Accepted Manuscript Title: An affinity improved single-chain antibody from phage display of a library derived from monoclonal antibodies detects fumonisins by immunoassay Author: Zu-Quan Hu He-Ping Li Ping Wu Ya-Bo Li Zhu-Qing Zhou Jing-Bo Zhang Jin-Long Liu Yu-Cai Liao PII: DOI: Reference:
S0003-2670(15)00190-7 http://dx.doi.org/doi:10.1016/j.aca.2015.02.014 ACA 233728
To appear in:
Analytica Chimica Acta
Received date: Revised date: Accepted date:
3-1-2015 6-2-2015 9-2-2015
Please cite this article as: Zu-Quan Hu, He-Ping Li, Ping Wu, Ya-Bo Li, Zhu-Qing Zhou, Jing-Bo Zhang, Jin-Long Liu, Yu-Cai Liao, An affinity improved single-chain antibody from phage display of a library derived from monoclonal antibodies detects fumonisins by immunoassay, Analytica Chimica Acta http://dx.doi.org/10.1016/j.aca.2015.02.014 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
An affinity improved single-chain antibody from phage display of a library derived from monoclonal antibodies detects fumonisins by immunoassay
Zu-Quan Hua,b,c, He-Ping Lia,d, Ping Wua,d, Ya-Bo Lia,d, Zhu-Qing Zhoud, Jing-Bo Zhanga,b, Jin-Long Liua,b, Yu-Cai Liaoa,b,e,*
Molecular Biotechnology Laboratory of Triticeae Crops, Huazhong Agricultural University, Wuhan 430070, China
College of Plant Science and Technology, Huazhong Agricultural University, Wuhan 430070, China
College of Biology & Engineering, Guizhou Medical University, Guiyang 550004, China
College of Life Science and Technology, Huazhong Agricultural University, Wuhan 430070, China
National Center of Plant Gene Research (Wuhan), Wuhan 430070, China
*Corresponding author: Tel./Fax: +86 27 8728 3008. E-mail:[email protected]
(Y.-C. Liao) Graphical abstract
An affinity-improved scFv antibody was compared to its parent monoclonal antibody.
scFv antibody-based detection of three fumonisin toxins in agricultural samples. Good agreement of novel antibody-based fumonisin detection and a chemical method. Favorable modeling of the scFv antibody structure complementary to fumonisins.
Fumonisin B analogs, particularly FB1, FB2, and FB3, are major mycotoxins found in cereals. Single-chain fragment variable (scFv) antibodies represent a promising alternative immunoassay system. A phage-displayed antibody library derived from four monoclonal antibodies (mAbs) generated against FB1 was used to screen high binding affinity scFv antibodies; the best candidate was designated H2. Surface plasmon resonance measurements confirmed that the H2 scFv displayed an 82-fold higher binding affinity than its parent mAb. Direct competitive enzyme-linked immunosorbent assay demonstrated that the H2 antibody could competitively bind to free FB1, FB2, and FB3, with an IC50 of 0.11, 0.04, and 0.10 μM, respectively; it had no cross-reactivity to deoxynivalenol, nivalenol and aflatoxin. Validation assays with naturally contaminated samples revealed a linear relationship between the H2 antibody-based assay results and chemical analysis results, that could be expressed as y = 1.7072x + 5.5606 (R2 = 0.8883). Homology modeling of H2 revealed a favorable binding structure highly complementary to the three fumonisins. Molecular docking analyses suggested that the preferential binding of the H2 scFv to FB2 was due to the presence of a hydrogen radical in its R1 position, leading to a proper electrostatic matching and hydrophobic interaction. The H2 scFv antibody can be used for the
rapid, accurate, and specific detection of fumonisin contamination in agricultural samples. Keywords: Fumonisin; Monoclonal recombinant antibody; Phage display; Binding kinetics; Direct competitive enzyme-linked immunosorbent assay 1. Introduction Fumonisins are a series of long-chain polyhydroxyl alkylamines esterified with two tricarballylic acid moieties (Fig. 1) that were first identified in 1988 from Fusarium verticillioides MRC826 isolated from maize . Twenty-one known phytopathogenic Fusarium species are known to frequently infect maize and other cereals in the field and during storage and produce fumonisins; F. verticillioides and F. proliferatum are the most problematic fumonisin producing species [2–4]. There are dozens of fumonisin analogs and isomers; these are separated into five groups: A, B, C, D, and P [3,5]. Among these, the fumonisin B analogs FB1, FB2, and FB3, are the most abundant naturally occurring mycotoxins. FB1 is the most prominent and is usually found at the highest levels in maize and other cultivated crops [3,6]. FB2 is produced massively higher quantities than FB1 by some newly identified Fusarium species that infect maize . FB1 can disrupt sphingolipid metabolism, leading to many animal diseases such as porcine pulmonary edema and equine leukoencephalomalacia [7,8]. FB1 is considered to stimulate the occurrence of human esophageal cancer [9–11] and liver cancer [11,12].The International Agency for Research on Cancer has classified FB1 as a class 2B toxic compound, possibly carcinogenic to humans . Fumonisins are heat-tolerant and can be stably present during food/feed processing and thus enter
