Development of a Convenient Competitive ELISA for the Detection of

Dec 6, 2012 - (2, 3) However, porcine tissue expresses a high proportion of the Gal-α-(1,3)-Gal epitope and elicits a vigorous anti-Gal-α-(1,3)-Gal ...
1 downloads 13 Views 790KB Size
Article pubs.acs.org/ac

Development of a Convenient Competitive ELISA for the Detection of the Free and Protein-Bound Nonhuman Galactosyl-α-(1,3)Galactose Epitope Based on Highly Specific Chicken Single-Chain Antibody Variable-Region Fragments Stephen Cunningham, Emily Starr, Iain Shaw, John Glavin, Marian Kane, and Lokesh Joshi* Glycoscience Group, National Centre for Biomedical Engineering Science, National University of Ireland, Galway, Ireland S Supporting Information *

ABSTRACT: The presence of the nonhuman galactosyl-α-(1,3)galactose (Gal-α-(1,3)-Gal) carbohydrate epitope on a number of recombinant therapeutic proteins has recently been reported, renewing interest in this immunogenic carbohydrate epitope. It is well-known that this motif is the primary contributing factor in hyperacute rejection of porcine organ xenograft, due to the existence of natural antibodies against this epitope in human serum. Though the number of epitopes on recombinant glycoproteins may be low when compared directly to whole tissue, circulating anti-Gal-α-R immunoglobulins can still induce anaphylaxis. Therefore, there is a need for rapid and convenient methods for detection and monitoring of this epitope in biopharmaceuticals produced in recombinant mammalian systems. To this end, we have generated immune-challenged chicken single-chain antibody variable-region fragment (scFv) libraries targeting the Galα-(1,3)-Gal motif and have selected a panel of scFv's that bind the target. We have used one of these antibodies to develop a competitive ELISA for both free and protein-bound Gal-α-(1,3)-Gal and have demonstrated that the ELISA is specific for the target and can be used to determine the loading of the target on glycoproteins. This competitive ELISA will provide a convenient method of detecting and quantifying Gal-α-(1,3)-Gal on therapeutic glycoproteins.

T

epitope required to induce anaphylaxis remain undetermined and may be both product-specific and patient-dependent. The production of recombinant antibodies and related biopharmaceuticals is now seen as a well-established and a routinely performed industrial process, as evidenced by the increase in biosimilar production plants over recent years. Control over glycosylation is a critical factor during recombinant therapeutic production because of its profound effect on protein function, allergenic and immunogenic properties, plasma clearance rates, and efficacy.11,12 It is now accepted that specific glycan structures adversely affect the safety of these recombinant products.10,11 The mammalian cell lines utilized for production, most commonly murine-derived cell lines SP2/0 and NS0 or Chinese hamster ovary (CHO) cells, permit “humanlike” glycosylation to occur.13 With the identification of Gal-α-(1,3)-Gal on recombinant proteins currently on the market, there is a need for convenient and rapid analytical approaches to monitor and quantify the levels of Gal-α-(1,3)-Gal on existing and future recombinant therapeutics for human use.14,15 Furthermore, the possibility of exploiting the Gal-α-(1,3)-Gal epitope/natural human

he immunological importance of the nonhuman galactosyl-α-(1,3)-galactose (Gal-α-(1,3)-Gal) epitope was initially highlighted by its direct involvement in hyperacute rejection (HAR) of porcine organ xenografts.1 Xenotransplantation has been widely considered as a direct means of overcoming the critical shortage of human donor organs, with porcine organs, in particular, being considered the most suitable.2,3 However, porcine tissue expresses a high proportion of the Gal-α-(1,3)-Gal epitope and elicits a vigorous anti-Gal-α(1,3)-Gal antibody response following transplantation of porcine xenografts into humans. 4 Undesirable immune responses to this epitope are also thought to be involved in the suboptimal clinical outcomes after implantation of acellular tissue matrices.5 Naturally occurring human antibodies directed against the Gal-α-(1,3)-Gal epitope are widespread in humans,6 comprising up to 3% of immunoglobulin (Ig) in human sera (mainly the IgG2 subclass).7,8 Their induction by environmental stimuli, commensal bacteria, and/or parasitic interaction has been proposed, and they putatively exert a natural barrier function.8,9 As well as its involvement in HAR, the presence of this immunogenic glycan moiety has also been attributed to an IgE-mediated allergic/hypersensitivity response observed in patients taking recombinant monoclonal antibody preparations, e.g., Erbitux (Cetuximab).10,11 The levels of Gal-α-(1,3)-Gal © 2012 American Chemical Society

Received: September 6, 2012 Accepted: December 6, 2012 Published: December 6, 2012 949

