affinity columns and the SPU surfaces (Tables 2 and 4). β-catenin was affinity purified using both GST-E-cadherin ICD and GST-APC2 with the SPU and Sepharose beads, but with different protein coverage between the two affinity surfaces. It is interesting to note that slightly better protein coverage for β-catenin was obtained from the GST-E-cadherin SPU sensor surface (18 peptides, 26% coverage) compared with the affinity beads (13 peptides, 23% coverage) (cf Tables 1, 3), although the surface immobilization level was much lower (approximately 320ng of GST-E-cadherin ICD immobilized on the SPU compared to 1 mg of receptor on the 2 mL affinity

research articles

Biosensor-Based Micro-Affinity Purification

columns). The SDS-PAGE results (Figues 4, 6) shows that the levels of β-catenin recovered were not as different as would be expected from the potential capacity of the two surfaces, suggesting that much smaller affinity columns could be used. Use of smaller columns would presumable also reduce the levels of nonspecific binding observed, due in part to the large surface area of these packings.36 Both the GST-E-cadherinIPC and GST-APC2 affinity columns also showed significant bleeding, resulting in contamination of the recovered binding proteins (Figure 6). No bleeding was observed using the SPU (Figure 4). As we had observed with the SPU experiments, R1-catenin was identified in the GST-E-cadherin-Sepharose affinity experiments, but not in the corresponding APC2-Sepharose purified eluates (cf Table 2,4). R2-catenin was not detected in the SPU eluate, but was identified using the E-cadherin beads (cf Table 2,4). R-catenins were only detected when E-cadherin ICD was immobilized onto both SPU and Sepharose affinity surfaces, suggesting again that different pools of β-catenin are interacting with the two proteins. R-actinin 4 was probably present due to its interaction with actin. Again, the known E-cadherin binding protein p120ctn and R-catenin binding protein vinculin were not detected using E-cadherin-ICD Sepharose beads. Several proteins were identified as binding to GST-APC2-Sepharose but not to the GST-E-cadherin Sepharose or the GST-Sepharose control. Among them are DDB1 DNA damage binding protein 1 (2 peptides), TF1B transcription factor (1 peptide), MCM5 DNA replication factor (CDC46 homolog, 5 peptides). Interestingly, DDB1 was recently characterized as part of a cullin-based ubiquitin ligase complex comprising DDB1, cullin 4A (CUL4A) and Regulator of Cullins-1 (ROC1).83-85 CUL4 is amplified in cancer86,87 and has been linked with DDB1 to the degradation of several substrates including replication licensing factor CDT1,85,88 c-Jun,89 STAT 1, and STAT2.90 APC has been recently shown to be downregulated by the ubiquitin proteasome.91 The APC domain involved in ubiquitination is localized between amino acids 960-1337. The APC domain we have used in our experiments (APC2 1011-1470) encompasses most of the region identified in ubiquitination. The affinity experiments suggest that DDB1 could be involved in APC ubiquitination. Furthermore, TF1B protein possesses a RING domain and may also be involved in ubiquitination. These proteins were only immunopurified by interacting directly with immobilized APC2 protein and not as β-catenin interacting proteins. However, these proteins were not identified using the APC2-SPU surface. This may be due to the lower protein loading of the SPU surface compared to affinity column (320 ng compared with 1 mg). More biochemical analyses need to be performed to confirm the role of these proteins in APC ubiquitination and degradation. Some proteins selectively bind to GST-E-cadherin ICDSepharose (A32B, NPS proteins, MA32 and SET) and GSTAPC2-Sepharose (SEC23, MCM5) but not to the GSTSepharose control (Table 4). However, their biological relevance is unclear. A relatively large number of proteins bind to GST-Sepharose compared to the GST-SPU surface. These proteins, which were not identified in our affinity SPU or Sepharose affinity experiments using GST-E-cadherin-ICD and GST-APC2, probably reflect nonspecific interaction to the GST or Sepharose support. This could suggest that that the sensor surface displayed less nonspecific interaction compared to chromatographic beads. However, the larger loading on the Sepharose beads seemed