into food/feed chains . To prevent these mycotoxins from entering food/feed chains and avoid the consumption of fumonisin-contaminated products, the rapid and accurate detection of fumonisins is essential. Several chemical techniques for the detection and analysis of fumonisins are available, but these all require expertise and advanced facilities and are thus limited to use in laboratories . Enzyme-linked immunosorbent assay (ELISA) technology provides a promising alternative, as it is portable and simple. Different polyclonal antisera [15,16] and monoclonal antibodies (mAbs) [17–21] have been developed for detecting fumonisin contamination in agricultural commodities. As compared to the labor-intensive and costly procedures of polyclonal and monoclonal antibody preparations, single-chain fragment variable (scFv) antibodies can be selected by phage display [22,23], expressed, and extracted simply through a bacterial expression system [24,25]. Moreover, scFvs can be isolated together with their coding sequences , evolved in vitro [24,26], engineered as variants, or fused genetically to other molecules [23,27,28]. However, the generation of recombinant antibodies from hybridoma cell lines is not easily accomplished due to high levels of aberrant mRNA molecules from rearranged, nonfunctional heavy- and light-chain genes [29,30]. ScFv antibodies directly cloned from monoclonal hybridoma cells often have lower binding affinities than their parent mAbs [20,31]. To date, five scFv antibodies against FB1 have been reported. Three scFvs derived from hybridoma cell lines showed 10- to 100-fold less reactivity as compared with their parent mAbs [20,32].One scFv antibody selected from the human synthetic antibody library had a KD of only 4.08
10–7 M . One recently reported scFv had a slightly lower affinity than its parent mAb, but displayed a poor cross-reactivity to FB2 , which may not be satisfactory for sample detection, as the guidelines for fumonisins in human foods and animal feeds require the efficient detection of total fumonisins (FB1 + FB2 + FB3) that invariably occur simultaneously in contaminated samples, with FB2 being the predominant compound in some instances [4,6]. Thus, the generation of scFv antibodies that are highly reactive to fumonisins and that can be applied for immunoassay detection remains a challenge. In this study, we combined hybridoma and phage display technologies to screen for fumonisin-specific recombinant antibodies with excellent properties. We firstly isolated fumonisin-specific hybridoma cell lines and then used them to construct a phage-displayed antibody library. Phage display coupled with highly stringent panning was used to isolate one high-affinity scFv antibody, designated H2. The objectives of this study were to characterize the features of the H2 scFv antibody and develop the H2 scFv-based assay for immunological detection of fumonisin contamination in agricultural products. 2. Experimental 2.1. Reagents and apparatus Fumonisins (FB1, FB2, and FB3), keyhole limpet hemocyanin (KLH), and bovine serum albumin (BSA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). TRizol reagent, Oligo(dT)12–18 primer, hypoxanthine-aminopterine-thymidine (HAT), and hypoxanthine-thymidine medium were purchased from Life Technologies (Grand
Island, NY, USA). The composition of HBS-EP running buffer was 10 mM HEPES (pH 7.4) with 150 mM NaCl, 3 mM EDTA, and 0.005% (v/v) surfactant P20. 96-well cell culture plates were purchased from Corning (Acton, MA). Cells were cultured in a CO2 incubator (Thermo-Fischer Scientific, OH). Affinity was determined by SPR measurements using a BIAcore T200 system (GE Healthcare, Uppsala, Sweden) and CM5 sensor chip was also purchased from GE Healthcare. UHPLC was performed using a Waters Acquity ultra-performance LC system (Waters, MA). Other reagents and apparatus were the same as those reported previously . 2.2. Antigen preparation and immunization FB1 was conjugated to KLH (FB1–KLH) for use as immunogen and conjugated to BSA (FB1–BSA) for use as a detection antigen with glutaraldehyde, as described previously . Five female Balb/c mice (6 week-old) were given four immunizations with 200 μL of 500 μg mL–1 FB1–KLH conjugate. In the first injection, antigen was emulsified with an equal volume of Freund’s complete adjuvant and injected subcutaneously. After 4 weeks, two enhanced immunizations were given with the same amounts of the conjugate emulsified with Freund’s incomplete adjuvant at 3-week intervals. The subcutaneous injection was carried out for the second and third times, while intravenous injection was used for the final immunization with 25 μg (200 μL) FB1–KLH conjugate. The blood was collected 10 days after the third immunization and sampled for antiserum titer assessment . The cell fusion for hybridoma production was done 3 days after the final immunization.