dx.doi.org/10.1021/ac302587q | Anal. Chem. 2013, 85, 949−955

Analytical Chemistry

Article

chicken, and expression of phage−scFv was induced using standard protocols.26 More details on these methods are provided in the Supporting Information (sections SI-1 to SI-4). Selection by Biopanning of scFv−Phage Libraries against Gal-α-(1,3)-Gal-BSA. Biopanning against Gal-α(1,3)-Gal-BSA was performed using Maxisorp immunotubes (Nunc). Briefly, tubes were coated overnight at 4 °C with 1 mL of 10 μg/mL Gal-α-(1,3)-Gal-BSA (Dextra, Reading, U.K.) and then blocked with phosphate-buffered saline (PBS), pH 7.2, containing 0.5% BSA (PBS-BSA). ScFv-displaying phage (1 × 1011 to 1 × 1012) suspended in 1 mL of PBS containing 1% ovalbumin (PBS-O) were then added and mixed by rotation for 1 h at room temperature, followed by repeated washing with PBS containing 0.1% Tween 20 (PBS-T), and then with PBS only. Bound phage were eluted with the addition of 1 mL of 100 mM glycine−HCl, pH 2.2, for 10 min. Eluted phage were neutralized using 500 μL of 1 M Tris−HCl, pH 8.8. Propagation of eluted phage and further manipulations were then carried out as previously described.26 Three rounds of panning were performed, with increasing selection stringency mediated by a progressive increase in washing steps from 10 in pan 1 to 15 in pans 2 and 3. ELISA Analysis of the scFv−Phage Population Pool after Each Round of Biopanning. Eluted scFv-displaying phage suspensions from each round of panning were assessed for Gal-α-(1,3)-Gal binding by direct ELISA analysis. The procedure was as described for analysis of serum response (Supporting Information, section SI-2), except that plates were coated with Gal-α-(1,3)-Gal-BSA, phage−scFv preparations were diluted 3-fold in PBS-BSA containing 0.1% Tween 20 (PBS-BT) before addition to the wells, and bound phage were detected by probing with horseradish peroxidase (HRP)conjugated anti-M13 antibody (GE Healthcare) in PBS-BT for 1 h at 37 °C. Single scFv−Phage Clones and Direct ELISA Analysis. Ninety-six random clones were picked from output plates from rounds 2 and 3 of panning and cultured in deep-well plates in a 1.5 mL volume and individual scFv-displaying phage rescued as described in section SI-4 in the Supporting Information. Binding of single scFv−phage to Gal-α-(1,3)-Gal-BSA was then determined in direct ELISA as described previously. Establishment of Individual scFv Clones for Stable Expression and Purification. Colonies selected on the basis of observed binding to Gal-α-(1,3)-Gal-BSA were subjected to phagemid purifications (NucleoBond, Macherey-Nagel). The plasmid preparations were transformed by electroporation into Escherichia coli strain TOP 10F′ (Invitrogen) and were then plated on Luria−Bertani (LB) carbenicillin agar to provide single colonies. Expression and purification of soluble scFv from individual colonies is described in section SI-5 in the Supporting Information. DNA Sequencing of Single scFv Clones. Plasmid DNA isolated from the bacterial pellets was subjected to multiple read sequencing reactions for both strands of the complete scFv inserts (Eurofins MWG operon). Multiple sequence alignment, translation, and recognition of complementarity-determining regions (CDRs) was performed using BLAST (http://www. ncbi.nlm.nih.gov/BLAST), ClustalW2 from EBI (http://www. ebi.ac.uk/Tools/msa/), and the CDR repository held at http:// www.bioinf.org.uk/abs. Evaluation of the Specificity of Purified scFv by Direct ELISA. Initially, a range of scFv concentrations were tested in direct ELISA against 10 μg/mL immobilized Gal-α-(1,3)-Gal-

antibody system to improve the efficacy of autologous vaccines is gaining increased attention recently, and this will also demand convenient assay tools.16 Specific binding assays provide robust analytical platforms once high-affinity and specific binding agents are available for the analytes. Lectins are the most widely used affinity reagents for carbohydrates, with less reliance on antibodies, due to limited availability of high-quality antibodies.17 Two lectins are commonly used for the detection of the Gal-α-(1,3)-Gal motif. Griffonia simplicifolia I isolectin B4 (GS-I-B4) detects terminal Gal-α-1-R (α-galactosyl residues, termed α-Gal) epitopes,18 but cannot distinguish between structures in which the terminal galactose is linked to different carbon atoms in the penultimate galactose on the carbohydrate chain (e.g., Gal-α-(1,2)-R, Gal-α(1,3)-R, or Gal-α-(1,4)-R). Binding of GS-I-B4 may also depend on whether the Gal-α-(1,3)-Gal is on a cell surface or on an isolated glycoprotein, as has been reported for a number of glycan recognition molecules.19−22 Marasmius oreades agglutinin (MOA) is also known to bind with Gal-α-1-Rterminated structures. Although a number of anti-Gal-α-(1,3)Gal antibodies have been described, only a small number are commercially available, including a polyclonal antibody raised in baboon and the M86 mouse IgM monoclonal antibody, which have found limited application to date.23,24 There are still significant challenges in the generation of high-quality antibodies targeting carbohydrate motifs because of their low immunogenicity, yet human serum and the serum of many animals contain a wide range of natural anticarbohydrate antibodies.17 Engineered single-chain antibody fragment (scFv) libraries generated from immunoglobulin cDNA, whether from naive or immune-challenged host systems, may provide access to these antibodies. Chickens, like humans, do not produce the Gal-α-(1,3)-Gal epitope and hence develop a strong immune response on exposure to this motif.25 The generation of chicken antibody libraries has been shown to be simpler than that of libraries from mammalian species, due to the peculiar mechanism of immunoglobulin gene diversification in birds.26 Chickens possess single functional immunoglobulin heavy-chain-variable region (VH) and light-chain-variable region (VL) genes, with diversity created by the high-frequency gene conversion mechanism operating continuously during B cell proliferation in the bursa. Here we describe the generation of a Gal-α-(1,3)-Gal-targeted phage-displayed scFv library and isolation of chicken scFv antibody fragments directed against this epitope. These scFv's were shown to be highly specific for the detection of the Gal-α-(1,3)-Gal motif when tested in direct ELISA format against a panel of related neoglycoconjugates. The scFv's were also used in competitive ELISA format, where they allowed the concentration of Gal-α-(1,3)-Gal to be determined in free solution and when present on the surface of a glycoprotein. To our knowledge, this is the first report of a panel of engineered scFv's against the nonhuman carbohydrate Gal-α-(1,3)-Gal motif and most importantly the first report of a convenient competitive ELISA for detection of this motif on glycoproteins.