to have enabled us to detect additional proteins such as R2catenin on E-cadherin-ICD-Sepharose and DDB1 and TF1B on APC2-Sepharose. The relative intensity of the bands obtained on SDS-PAGE with sensitive Coomassie staining following recovery from the Sepharose beads was only slightly stronger than that obtained from the SPU (cf Figures 4,6), suggesting that a much smaller bead volume could have been sufficient. The above experiments clearly show that, even if a control support is used in affinity-based experiments, additional biochemical experiments may be required to distinguish between specific and nonrelevant binding to the immobilized receptor. Optimally, elution competition with high concentrations of the soluble target would provide the best way to distinguish between specific and nonspecific binding, as used in other immunodetection methods (e.g., immunohistochemistry). However, large (mg) quantities of protein would be needed for such an approach.

Conclusion Using the SPU and immobilized GST-E-cadherin-ICD and GST-APC2 we have been able to affinity purify, and characterize by MS/MS analysis, a number of the known binding partners for these proteins. Similar results were obtained for replicate experiments, showing the excellent reproducibility of the method. Our results also suggest that DDB1, which was recently characterized as been part of a cullin based ubiquitin ligase complex may also be involved with APC ubiquitination. The results obtained with the SPU compare favorably with those obtained using affinity purification on Sepharose beads. However, the amounts of protein required for SPU immobilization are much lower (ng-µg) levels, allowing this technique to be used when protein levels are limited (e.g., native proteins, clinical samples, phosphorylation studies). As we have noted previously, the nonspecific binding and bleeding of the immobilized target protein appears to be lower with the sensor surfaces compared with the corresponding affinity supports, although the larger capacity of the affinity supports allows for detection of some additional proteins.

Acknowledgment. The authors were supported, in part, by NHMRC Program Grant No. 280912. References (1) Alberts, B. Cell 1998, 92, 291-294. (2) Zhu, H.; Bilgin, M.; Snyder, M. Annu. Rev. Biochem. 2003, 72, 783-812. (3) Bauer, A.; Kuster, B. Eur. J. Biochem. 2003, 270, 570-578. (4) Yanagida, M. Functional proteomics; current achievements. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 2002, 771, 89106. (5) Nice, E. C.; Lackman, M.; Smyth, F.; Fabri, L.; Burgess, A. W. J. Chromatogr. 1994, 660, 169-185. (6) Nice, E. C.; Catimel, B.; Lackmann, M.; Stacker, S.; Runting, A.; Wilks, A.; Nicola, N.; Burgess, A. W. Lett. Pept. Sci. 1997, 4, 107120. (7) Bartley, T. D.; Hunt, R. W.; Welcher, A. A.; Boyle, W. J.; Parker, V. P.; Lindberg, R. A.; Lu, H. S.; Colombero, A. M.; Elliott, R. L.; Guthrie, B. A.; Holst, P. L.; Skrine, J. D.; Toso, R. J.; Zhang, Fernandez, E.; Trail, G.; Varnum, B.; Y. Yarden, Y.; Hunter T.; Fox, G. M. Nature 1994, 368, 558-560. (8) Stitt, T. N.; Conn, G.; Gore, M.; Lai, C.; Bruno, J.; Radziejewski, C.; Mattsson, K.; Fisher, J.; Gies, D. R.; Jones, F.; Masiakowski, P.; Rian, T. E.; Tobkes, N. J.; Chen, D. H.; DiStefano, P. S.; Long, G. L.; Basilico, C.; Goldfarb, M. P.; Lemke, G.; Glass, D. J.; Yancopoulos, G. D. Cell 1995, 80, 661-670. (9) Lackmann, M.; Bucci, T.; Mann, R. J.; Kravets, L. A.; Viney, E.; Smith, F.; Moritz, R. L.; Carter, W.; Simpson, R. J.; Nicola, N. A.; Mackwell, K.; Nice, E. C.; Wilks, A. F..; Boyd, A. W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 2523-2527.