2.3. Generation of the mAbs against FB1 and construction of phage display library Hybridoma production, identification of the mAbs against FB1 and construction of phage display library from the mAbs were detailed in Supplementary information. 2.4. Panning by phage display A 1.5-mL aliquot of bacterial cells from the constructed monoclonal recombinant antibody library was pipetted for preparation of rescued phages, as described previously . Solid phase panning was applied to screen specific scFv antibodies, as described previously , with some modifications. Briefly, the microtiter plate wells were coated with 100 μL of FB1–BSA conjugate (20, 15, and 10 μg mL–1 in the first, second, and third rounds, respectively) and blocked with 2% (w/v) BSA in PBS. After three washings with PBST and PBS, 100 μL of prepared phages was added and incubated at 37 °C for 2 h. The wells were washed five times with PBST and PBS in the first round, and increased to 10 and 15 times in the second and third rounds of panning, respectively. Simultaneously, the concentration of Tween-20 in PBST was increased from 0.1% (v/v) to 0.5% (v/v) in the third round of selection. The bound phages were collected by adding 100 μL of 100 mM triethylamine (pH 12.0) at room temperature for 10 min without shaking. Then, 50 μL of 1 M Tris-HCl (pH 7.4) was added immediately for neutralization and 200 μL of the log-phase E. coli XL1-Blue MRF’ was added for phage infection at 37 °C for 30 min. The bacteria were cultured on TYE agar plates containing 1% (w/v) glucose and 100 μg mL−1 ampicillin at 37 °C overnight.
2.5. Identification of clones by phage ELISA and sequencing Phage ELISA was used to identify the individual clones after the third round of panning. Randomly selected clones were inoculated in 2 TY medium and infected with M13KO7 helper phages. The rescued phages were applied in phage ELISA detection against 5 μg mL–1 FB1–BSA, according to Liu et al. , and blocked with 2% (w/v) BSA in PBS. The positive clones were sequenced using the pHENpel and pHENmyc primers. Sequences were analyzed using BioEdit software (Ibis Biosciences, Carlsbad, CA, USA). 2.6. Soluble scFv-ELISA Soluble scFv antibodies were expressed and purified in bacteria, as described previously . The microtiter plate wells were coated with 100 μL of FB1–BSA (1 μg mL–1) and blocked with 150 μL of 2% (w/v) BSA at 37 °C for 2 h. Next, 200 nM purified scFv antibody was added and incubated for 1.5 h at 37 °C. After three washings with PBST and PBS, the wells were incubated with 100 μL of anti-His antibody (Tiagen, 1:5000 dilution in blocking solution). The absorbance values were detected with AP-labeled goat anti-mouse IgG antibody at 405 nm. Negative controls omitting FB1–BSA were set up, and all measurements were done independently, in triplicate. 2.7. SPR measurements of binding kinetics The binding capacity of H2 scFv antibody and its parent mAb was determined by SPR measurements using a BIAcore T200 system. The FB1–BSA, approximately 65
resonance units, was immobilized on the surface of a CM5 sensor chip. Different concentrations of each antibody, diluted in HBS-EP running buffer, were injected serially over the surface at a flow rate of 30 μL min–1 at 25 °C. The association and dissociation phases were monitored for 4 and 10 min, respectively. The chip was regenerated with 10 mM glycine-HCl (pH 2.0). The data were analyzed using BIAcore T200 evaluation software. 