MATERIALS AND METHODS Chicken Immunization and Phage-Display Library Generation. Two chickens were immunized with a mixture of glycoproteins, including the neoglycoprotein Gal-α-(1,3)Gal-BSA (BSA = bovine serum albumin), using a standard immunization regime, and serum responses were determined by direct ELISA. The scFv libraries were constructed from each 950

dx.doi.org/10.1021/ac302587q | Anal. Chem. 2013, 85, 949−955

Analytical Chemistry

Article

Figure 1. Schematic alignment of CDRs from the VL and VH regions from six scFv sequenced clones demonstrating amino acid variants. CDRs were identified by locating flanking conserved regions listed at http://www.bioinf.org.uk/abs. The linker region is given in blue. Amino acid substitutions and sequence variation are highlighted in yellow.

gentibiose, D-(+)-trehalose, D-(+)-turanose, L-fructose, D(+)-glucose, sucrose, galacto-N-biose, and 4β-galactobiose) and Dextra (D-cellibiose, Gal-α-(1,3)-Gal-β-(1,4)-Gal, Gal-α(1,3)-Gal-β-(1,4)-Glc, Gal-α-(1,3)-Gal-β-(1,4)-Gal-α-(1,3)-Gal, laminarbiose, Xyl-β-(1,4)-Xyl (xylobiose), GlcNAc, galactose, and LacNAc) analyzed at a range of concentrations. The percent cross-reactivity was determined by comparing the ED50 (the 50% effective dose, i.e., the dose of free sugar causing 50% displacement of the scFv) values. Analysis of Glycoproteins in the Competitive ELISA. The neoglycoproteins Gal-α-(1,3)-Gal-BSA and Gal-α-(1,3)Gal-β-(1,4)-GlcNAc-HSA were prepared at concentrations ranging from 20 μg/mL to 5 mg/mL in PBS, pH 7.2, and assayed in the competitive ELISA. Two glycoproteins, Erbitux (Cetuximab), a recombinant antibody produced in murine cell lines (Merck, Middlesex, U.K.; lot no. 133380), and the natural glycoprotein laminin, from the basement membrane of Engelbreth−Holm−Swarm (EHS) mouse sarcoma (Sigma, Poole, U.K.; lot no. L2020-031M4058), were assayed at concentrations ranging from 250 μg/mL to 5 mg/mL and 20 μg/mL to 10 mg/mL, respectively. Both proteins in native form were found to be too viscous to be analyzed in the assay, but gave reliable results following heat treatment at 70 °C for 10 min. Previously, laminin was enzymatically digested prior to lectin profiling to permit glycan recognition.29

BSA to determine the optimal subsaturating concentration of each scFv. The specificity of all three scFv's was assessed at a concentration of 4 μg/mL against a panel of 13 neoglycoconjugates (NGCs) comprising Gal-α-(1,3)-Gal-BSA, Gal-α-(1,3)-Gal-β-(1,4)-N-acetyl-D-glucosamine-human serum albumin (Gal-α-(1,3)-Gal-β-(1,4)-GlcNAc-HSA), Gal-α-phenylisothiocyanate-BSA (Gal-α-PITC-BSA), Gal-β-PITC-BSA, Gal-α-(1,2)-Gal-BSA, Gal-β-(1,4)-Gal-BSA, xylosyl-β-BSA (Xyl-β-BSA), fucosyl-α-4-aminophenyl-BSA (Fuc-α-4APBSA), Fuc-β-4AP-BSA, N-acetylneuraminic acid-4AP-BSA (Neu5Ac-4AP-BSA), GlcNAc-β-BSA, lactosyl-β-4AP-BSA (Lac-β-4AP-BSA), and N-acetyl-D-lactosamine-BSA (LacNAcBSA) and two negative controls, HSA and BSA. All conjugate preparations were purchased from Dextra or synthesized within the laboratory.27,28 The direct ELISA was performed as described previously with the exception that HRP-conjugated anti-hemagglutinin antibody (Roche, West Sussex, U.K.) was used for detection of soluble scFv. Biotinylated GS-I-B4 lectin (EY Laboratories Inc., San Mateo, CA) and monoclonal antibody, M86, raised against the α-Gal epitope (Gal-α-(1,3)Gal-β-(1,4)-GlcNAc-R) (Enzo Life Sciences, Inc.) were evaluated in parallel, using streptavidin-conjugated HRP (Thermo Fisher Scientific) and a polyclonal rabbit antimouse Ig HRP (P 0260, Dako), respectively. Development of scFv-Based Competitive ELISAs for Gal-α-(1,3)-Gal. Each scFv was evaluated in a competitive ELISA format using free Gal-α-(1,3)-Gal as the standard. Microtiter plate wells were coated and blocked as previously described. Standard solutions were prepared in PBS (pH 7.2), ranging in concentration from 2 μg/mL (5.84 μM) to 20 mg/ mL (58.4 mM). For the assay, 50 μL aliquots of each standard and buffer blank were added to designated wells in triplicate, followed by 50 μL of a subsaturating concentration (1 μg/mL) of the appropriate scFv, giving a final standard concentration in the assay of 1 μg/mL (2.92 μM) to 10 mg/mL (29.2 mM). The plate was incubated for 1 h at 37 °C with shaking. After washing, bound scFv was detected as previously described. The ratios of the absorbance for each standard versus the absorbance given by the blank (B/Bo) were plotted versus the standard concentration, and a standard curve was plotted using SigmaPlot (v11, Systat Software Inc.). The detection limits were determined as the concentration corresponding to the mean response for the zero standard minus 3 times the standard deviation (SD). Specificity of the Competitive ELISA. The specificity of the competitive ELISA given by each scFv was determined by comparing the displacement of scFv shown by the standard Gal-α-(1,3)-Gal with that shown by a panel of 21 sugars from Sigma-Aldrich (melibiose, D-raffinose, lactulose, palatinose, β-