Journal of Proteome Research • Vol. 4, No. 5, 2005 1655

research articles (10) Davis, S.: Aldrich, T. H.; Jones, P. F.; Acheson, A.; Compton, D. L.; Jain, V.; Ryan, T. E., Bruno, J.; Radziejewski, C.: Maisonpierre, P. C.; Yancopoulos, G. D. Cell 1996, 87, 1161-1169. (11) Sakano, S.; Serizawa, R.; Inada, T.; Iwama, A.; Itoh, A.; Kato, C.; Shimizu, Y.; Shinkai, F.; Shimizu, R.; Kondo, S.; Ohno M.; Suda, T. Oncogene 1996, 13, 813-822. (12) Fitz, L. J.; Morris, C.; Towler, T.; Long, A.; Burgess, P.; Greco, R.; Wang, J.; Gassaway, R.; Nickbarg, E.; Kovacic, S.; Ciarletta, A.; Giannotti, J. A.; Finnerty, H.; Zollner, R.; Beier, D. R.; Leak, L. V.; Turner K. J.; Wood, C. Oncogene 1997, 15, 613-618. (13) Catimel, B.; Ritter, G.; Welt, S.; Old, L. J.; Cohen, L.; Nerrie, M. A.; White, S. J.; Heath, J. K.; Demediuk, B.; Domagala, T.; Lee, F. T.; Scott, A. M.; Tu, G. F.; Ji, H.; Moritz, R. L.; Simpson, R. J.; Burgess A. W.; Nice, E. C. J. Biol. Chem. 1996, 271, 25664-25670. (14) Seok, Y. J.; Sondej, M.; Badawi, P.; Lewis, S.; Briggs, M. C.; Jaffe H.; Peterkofsky, A. J. Biol. Chem. 1997, 272, 26511-26521. (15) Catimel, B.; Rothacker, J.; Nice, E. C. J. Biochem. Biophys. Methods 2001, 49, 289-312. (16) Nedelkov, D.; Nelson, R. W. Trends Biotechnol. 2003, 21, 301305. (17) Krone, J. R.; Nelson, R. W.; Dogruel, D.; Williams, P.; Granzow, R. Anal. Biochem. 1997, 244, 124-132. (18) Nelson R. W.; Krone, J. R. Anal. Chem. 1997, 69, 4363-4368. (19) Nelson R. W.; Krone, J. R.; Jansson, O ¨ . Anal. Chem. 1997, 69, 4369-4374. (20) Nelson R. W.; Krone, J. R. J. Mol. Recogn. 1999, 12, 77-93. (21) Nedelkov, D.; Nelson, R. W. J. Mol. Recogn. 2000, 13, 140-145. (22) Nelson, R. W.; J. W. Jarvik, J. W.; Taillon E.; Tubbs, K. A. Anal. Chem. 1999, 71, 2858-2865. (23) Nelson, R. W.; Nedelkov, D.; Tubbs, K. A.; Electrophoresis 2000, 21, 1155-1163. (24) Natsume, T.; Taoka M.; Manki, H.; Kume, S.; Isobe, T.; Mikoshiba, K. Proteomics 2002, 2, 1247-1253. (25) Nedelkov, D.; Rasooly. A.; Nelson, R. W. Int. J. Food Microbiol. 