2.8. Dc-ELISA The mMicroplate wells were coated with 100 μL of 10 μg mL–1 goat anti-mouse IgG and blocked with 2% (w/v) BSA in PBS. Anti-His antibody was diluted with blocking solution at 1:3000 and added into wells for 1.5 h at 37 °C. After three washings with PBST and PBS, 200 nM purified scFv antibody was pipetted into each well and incubated for 1.5 h. 50 μL of fumonisin (2000, 200, 20, 2, 0.2 ng mL–1 FB1, FB2, or FB3), or 10 μM non-fumonisin mycotoxins (deoxynivalenol, nivalenol and aflatoxin B1), or PBS (as a blank control omitting mycotoxins) mixed with FB1–HRP conjugate (50 μL, prepared as previously described  and diluted 1:2000 in PBS containing 0.5% (w/v) BSA) was then added to wells and incubated for 1.5 h at 37 °C. After three washings with PBST and PBS, 100 μL of tetramethylbenzidine substrate was added and incubated for 25 min for color development. The reaction was stopped with 50 μL of H2SO4 (2 M), and the absorbance at 450 nm was measured. The standard curve was prepared using purchased FB1 and conjugated FB1–HRP as described above for sample analyses of different fumonisins in the Dc-ELISA detection.
2.9. Sample preparation and validation assay Healthy and naturally contaminated maize and rice samples were collected from local markets in Wuhan and tested by Dc-ELISA and ultra-high-performance liquid chromatography-tandem mass spectrometry (UHPLC–MS/MS) . For recovery analysis, 1 g of ground healthy maize suspended in 5 mL PBS containing 0.05% (v/v) Tween-20 was vigorously mixed with different amounts of FB1, FB2 or FB3 (200, 500, 1000 ng mL–1). After centrifugation at 10000 ×g for 10 min, a 50-μL aliquot of the matrix was mixed with equal volume of FB1–HRP conjugate for Dc-ELISA detection. For correlation coefficient analysis between the developed Dc-ELISA and UHPLC–MS/MS results, 16 naturally contaminated maize (7 samples) and rice (9 samples) from the market were used. One gram of ground sample was extracted with 4 mL extraction solvent (acetonitrile/water = 84:16, v/v) and cleaned-up by a silica gel cartridge and a MultiSep® 211 column (Romer Labs, Washington, MO, USA)  prior to UHPLC–MS/MS analysis . The extracts without cleaning-up were diluted 100 times with 0.05% (v/v) PBST and a 50-μL aliquot was used for the Dc-ELISA analysis . The experiments were performed in duplicate. 2.10. Modeling of the scFv antibody BLAST searches in the Protein Data Bank (NCBI) were performed to identify suitable templates for determining the three-dimensional structure of the scFv by homology modeling. The H2 scFv antibody structure was generated using MODELER (9v10) modeling software  based on structural alignment of four
three-dimensional structural models. The templates of 1DZB, 2GHW, 1F3R, and 3UMT with the (G4S)3 linker and more than 50% consensus sequence were used in the modeling (Fig. S1 and Table S2). The model with the lowest energy was chosen for energy minimization using assisted model building with force field ff99SB. Both the steepest descent steps and the conjugate gradient steps were set to 1000 during scFv antibody modeling. 2.11. Molecular docking The FB1, FB2, and FB3 molecules (PubChem accession nos.: 2733487, 2733489, and 3034751) were obtained from the PubChem Compound database. The SDF format of the ligands was converted to the Protein Data Bank format, and their structures were optimized using OpenBabel . To understand the molecular interactions between fumonisins and the H2 antibody, flexible small molecule-rigid protein docking experiments were performed using AutoDock 4.0 . Ten Lamarckian genetic algorithm runs were performed with default parameter settings, except that the maximum number of evaluations was set to 2,500,000. The docked confirmations were ranked by their energy, and the lowest energy complex of ligand and protein was viewed using PyMOL and subjected to further analyses.