RESULTS Immune Response and Library Construction. Analysis of the serum antibody response following immunization with Gal-α-(1,3)-Gal-BSA indicated a strong response to the target glycan motif in both immunized chickens (Figure S-1a, Supporting Information). ScFv libraries displayed on filamentous phage were generated from the combined RNA extracted from the spleen and bone marrow of each chicken.26 Colony polymerized chain reaction (PCR) revealed that all individual clones tested contained full-length inserts, with sequence diversity confirmed by BstN1 restriction mapping of amplified products (Figure S-1b). The initial size of the two libraries was estimated at approximately 5 × 107 transformants. Libraries were combined and panned as a single scFv−phage library. Isolation of Anti-Gal-α-(1,3)-Gal scFv−Phage Particles. Three rounds of biopanning were performed at room temperature against Gal-α-(1,3)-Gal-BSA. Following the second round of panning, the presence of an enriched population of phage-displayed scFv's demonstrating binding to Gal-α-(1,3)-Gal-BSA was confirmed by direct ELISA (Figure S-1c, Supporting Information). No further enrichment was observed after the third round, and there was minimal cross reactivity against the carrier protein BSA or HSA. From the 951

dx.doi.org/10.1021/ac302587q | Anal. Chem. 2013, 85, 949−955

Analytical Chemistry

Article

output phage of the second and third panning rounds, 96 individual phage clones were selected at random, cultured individually, and analyzed for specific binding activity. Of the clones tested, 74 showed binding greater than 3 times the background to the immobilized NGC (Figure S-2, Supporting Information). Six of the higher binding clones (A4, A11, D9, F1, G12, and H3) were chosen for sequencing, and their encoding phagemids were transformed into TOP10 cells for inducible expression and generation of soluble scFv for downstream affinity purification. Sequence Analysis of Isolated Clones. Analysis of the nucleotide sequences of the VH and VL regions of the six scFv's demonstrated a high degree of consensus with germline chicken VH and VL regions. All clones exhibited considerable divergence in the VL CDR1, CDR2, and CDR3, with less variability in the VH region. Indeed, the VH CDR1 was conserved across all clones (Figure 1). The sequence data suggest that the scFv's isolated were generated in an antigen-driven response, with particular reference to the VL regions, rather than being naive antibody sequences. Four unique sequences, representing clones A11/D9, F1/G12, H3, and A4, were identified (Figure 1). Within the VL CDRs, single transitions occurred in a number of positions: (i) VL CDR1 where four codon transitions were identified alternating from tyrosine (Y) to histidine (H) to asparagine (N) to lysine (K), (ii) in VL CDR2 where asparagine (N) to glutamine (Q) to lysine (K) occurred, and (iii) VL CDR3 where threonine (T) to serine (S) to glycine (G) to alanine (A) was observed. In contrast, fewer polymorphic sites were observed across the VH regions, with A11/D9, H3, and A4 showing consensus across CDR1, CDR2, and CDR3. F1/G12 differed from the other clones within CDR3. Across the linker region, two substitutions from the germline sequence were observed (serine (S) to glutamine (Q) and glycine (G) to serine (S)) along with two serine (S) deletions shortening the linker region of A11/D9. Such observations of linker region alterations have been reported within similar libraries previously with no effect on library efficiency.30 Expression and Purification of Soluble scFv. The four unique scFv clone sequences represented by A4, A11, G12, and H3 were carried forward for expression and purification of soluble scFv. Three were successfully expressed in bacterial culture (A4, A11, and G12) and purified using Ni−NTA (nitrilotriacetic acid) affinity chromatography, and the purity was confirmed by gel electrophoresis (Figure S-3, Supporting Information). Yields of purified scFv were approximately 6.5 mg/L of bacterial culture medium. All purified scFv demonstrated stability at 4 °C for periods of up to 6 months in PBS. One clone, H3, failed to purify under both native and denatured conditions. ScFv Specificity Evaluation by Direct ELISA. The specificities of the purified scFv clones A4, A11, and G12 were tested in a direct ELISA format against a range of Gal-αR-related NGCs. All three scFv's were specific for Gal-α-(1,3)Gal-BSA, binding with slightly higher intensity to Gal-α-(1,3)Gal-β-(1,4)-GlcNAc-HSA compared to Gal-α-(1,3)-Gal-BSA (Figure 2). The binding to the disaccharide and trisaccharide epitopes and not to the monosaccharide Gal-α-PITC-BSA or the Gal-α-(1,2)-Gal-BSA and Gal-α-(1,4)-Gal-BSA structures demonstrates that the α-(1,3) linkage is a requirement for binding. Similar binding specificities for the Gal-α-(1,3)-Gal epitope were shown by all three scFv's tested, regardless of sequence diversity in the CDRs (Figure 1). No binding was

Figure 2. Specificity profile of the three anti-Gal-α-(1,3)-Gal scFv's in direct ELISA. Wells were coated as follows: (1) Gal-α-(1,3)-Gal-BSA, (2) Gal-α-(1,3)-Gal-β-(1,4)-GlcNAc-HSA, (3) Gal-α-PITC-BSA, (4) Gal-β-PITC-BSA, (5) Gal-α-(1,2)-Gal-BSA, (6) Gal-β-(1,4)-Gal-BSA, (7) Xyl-β-4AP-BSA, (8) Fuc-α-4AP-BSA, (9) Fuc-β-4AP-BSA, (10) Neu5Ac-4AP-BSA, (11) GlcNAc-β-BSA, (12) Lac-β-4AP-BSA, (13) LacNAc-BSA, (14) HSA, (15) BSA, and (16) BSA-4AP. The error bars indicate the standard error of the mean of triplicate determinations.