2000, 60, 1-13. (26) Nedelkov, D.; Nelson, R. W. Appl. Environ. Microbiol. 2003, 69, 5212-5215. (27) Nelson, R. W.; Nedelkov, D.; Tubbs, K. A. Anal. Chem. 2000, 72, 404A-411A. (28) Nedelkov, D.; Nelson, R. W. Anal. Chim. Acta 2000, 423, 1-7. (29) Nedelkov, D.; Nelson, R. W. Biosens. Bioelectron. 2001, 16, 10711078. (30) Nedelkov, D.; Nelson, R. W. Am. J. Kidney Dis. 2001, 38, 481487. (31) Nedelkov, D.; Nelson, R. W. Biopharm. 2001, 28-33. (32) Nedelkov, D.; Nelson, R. W. Proteomics 2001, 1, 1441-1446. (33) Nedelkov, D.; Nelson, R. W.; Kierna, U, A,; Niederkofler, E. E.; Tubbs, K. A. FEBS Lett. 2000, 536, 130-134. (34) Nedelkov, D.; Nelson, R. W. J. Mol. Recognit. 2003, 16, 9-14. (35) Nedelkov, D.; Nelson, R. W. J. Mol. Recognit. 2003, 16, 15-19. (36) Catimel, B.; Weinstock, J.; Nerrie, M. T.; Nice, E. C. J. Chromatogr. A 2000, 869, 261-273. (37) Catimel, B.; Rothacker, J.; Nice, E. C. Life Sci. 2002, 14, 24-30. (38) Fitz, L.; Cook, S.; Nickbarg, E.; Wang, J. H.; Wood, C. R. BIA J. 1998, 5, 23-25. (39) Williams C.; Addona, T. A. TIBTECH 2000, 18, 45-48. (40) Natsume, T.; Nakayama, H.; Jansson, O ¨ .; Isobe, T.; Takio K.; Mikoshiba, K. Anal.Chem. 2000, 72, 4193-4198. (41) Natsume, T.; Nakayama, H.; Toshiaki, I. Trends Biochem. 2001, 19, 528-533. (42) Seok, H. J.; Hong, M. Y.; Kim, J.; Han, M. K.; Lee, D.; Lee, J. H.; Yoo, J. S.; Kim, H. S. Anal. Biochem. 2005, 337, 294-307. (43) So¨nksen, C. P.; Nordhoff, E.; Jansson, O ¨ .; Malmqvist, M.; Roepstorff, P. Anal. Chem. 1998, 70, 2731-2736. (44) Mattei, B.; Cervone, F.; Roepstorff, P. Comput. Funct. Genom. 2001, 2, 359-364. (45) Gilligan J. J.; Schuck, P.; Yergey, A. L. Anal. Chem. 2002, 74, 20412047. (46) Roepstorff, P.; So¨nksen, C. P. BIA J. 1999, 6, 9-11. (47) Zhukov, A.; Schurenberg, M.; Jansson, O ¨ .; Areskoug, D.; Buijs, J. J. Biomol. Technol. 2004, 15, 112-119. (48) Lopez, F.; Pichereaux, C,; Burlet-Schiltz, O.; Pradayrol, L.; Monsarrat, B.; Esteve, J. P. Proteomics 2003, 3, 402-412. (49) Kikuchi, J.; Furukawa, Y.; Hayashi, N. Mol. Biotechnol. 2003, 23, 203-212.