3. Results and discussion 3.1. Generation and evaluation of the mAbs against FB1 Fumonisin B analogs have an amino group in the C2 position of their polyhydroxyl alkylamine chains (Fig. 1). To boost the immune response, KLH was
used as a carrier protein and conjugated to the amino group of FB1. After three immunizations, antisera from the immunized mice were sampled and measurement of their titers was performed using indirect ELISA. The results showed that all treated mice produced specific polyclonal antibodies against FB1, displaying a robust immune response up to 1:512000. The spleen cells from the mice that displayed the highest immune responses were fused with Sp2/0 myeloma cells, and supernatants prepared from the resulting cultures were analyzed by indirect ELISA. The positive hybridoma cells were subcloned for subsequent selections, generating four hybridoma cell lines secreting fumonisin-specific mAbs; these were designated as 1A9, 1C11, 2C5, and 2D12. Ascites fluid from each mAb was prepared for purification of the mAbs. The avidity of the purified mAbs was determined against FB1–BSA by indirect ELISA, revealing that 1C11 had the highest avidity towards FB1, followed by 2C5, 2D12, and 1A9 (Fig. S2). 3.2. Identification of scFv clones amplified directly from the mAbs Amplification of the VL domain from each mAb with two sets of light-chain-specific primers (Table S1) showed a single DNA fragment to be amplified only with the MVLF1–5 and MVLB1–3 primers, indicating that all of the mAbs contained a kappa light chain. For each mAb, the amplified VL domain with its VH domain were linked by a linker sequence (G4S)3 by splicing with overlap extension PCR and then immobilized into bacteria. Ten clones for each mAb were assayed by soluble scFv-ELISA; unfortunately, rather low or no reactivity against FB1–BSA was detected. Further sequence analyses of five clones from each mAb
demonstrated the presence of aberrant rearrangement, frame shift, or stop mutations in the sequences (Fig. S3), supporting the view that it is difficult to isolate functional scFv antibodies directly amplified from hybridoma cell lines [29,30]. 3.3. Construction and panning of a monoclonal recombinant antibody library To select scFv antibodies with high affinity and to circumvent the problem of aberrant mRNA from rearranged coding genes of hybridoma cells, an antibody library containing 1.3 106 clones was generated with the recombinant phagemids derived from four mAbs; this library was used for panning by phage display. Equal amounts of the recombinant phagemids from four mAbs were mixed before transformation in order to assure the same number of initial phagemids under the same conditions was introduced into the bacterial cells. Thus, equivalent phage numbers from each cell line were used for panning of binders by phage display. PCR analyses of 20 randomly selected clones from the library showed that each carried the expected DNA fragment (data not shown). After phage rescue and recovery, approximately 1013 phage particles were used for screening against FB1–BSA. Three rounds of panning were performed. The amounts of input and output phages are shown in Table S3. In the subsequent procedures, the washing times with PBST and PBS were increased while the concentration of coated antigens was reduced in each round of panning. The ratio of output and input phages of the second round (4.05 10–7) increased nearly 10 times relative to the first round (4.50 10–8). However, the phage recovery was reduced in the third round of selection (8.99 10–9). We hypothesize that this atypical reduction in the third round may have resulted from the highly stringent conditions used for
panning, particularly the 5-fold increment of Tween-20 in the PBST . The panned library was used as a pool for subsequently identifying scFv antibodies with high affinity. 3.4. Phage ELISA and sequencing analysis of phage display-derived clones Ninety-five clones resulting from the final panned library were randomly selected and their affinities toward FB1–BSA were identified by phage ELISA. As shown in Figure 2, 81 clones gave a positive reaction with varied signal intensity; 20 of these clones were sequenced based on the strength of signal intensities (high, intermediate and low intensities). Alignments of these sequences with the VH and VL domains directly amplified from the four mAbs clearly revealed two classes of scFv antibodies (Fig. S4). 17 clones (labeled with green stars in Fig. 2) were derived from the 1C11 mAb, and the remaining three antibodies (labeled with red circles) were derived from 2C5. No phage clone was obtained from the 1A9 or 2D12 mAbs. We noted that all of the scFv antibodies derived from the 1C11 mAb had higher binding affinities than those derived from the 2C5 mAb. Further sequence analyses indicated that scFv antibody sequences derived from the same mAb often had some variation in their nucleotides. Interestingly, among the 17 clones derived from the 1C11 mAb, three clones (E8, F9, and H2) had identical nucleotide sequences. Finally, six clones (B1, F8, F10, G3, G9, and H2) from 1C11 and the three clones (A11, G8, and H7) from 2C5 (Fig. 2) were selected for soluble scFv-ELISA with antibodies obtained through a bacterial expression system. The two groups of scFv antibodies derived from the 1C11 and 2C5 displayed 5.7% and 4% variation, respectively, in their amino