observed to the other structures tested in the direct ELISA (Gal-α-PITC-BSA, Gal-β-PITC-BSA, Gal-α-(1,2)-Gal-BSA, Gal-α-(1,4)-Gal-BSA, Xyl-β-BSA, Fuc-α-4AP-BSA, Fuc-β-4APBSA, Neu5Ac-4AP-BSA, GlcNAc-β-BSA, Lac-β-4AP-BSA, LacNAc-BSA) or to the carrier proteins HSA, BSA, and BSA4AP. By comparison, the lectin GS-I-B4 showed significantly higher binding to the trisaccharide NGC compared to the disaccharide NGC and also bound to all Gal-α-R-containing NGCs (Figure S-4, Supporting Information), in consensus with previous reports.23,31 Therefore, the scFv's generated demonstrate a higher specificity than the GS-I-B4 lectin for Gal-α(1,3)-Gal. In parallel with the lectin GS-I-B4 and the monoclonal M86, all scFv's were able to detect the target motif on the natural glycoprotein murine laminin in direct ELISA (data not shown). Murine laminin is known to display the Gal-α-(1,3)-Gal motif.32 ScFv-Based Competitive ELISA. Competitive ELISAs were established with each of the three scFv's examined and free Gal-α-(1,3)-Gal disaccharide as the standard over the final assay concentration range of 1 μg/mL to 10 mg/mL (2.92 μM to 29.2 mM). The curves given by scFv's G12 and A4 were very similar with ED50 values of 158 and 311 μg/mL (0.46 and 0.91 mM), respectively (Figure 3). ScFv A11 gave a less sensitive standard curve (ED50 = 1225 μg/mL, 3.58 mM; Figure 3). No standard curve could be generated with either the lectin GS-IB4 or monoclonal M86 within the same concentration range. A panel of 21 related sugars were tested in the scFv-based competitive assays. Of the sugars tested, only those bearing the Gal-α-(1,3)-Gal epitope demonstrated inhibition at detectable levels. The concentrations of the Gal-α-(1,3)-Gal-containing trisaccharides and tetrasaccharide that caused 50% displacement (i.e., the ED50) were lower for all three scFv's compared to the ED50 for the Gal-α-(1,3)-Gal disaccharide (Table 1), especially with scFv A11. This suggested that the scFv's had a higher affinity for the larger sugars than for the disaccharide. No inhibition was observed in the presence of any monosaccharide within the ranges indicated, supporting the conclusion from the direct assay that monosaccharide structures are not capable of inhibiting binding of any of the scFv's to the Gal-α-(1,3)-Gal structure. ScFv A4 gave the most consistent readings and was selected for further evaluation. Figure 4 shows the A4 composite standard curve, giving an ED50 of 327 μg/mL (0.955 mM) and a detection limit of 3.9 ng/mL (∼10 nM). 952

dx.doi.org/10.1021/ac302587q | Anal. Chem. 2013, 85, 949−955

Analytical Chemistry

Article

Figure 3. Standard curves for competitive ELISA given by scFv clones G12, A4, and A11. Plates were coated with immobilized Gal-α-(1,3)Gal-BSA (10 μg/mL) and incubated with a 2.5 μg/mL concentration of the scFv's in the presence of increasing concentrations of free Gal-α(1,3)-Gal. The x-axis shows the final Gal-α-(1,3)-Gal concentration in the assay well.

Figure 4. Composite standard curve given by scFv A4 from nine independent assays. The x-axis shows the final Gal-α-(1,3)-Gal concentration in the assay well, and error bars indicate the standard error of the mean.

Table 2. Comparison of the Molar Ratios of Gal-α-(1,3)-Gal to Protein in Selected Glycoproteins

Table 1. Specificity Study of the Three Competitive ELISAs Based on the Anti-Gal-α-(1,3)-Gal-BSA scFv's

Gal-α-(1,3)-Gal:protein molar ratio

ED50 (mM) free sugar

Mr

scFv A11

Gal-α-(1,3)-Gal Gal-α-(1,3)-Gal-β-(1,4)-Gal Gal-α-(1,3)-Gal-β-(1,4)-Glc Gal-α-(1,3)-Gal-β-(1,4)-Gal-α-(1,3)Gal

342.3 504.4 504.4 666.5

3.58 0.08 0.07 0.14

scFv A4

scFv G12

0.91 0.16 0.12 0.26

0.46 0.21 0.20 0.12

glycoprotein

competitive ELISA

MALDI/other

Gal-α-(1,3)-Gal-BSA Gal-α-(1,3)-Gal-BSA laminin Erbitux

9.4 19.7 38.9 3.9

16 (10−25)a 23 (15−31)a NDb 1.13c

a

Mean (range) determined by MALDI analysis. bThe Gal-α-(1,3)-Gal content of native laminin has not been reported. cValue determined following detergent denaturation and enzymatic release of oligosaccharides.33

The assay was able to detect the Gal-α-(1,3)-Gal motif when protein bound in the form of NGCs and on both the glycoproteins tested, Erbitux and laminin (Table 2). The disaccharide NGC interacted with the scFv's in a manner similar to that of the free sugar, as indicated by the parallelism of the response obtained over the concentration range of 0.25− 1 mg/mL (data not shown). Table 2 gives the disaccharide:protein molar ratios of the glycoproteins, estimated from the results of the competitive ELISA, compared to reported values where available. The ELISA-estimated values for the NGCs were very close to or within the range of the MALDI results provided by the manufacturers. Both laminin and Erbitux in their native state were too viscous to be analyzed in the competitive ELISA. However, heat treatment reduced the

viscosity, and assay results indicated that there were 39 mol of the Gal-α-(1,3)-Gal residue/mol of murine laminin and approximately 3.9 mol/mol of Erbitux, in the same range as recently reported by Lammert et al.33 The content of Gal-α(1,3)-Gal in murine laminin has not previously been reported.