1656

Journal of Proteome Research • Vol. 4, No. 5, 2005

Catimel et al. (50) Williams, C. Cur. Opin. Biotechnol. 2000, 11, 42-46. (51) Mattow, J.; Schmidt, F.; Hohenwarter, W.; Siejak, F.; Schaible, U. E.; Kaufmann, S. H. Proteomics 2004, 4, 2927-2941. (52) Cargile, B. J.; Bundy, J. L.; Stephenson, J. L. Jr. J. Proteome. Res. 2004, 3, 1082-1085. (53) Polakis, P. Biochim. Biophys. Acta 1997, 1332, 127-147. (54) Nice, E. C.; Lackmann, M.; Smyth, F.; Fabri, L.; Burgess, A. W. J. Chromatogr. 1994, 660, 169-185. (55) Catimel, B.: Faux, M. C.: Rothacker, J.: Otvos Jr., L.; Wade, J. D.; Nice, E. C.; Burgess, A. W. J. Peptide Res. 2002, 58, 204-212. (56) Wong, C.; Sridhara, S.; Bardwell, J. C.; Jakob, U. Biotechniques 2000, 28, 426-432. (57) Stenberg, E.; Persson, B.; Roos, H.; Urbaniczki, J. J. Colloid Interface 1990, 43, 513-526. (58) Gayet, J.; Zhou, X. P.; Duval, A.; Rolland, S.; Hoang, J. M.; Cottu, P.; Hamelin, R. Oncogene 2001, 20, 5025-5032. (59) Nice, E. C.; Catimel, B. Bioessays 1999, 21, 339-352. (60) Rich, R. L.; Myszka, D. G. J. Mol. Recognit. 2005, 18, 1-39. (61) Yap, A. S.; Brieher, W. M.; Gumbiner, B. M. Annu. Rev. Cell Dev. Biol. 1997 13, 119-146. (62) Takeichi, M. Development 1988, 102, 639-655. (63) Yamada, K. M.; Geiger, B. Curr. Opin. Cell Biol. 1997, 9, 76-85. (64) Knudsen, K. A.; Soler, A. P.; Johnson, K. R.; Wheelock, M. J. J. Cell Biol. 1995, 130, 67-77. (65) Nieset, J. E.; Redfield, A. R.; Jin, F.; Knudsen, K. A.; Johnson, K. R.; Wheelock, M. J. Cell. Sci. 1997, 110, 1013-1022. (66) Perez-Moreno, M.; Jamora, C.; Fuchs, E. Cell 2003, 112, 535548. (67) Weiss, E. E.; Kroemker, M.; Rudiger, A. H.; Jockusch, B. M.; Rudiger, M. J. Cell Biol. 1998, 141, 755-764. (68) Wong, A. S. T.; Grumbiner, B. M. J. Cell. Biol. 2003, 161, 11911203. (69) M. Peifer, M.; Polakis, P. Science 2000, 287, 1606-1609. (70) Miller J. R.; Moon, R. T. J. Cell Biol. 1997, 139, 229-243. (71) Gottardi, C. J.; Gumbiner, B. M. J. Cell Biol. 2004, 167, 339-349. (72) Huber, A. H.; Weis, W. I. Cell 2001, 105, 391-402. (73) Gooding, J. M.; Yap, K. L.; Ikura, M. Bioessays 2004, 26, 497-511. (74) Shapiro, L. Nat. Struct. Biol. 2001, 8, 484-7. (75) Xing, Y.; Clements, W. K.; Le Trong, I.; Hinds, T. R.; Stenkamp, R.; Kimelman, D.; Xu, W. Mol. Cell 2004, 15, 523-533. (76) Xing, Y.; Clements, W. K.; Kimelman, D.; Xu, W. Genes Dev. 2003, 17, 2753-2764. (77) Spink, K. E.; Fridman, S. G.; Weis, I. W. EMBO J. 2001, 20, 62036212. (78) Lickert, H.; Bauer, A.; Kemler, R.; Stapper, J. J. Biol. Chem. 2000, 275, 5090-5095. (79) Roura, S.; Miravet, S.; Piedra, J.; Garcia de Herreros, A.; Dunach, M. J. Biol. Chem. 1999, 274, 36734-36740. (80) Faux, M. C.; Ross, J. L.; Meeker, C.; Johns, T.; Ji, H.; Simpson, R. J.; Layton, M. J.; Burgess, A. W. J. Cell Sci. 2004, 117, 427-439. (81) Pockley, A. G.; Exp. Rev. Mol. Med. 2001, http://www.expertreviews.org/01003556h.htm (82) Catimel, B.; Rothacker, J.; Nice, E. C. J. Biochem. Biophys. Methods 2001, 49, 289-312. (83) Shiyanov.; Nag, A. P.; Raychaudhuri, P. J. Biol. Chem. 1999, 274, 5309-35312. (84) Chen, X.; Zhang, Y.; Douglas, L.; Zhou, P. J. Biol. Chem. 2001, 276, 48175-48182. (85) Hu, J.; McCall, C. M.; Ohta, T.; Xiong, Y. Nat. Cell. Biol. 2004, 6, 1003-1009. (86) Chen, L. C.; Manjeshwar, S.; Lu, Y.; Moore, D.; Ljung, B. M.; Kuo, W. L.; Dairkee, S. H.; Wernick, M.; Collins, C.; Smith, H. S. Cancer Res. 1998, 58, 3677-3683. (87) Yasui, K.; Arii, S.; Zhao, C.; Imoto, I.; Ueda, M.; Nagai, H.; Emi, M.; Inazawa, J. Hepatology 2002, 35, 1476-1484. (88) Higa, L. A.; Mihaylov, I. S.; Banks, D. P.; Zheng, J.; Zhang, H. Nat. Cell. Biol. 2003, 5, 1008-10015. (89) Wertz, I. E.; O’Rourke, K. M.; Zhang, Z.; Dornan, D.; Arnott, D.; Deshaies, R. J.; Dixit, V. M. Science 2004, 303, 1371-1374. (90) Ulane, C. M.; Horvath, C. M. Virology 2002, 304, 160-166. (91) Choi, J.; Park, S. Y.; Costantini. F.; Jho, E. H.; Joo, C. K. J. Biol. Chem. 2004, 279, 49188-49198.

PR050132X