acid sequences (Fig. S5). These results demonstrated that the stringent conditions employed in this study could efficiently enrich high-affinity antibodies and effectively discard low-affinity ones antibodies a pool, as has been observed by others [40,41]. Our results thus further demonstrate that phage display is a powerful method to select high affinity recombinant antibodies. 3.5. Soluble scFv-ELISA of the selected scFvs Soluble expression of the nine selected scFvs was induced in E. coli strain XL1-Blue with IPTG, and the periplasmic proteins were extracted by osmotic shock [24,25]. Soluble scFv-ELISAs illustrated that there were significant differences in the scFvs assayed (Fig. 3). The six phage clones derived from 1C11 displayed a higher reactivity than the three clones from 2C5. However, variations were also apparent within the clones from the same hybridoma cell line, 1C11 or 2C5, a result that has been noted by other studies . These divertent reactivities result from the fact that variation of a few amino acids in the scFvs can profoundly affect the binding capacity of an antibody [24,26]. Among the scFvs from 1C11, the H2 antibody had the highest reactivity. The discrepancy between phage ELISA (Fig. 2) and soluble scFv-ELISA (Fig. 3) for the relative OD values for some scFvs is a phenomenon that may be related to the improper folding of the scFv-gIII fusion in phage particles; this has been observed previously . A high reactivity of the soluble form of a scFv antibody is known to be a vital factor for its potential use in downstream biotechnological applications. Therefore, the H2 antibody was chosen for further analysis. The H2 coding sequence (GenBank Accession Number KC952002) and its deduced amino
acid sequence are presented in Figure S6. 3.6. Kinetics analyses of the H2 scFv and its parent mAb To ascertain whether the isolated H2 scFv antibody retains satisfactory binding properties as compared with its parent mAb, the H2 scFv and the 1C11 mAb were used for comparative SPR measurements using a BIAcore T200 system with a CM5 chip immobilized with FB1–BSA. The result showed that the association rate constants (ka) for the 1C11 mAb (9.145 105 M–1 s–1) and the H2 scFv antibody (7.178 105 M–1 s–1) were comparable. More importantly, the H2 antibody had a markedly lower dissociation rate constant (kd = 8.629 10–4 s–1) than did the parent 1C11 mAb (kd = 9.000 10–2 s–1), a 104-fold difference. Thus, the equilibrium dissociation constants (KD) were 1.20 10–9 M for the H2 scFv antibody and 9.84 10–8 M for the 1C11 mAb, respectively, demonstrating an 82-fold increment of the binding kinetics for the former. These results indicated that the scFv antibody H2 had a dramatically improved affinity toward FB1 as compared to the 1C11 mAb. The comparative binding results shown in Figure 4 illustrate that the H2 scFv and its parent 1C11 mAb injected at various concentrations (4, 16, 64, 256, and 1024 nM) could bind to FB1–BSA; the H2 scFv gave a stably stronger signal than did the parent mAb 1C11 throughout the dissociation time course experiments. The high affinity binding kinetics observed for the H2 scFv antibody were 340-fold higher than those reported previously for scFvs (KD = 4.08 10–7 M) against FB1 . The strategy of combining mAbs with phage display was impressively effective: the H2 scFv antibody had an 82-fold higher binding affinity than the original mAb from which it
was derived. The selected H2 scFv can therefore be considered to be a promising recombinant antibody molecule for basic and applied research. Further analysis of competition and cross-reactivity to other fumonisin analogs will provide a deeper understanding of its potential applications. 3.7. Cross-reactivity analyses To confirm that the H2 scFv antibody recognizes free fumonisin FB1 and its analogs with the same backbone structures, Dc-ELISAs were conducted with FB1, FB2, and FB3 as competitors. The H2 scFv antibody was not only able to bind to free fumonisin FB1, but also to FB2 and FB3 (Fig. 5), in the presence of their competitor, the FB1–HRP conjugate. The amounts of FB1, FB2, and FB3 required to inhibit the binding of the antibody to the FB1–HRP conjugate by 50% were found to be 0.11, 0.04, and 0.10 μM, respectively. These results indicate that the antibody bound preferentially to FB2. These low IC50 values for all three fumonisin mycotoxins indicated that the H2 scFv antibody can efficiently detect FB1, FB2, and FB3 at low concentrations, suggesting that the H2 antibody is highly promising for use in immunological detection. The H2 antibody had no cross-reactivity with non-fumonisin mycotoxins such as deoxynivalenol, nivalenol and aflatoxin B1 that are frequently present together with fumonisins in agriculture samples. No cross-reactivity was observed even when the concentrations of non-fumonisin mycotoxins were as high as 5 μM (Table S4). Therefore, the H2 scFv antibody can efficiently detect FB1, FB2, and FB3, the most predominant fumonisins, with high specificity.