DISCUSSION Although known for over 30 years, there is still considerable interest in the Gal-α-(1,3)-Gal motif, commonly found as a terminal motif on glycoproteins and glycolipids in a range of species, with the exception of humans and chickens, which lack the enzyme involved in the synthesis of the disaccharide.34 This 953

dx.doi.org/10.1021/ac302587q | Anal. Chem. 2013, 85, 949−955

Analytical Chemistry

Article

however, as all were highly specific for carbohydrates containing the Gal-α-(1,3)-Gal motif in direct assay. Competitive binding assays are widely used for determination of low molecular weight compounds in complex samples, with often no viable alternative. They represent a versatile, robust assay configuration, being adaptable for highthroughput or single-use, point-of-care formats. Competitive assays are, however, demanding in terms of the binding agent, with assay sensitivity being directly dependent on the affinity of the binder for the analyte.48 We have demonstrated that all three scFv's that were purified could be used in competitive assays for Gal-α-(1,3)-Gal, allowing detection of the disaccharide at a low nanogram per milliliter level (detection limit estimated at 3.9 ng/mL, ∼10 nM), which exceeds reported detection limits of most lectins, typically at the microgram per milliliter level, when tested in similar plate assays.49 The exceptional specificity of the scFv's was also retained in this format. We have shown that the assay can detect the Gal-α-(1,3)-Gal motif in the context of a glycoprotein and indeed have shown that the assay can be used to estimate the quantity of the motif within a native glycoprotein, laminin, and a recombinant glycoprotein preparation, Erbutix. Thus, the assay provides the option of either direct analysis of the motif on a glycoprotein of interest or measurement of the free disaccharide in solution.

glycan structure still remains relevant to xenotransplantation efforts, and the potential to exploit the relatively large quantities of natural anti-Gal-α-(1,3)-Gal antibodies present in human serum to increase immunogenicity in cancer immunotherapy has been widely explored.16,35 Modification of cancer cells or cancer-associated molecules, such as mucins, with the Gal-α(1,3)-Gal epitope has been shown to increase uptake by antigen-presenting cells and enhance the immune response.35,36 The ability to measure the loading of the motif on cells is important for this work, but the recently reported occurrence of this immunogenic motif on glycoprotein biopharmaceuticals, with resulting adverse reactions in patients, has particularly highlighted the need for specific binding agents and a convenient analytical method for Gal-α-(1,3)-Gal.10,37 The lectin GS-1-B4, which is specific for α-Gal residues, is most commonly used for the detection of the Gal-α-(1,3)-Gal epitope.18 However, GS-1-B4 binds to a range of terminal, nonreducing α-linked Gal residues and is therefore not a reliable marker for Gal-α-(1,3)-Gal.19 The lectin from mushroom MOA has been reported to be more specific for Gal-α(1,3)-Gal and Gal-α-(1,3)-Gal-β-(1,4)-GlcNAc epitopes, but yet binds strongly to the B blood group antigen, which contains an L-Fuc α-(1,2)-linked to the Gal at the reducing end.38 Naturally occurring antibodies in human serum against this and other nonhuman glycan epitopes have not been reliably used as analytical tools.17,39 Traditional immunization approaches are not very reliable for generation of high-affinity antibodies against specific carbohydrate moieties and do not often yield antibodies with characteristics suitable for assay development. The use of phage-display technology with recombinant antibody fragment libraries offers more promise, giving access to a larger repertoire of antibodies, including anti-selfantibodies that would not normally be available due to tolerization. However, only a few successful examples have been published to date.40−44 Here, we describe the generation of phage-displayed scFv libraries from chickens immunized against the Gal-α-(1,3)-Gal epitope and the isolation of a panel of chicken scFv's targeting the motif that were successfully used to develop the first competitive ELISA capable of detecting and measuring the motif in free or protein-bound form. The selection of the chicken for immunization and library generation was based upon the known absence and resulting immunogenicity of this epitope in chicken.45 Phage-displayed scFv library generation from chickens also means a much smaller number of primer sets are needed for amplification of the antibody gene repertoire than with most other species, because of the particular arrangement of immunoglobulin genes in the avian.26 Immune chicken scFv libraries are also becoming increasingly popular for production of recombinant antibodies against a variety of mammalian protein targets, because of evolutionary distance.30,46 The rapid selection of specific scFv's from the libraries generated suggests a strong representation of antibodies against the Gal-α-(1,3)-Gal epitope in the libraries and demonstrates the advantages of using immune libraries where possible when isolating specific scFv's. It was interesting to observe that the major differences in the scFv sequences obtained were in the light chain rather than the heavy chain. This supports the importance of the light chain in determining the performance characteristics of chicken scFv's, in agreement with previous reports of improvements in scFv affinities following light chain shuffling.30,47 The differences in light chain sequences did not influence the specificity of the scFv's,



CONCLUSION This is to our knowledge the first report of a competitive assay for a small carbohydrate motif in both free and conjugated forms, opening the way for a new approach to carbohydrate analysis. This assay provides a level of specificity not possible with conventional methods of analysis, such as HPLC/HILIC (hydrophilic interaction liquid chromatography) with exoglycosidase digestion arrays or LC−mass spectrometry.50 The compatibility of competitive binding assays with complex samples and the stability of the isolated scFv's mean that this assay could provide a convenient method for routine screening of recombinant therapeutics, with a major benefit for the biopharmaceutical industry and enhanced safety for patients. It is also expected that the scFv's generated will contribute to ongoing research on the Gal-α-(1,3)-Gal xenotransplantation antigen, its role in many parasitic diseases, and its novel exploitation in tumor vaccine development.16,51,52



ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Phone: +353 91 495768. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The research leading to these results has received funding from Bristol-Myers Squibb (BMS), Syracuse, NY, the Industrial Development Authority, Ireland, under the Centre for BioAnalytical Sciences (CBAS) project (Code 116294), Science Foundation Ireland under Grant 08/SRC/B1393 for Alimentary Glycoscience Research Cluster (AGRC), and the 954

dx.doi.org/10.1021/ac302587q | Anal. Chem. 2013, 85, 949−955

Analytical Chemistry

Article

(27) Kilcoyne, M.; Gerlach, J. Q.; Kane, M.; Joshi, L. Anal. Methods 2012, 4, 2721−2728. (28) McBroom, C.; Samanen, C.; Goldstein, I. In Methods in Enzymology; Ginsburg, V., Ed.; Academic Press: New York, 1972; pp 212−219. (29) Rao, C. N.; Goldstein, I. J.; Liotta, L. A. Arch. Biochem. Biophys. 1983, 227, 118−124. (30) Finlay, W. J. J.; Shaw, I.; Reilly, J. P.; Kane, M. Appl. Environ. Microbiol. 2006, 72, 3343−3349. (31) Kirkeby, S.; Moe, D. Xenotransplantation 2002, 9, 260−7. (32) Arumugham, R.; Hsieh, T.; Tanzer, M.; RA, L. Biochim. Biophys. Acta 1986, 833, 112−126. (33) Lammerts van Bueren, J. J.; Rispens, T.; Verploegen, S.; Van der Palen-Merkus, T.; Stapel, S.; Workman, L.; James, H.; Van Berkel, P.; Van de Winkel, J.; Platts-Mills, T.; Parren, P. Nat. Biotechnol. 2011, 29, 574−576. (34) Stanley, W.; Peters, B.; Blake, D.; Yep, D.; Chu, E.; Goldstein, I. Proc. Natl. Acad. Sci. U.S.A. 1979, 76, 303−307. (35) Manches, O.; Plumas, J.; Lui, G.; Chaperot, L.; Molens, J.; Sotto, J.; Bensa, J.; Galili, U. Haematologica 2005, 90, 625−634. (36) Deguchi, T.; Tanemura, M.; Miyoshi, E.; Nagano, H.; Machida, T.; Ohmura, Y.; Kobayashi, S.; Marubashi, S.; Eguchi, H.; Takeda, Y.; Ito, T.; Mori, M.; Doki, Y.; Sawa, Y. Cancer Res. 2010, 70, 5259−5269. (37) Pointreau, Y.; Commins, S.; Calais, G.; Watier, H.; Platts-Mills, T. Clin. Oncol. 2012, 30, 334. (38) Winter, H. C.; Mostafapour, K.; Goldstein, I. J. Biochemistry 2002, 277, 14996−15001. (39) Huflejt, M. E.; Vuskovic, M.; Vasiliu, D.; Xu, H.; Obukhova, P.; Shilova, N.; Tuzikov, A.; Galanina, O.; Arun, B.; Lu, K.; Bovin, N. Mol. Immunol. 2009, 46, 3037−49. (40) Matsumoto-Takasaki, A.; Horie, J.; Sakai, K.; Furui, Y.; Sato, R.; Kawakami, H.; Toma, K.; Takayanagi, A.; Shimizu, N.; FujitaYamaguchi, Y. Biosci. Trends 2009, 3, 87−95. (41) Schoonbroodt, S.; Steukers, M.; Viswanathan, M.; Frans, N.; Timmermans, M.; Wehnert, A.; Ladner, R. C.; Hoet, R. M.; Nguyen, M. Journal Immunol. 2008, 181, 6213−6221. (42) Sakai, K.; Shimizu, Y.; Chiba, T.; Matsumoto-Takasaki, A.; Kusada, Y.; Zhang, W.; Nakata, M.; Kojima, N.; Toma, K.; Takayanagi, A.; Shimizu, N.; Fujita-Yamaguchi, Y. Biochemistry 2007, 46, 253−62. (43) Zhang, W.; Matsumoto-Takasaki, A.; Kusada, Y.; Sakaue, H.; Sakai, K.; Nakata, M.; Fujita-Yamaguchi, Y. Biochemistry 2007, 46, 263−70. (44) Sakai, K.; Yuasa, N.; Tsukamoto, K.; Takasaki-matsumoto, A.; Yajima, Y.; Sato, R.; Kawakami, H.; Mizuno, M.; Takayanagi, A.; Shimizu, N.; Nakata, M.; Fujita-yamaguchi, Y. J. Biochem. 2010, 147, 809−817. (45) Parmentier, H. Poult. Sci. 2008, 87, 918−926. (46) Bowes, T.; Hanley, S. a; Liew, A.; Eglon, M.; Mashayekhi, K.; O’Kennedy, R.; Barry, F.; Taylor, W. R.; O’Brien, T.; Griffin, M. D.; Finlay, W. J. J.; Greiser, U. J. Biomol. Screening 2011, 16, 744−54. (47) Fitzgerald, J.; Lugovskoy, A. mAbs 2011, 3, 299−309. (48) Le Berre, M.; Kane, M. Anal. Lett. 2006, 39, 1587−1598. (49) Tateno, H.; Nakamura-Tsuruta, S.; Hirabayashi, J. Nat. Protoc. 2007, 2, 2529−37. (50) Doherty, M.; McManus, C.; Duke, R.; Rudd, P. Methods Mol. Biol. 2012, 899, 293−313. (51) Galili, U. Immunology Cell Biol. 2005, 83, 674−86. (52) Wohlschlager, T.; Butschi, A.; Zurfluh, K.; Vonesch, S. C.; auf dem Keller, U.; Gehrig, P.; Bleuler-Martinez, S.; Hengartner, M. O.; Aebi, M.; Künzler, M. J. Biol. Chem. 2011, 286, 30337−43.