3.8. Validation Assay To prove that the H2 scFv antibody can be used for the detection of fumonisin contamination in agricultural samples, validation assays were initially carried out by investigating the recovery and coefficient of variation (CV) of FB1, FB2, and FB3 in maize samples. As shown in Table 1, the average recovery efficiencies of intra assay ranged from 92.8% to 104.1% for FB1, from 85.8% to 97.1% for FB2, and from 82.0% to 88.0% for FB3; the inter assay recoveries were 85.7% to 102.2 % for FB1, from 88.5% to 98.3% for FB2, and from 85.7% to 92.9% for FB3 respectively, with CV values of less than 15%. These results indicated that H2 scFv-based immunoassays could efficiently and consistently detect three fumonisins present in food samples. To further validate the utility of the H2 scFv-based immunoassays with fumonisin-contaminated cereals, a total of 16 naturally contaminated samples (7 maize and 9 rice) were tested. Both H2 scFv-based Dc-ELISA detection and chemical analysis with UHPLC–MS/MS were conducted. As shown in Figure 6, the results measured by the Dc-ELISA were in agreement with those of the coupled mass spectrometry-based method that measured the amounts of FB1 and FB2 in the samples; the coefficient of correlation R2 for the two methods was 0.89. There is a deviation between the ELISA and UHPLC–MS/MS analyses; the Dc-ELISA revealed higher values than the chemical analysis, possibly due to matrix effects, losses during a cleaning-up process for chemical analysis , the greater sensitivity of ELISA than chemical analysis , as well as the fact that UHPLC–MS/MS analyses only
determined FB1 and FB2 whereas the Dc-ELISA detected all fumonisin analogs from the naturally contaminated samples. Similar phenomena have been observed in other immunological assays of fumonisns by polyclonal, monoclonal and scFv antibodies compared with chemical analyses [16,18,34]. Nevertheless, the current results indicated that the H2-based assay can effectively detect fumonisin-contaminated samples in nature. Phosphate buffer extracts may be a good option for sample preparation for immunoassays owing to its simplicity and wide applications. 3.9. Molecular docking and putative rational for the binding of the H2 scFv antibody to fumonisins To understand the molecular mechanism of the highly specific binding interactions between fumonisins and the H2 scFv antibody, a three-dimensional structure of the H2 scFv antibody was built by homology modeling. This modeled structure exhibited the characteristic scFv regions of immunoglobulin folding (Fig. 7) that consists exclusively of antiparallel β-sheets connected by loops . This clearly indicated that a favorable binding surface was formed by the six complementarity-determining regions (CDRs) in the H2 antibody. Further molecular docking analysis of the H2 scFv antibody with the FB1, FB2, and FB3 ligands was separately carried out. These studies revealed that the CDRs of both the heavy and light chains directly face the all of the fumonisin molecules in the antibody-fumonisin complex (Figs. 7 and S7). Electrostatic matching between the H2 scFv antibody and FB1, FB2, and FB3 was analyzed individually (Fig. 8A). The abundant negative charges carried by the carboxyl radicals in the fumonisins in neutral solution were well matched with the
positive charge regions of the H2 antibody surface. Likewise, the amidogen protonation and hydroxyl radical in the C3 position of the fumonisins make each molecule carry positive charges that can interact coulombically with the negative charge regions of the H2 antibody. However, the hydroxyl radical in the R1 position of FB1 and FB3 is close to the carboxyl that carries strong negative charges that could occupy the hydrogen atom of the hydroxyl. Under such a scenario, the negative charge in the R1 position of FB1 and FB3 does not match with the negative charge domains of the H2 scFv antibody, resulting in an adverse effect on the interaction. In contrast, a hydrogen radical is present in the R1 position