European Union Seventh Framework Programme (FP7/2007− 2013) under Grant Agreement 260600 (“GlycoHIT”). We thank C. F. Barbas at the Scripps Research Institute for providing the phage-display pCOMB vectors used in this study and Dr. Michelle Kilcoyne, National University of Ireland, Galway, for conjugate preparations.



REFERENCES

(1) Joziasse, D. H.; Oriol, R. Biochim. Biophys. Acta 1999, 1455, 403− 18. (2) Halperin, E. C. Int. J. Cancer 2001, 96, 76−89. (3) Thompson, P.; Badell, I. R.; Lowe, M.; Turner, A.; Cano, J.; Avila, J.; Azimzadeh, A.; Cheng, X.; Pierson, R.; Johnson, B.; Robertson, J.; Song, M.; Leopardi, F.; Strobert, E.; Korbutt, G.; Rayat, G.; Rajotte, R.; Larsen, C. P.; Kirk, A. D. Am. J. Transplant. 2012, 12, 1765−75. (4) Chen, G.; Qian, H.; Starzl, T.; Sun, H.; Garcia, B.; Wang, X.; Wise, Y.; Liu, Y.; Xiang, Y.; Copeman, L.; Liu, W.; Jevnikar, A.; Wall, W.; Cooper, D. K. C.; Murase, N.; Dai, Y.; Wang, W.; Xiong, Y.; White, D. J.; Zhong, R. Nat. Med. 2005, 11, 1295−8. (5) Xu, H.; Wan, H.; Zuo, W.; Sun, W.; Owens, R. T.; Harper, J. R.; Ayares, D. L.; McQuilian, D. Tissue Eng., Part A 2009, 15, 1807−1819. (6) Macher, B.; Galili, U. Biochim. Biophys. Acta 2008, 1780, 75−88. (7) Parker, W.; Lin, S. S.; Yu, P. B.; Sood, A.; Nakamura, Y. C.; Song, A.; Everett, M. Lou; Platt, J. L. Glycobiology 1999, 9, 865−873. (8) Yu, P.; Holzknecht, Z.; Bruno, D.; Parker, W.; Platt, J. J. Immunol. 1996, 157, 5163−5168. (9) Mangold, A.; Lebherz, D.; Papay, P.; Liepert, J.; Hlavin, G.; Lichtenberger, C.; Adami, A.; Zimmermann, M.; Klaus, D.; Reinisch, W.; Ankersmit, H. J. TPS 2011, 43, 3964−3968. (10) Chung, C.; Mirakhur, B.; Chan, E.; Le, Q.; Berlin, J.; Morse, M.; Murphy, B.; Satinover, S.; Hosen, J.; Mauro, D.; Slebos, R.; Zhou, Q.; Gold, D.; Hatley, T.; Hicklin, D.; Platts-Mills, T. N. Engl. J. Med. 2008, 358, 1109−1117. (11) Dingermann, T. Biotechnol. J. 2008, 90−97. (12) Presta, L. G. J. Allergy Clin. Immunol. 2005, 116, 731−736. (13) Beck, A.; Wagner-Rousset, E.; Bussat, M.-C.; Lokteff, M.; Klinguer-Hamour, C.; Haeuw, J.-F.; Goetsch, L.; Wurch, T.; Dorsselaer, A. V.; Corvaia, N. Curr. Pharm. Biotechnol. 2008, 9, 482−501. (14) Read, E. K.; Park, J. T.; Brorson, K. A. Biotechnol. Appl. Biochem. 2011, 58, 213−9. (15) Bosques, C. J.; Collins, B. E.; Meador, J. W.; Sarvaiya, H.; Murphy, J. L.; Dellorusso, G.; Bulik, D. a; Hsu, I.-H.; Washburn, N.; Sipsey, S. F.; Myette, J. R.; Raman, R.; Shriver, Z.; Sasisekharan, R.; Venkataraman, G. Nat. Biotechnol. 2010, 28, 1153−6. (16) Posekany, K. J.; Wiley, J. E.; Gagnon, G. A. Anticancer Res. 2009, 29, 2387−2392. (17) Bovin, N.; Obukhova, P.; Shilova, N.; Rapoport, E.; Popova, I.; Navakouski, M.; Unverzagt, C.; Vuskovic, M.; Huflejt, M. Biochim. Biophys. Acta 2012, 1820, 1373−82. (18) Goldstein, I.; Winter, H. Subcell Biochem. 1999, 32, 127−141. (19) Kirkeby, S.; Moe, D. Immunol. Cell Biol. 2001, 4, 121−127. (20) Gerlach, J.; Kilcoyne, M.; Eaton, S.; Bhavanandan, V.; Joshi, L. In The Molecular Immunology of Complex Carbohydrates-3; Wu, A., Ed.; Springer: New York, 2011; pp 257−269. (21) Manimala, J. C.; Roach, T. a; Li, Z.; Gildersleeve, J. C. Glycobiology 2007, 17, 17C−23C. (22) Manimala, J. C.; Roach, T. a; Li, Z.; Gildersleeve, J. C. Angew. Chem. 2006, 45, 3607−10. (23) Gabius, H.; Kirkeby, S.; Andre, S. Clin. Biochem. 2004, 37, 36− 41. (24) Galili, U.; LaTemple, D.; Radic, M. Transplantation 1998, 65, 1129−32. (25) McKenzie, I.; Patton, K.; Smit, J.; Mouhtouris, E.; Xing, P.; Myburgh, J.; Sandrin, M. Transplantation 1999, 67, 8684−870. (26) Andris-Widhopf, J.; Rader, C.; Steinberger, P.; Fuller, R.; Barbas, C. F. J. Immunol. Methods 2000, 242, 159−81. 955

dx.doi.org/10.1021/ac302587q | Anal. Chem. 2013, 85, 949−955