of FB2, so no such adverse impact exists. Further analyses of hydrophobic interactions showed that the H2 scFv antibody is in good complementary with the three fumonisin molecules in their respective hydrophobic and hydrophilic regions (Fig. 8B). The hydrogen radical in the R1 position of FB2 matches well with the hydrophobic domains of the antibody, forming a favorable interaction, whereas the hydrophilic hydroxyl radical in the R1 position of FB1 and FB3 does not match with the hydrophobic domains of the H2 antibody. Therefore, the H2 scFv antibody has a higher affinity to FB2 than to FB1 or to FB3. It has a comparable binding affinity to both FB1 and FB3. These results are in good agreement with the results observed in our Dc-ELISA experiments (Fig. 5). Analyses of the hydrogen bond interactions between the H2 antibody and the fumonisins FB1 and FB2 showed that nine hydrogen bonds are formed (one in Ala-35 of CDRH1, one in Lys-55 of CDRH2, two in Tyr-62 of CDRH2, three in Asp-64 of CDRH2, one in Thr-105 of CDRH3, and one in Tyr-241 of CDRL3) (Figs. 7 and
S7A). With FB3, the hydroxyl in the R2 position is substituted by a hydrogen radical. Thus, only one hydrogen bond is formed between Tyr-62 of CDRH2 in the H2 antibody and the R2 position of FB3 and there are only eight hydrogen bonds formed with the antibody (Fig. S7B). We assumed that these hydrogen bonds located in the CDRs, particularly those in the CDRH2 region, contribute greatly to the interaction of the H2 antibody with the fumonisins; such interactions have been shown to be critical for the binding of other antibody-ligand complexes [37,42]. The different numbers of hydrogen bonds formed between the H2 antibody and FB1 (FB2) (nine) and FB3 (eight) may not greatly contribute to the binding discrepancy observed in the IC50 assay for FB1, FB2, and FB3. The electrostatic, hydrophobic, and hydrogen bonding properties of the H2 scFv antibody together form a favorable binding pocket that appears to be highly complementary to the fumonisins and to confer excellent binding properties. Further, the hydrogen radical present in the R1 position of FB2 encourages proper electrostatic matching and hydrophobic interactions with the H2 antibody, giving rise to the preferential interaction with FB2 observed in the competition assays. 4. Conclusions Generation of sensitive reagents for the accurate and specific detection of potentially carcinogenic fumonisins is vitally important to prevent the consumption of contaminated agricultural products. Immunoassays are a simple, rapid, and cost-effective technology that can be used to monitor fumonisin contaminations. Single-chain fragment variable antibodies possess several advantages over polyclonal and monoclonal antibodies. However, the few known scFv antibodies against
fumonisins displayed either low affinity or limited reactivity to fumonisin analogs; there is currently no scFv antibody available for the satisfactory detection of fumonisins. We constructed a phage display antibody library with mRNA from four mAbs generated against the FB1 mycotoxin and then isolated and cloned the gene for an affinity improved fumonisin-specific scFv antibody by phage display. The isolated H2 scFv antibody has an 82-fold higher affinity for FB1 than does its parental mAb, and it binds with a 340-fold higher affinity than any of the other scFvs reported thus far. The H2 scFv antibody was used to develop a sensitive direct competitive immunoassay for the detection of three predominant fumonisins, FB1, FB2, and FB3. This scFv does not cross-react to other mycotoxins that are often concurrently present in agriculture samples along with fumonisins. Molecular modeling and docking analyses revealed that the H2 antibody carries a favorable binding structure that is highly complementary to the three fumonisins. The presence of a hydrogen radical in the R1 position of FB2 may be responsible for its proper electrostatic matching and hydrophobic interaction with the antibody. Further recovery analyses and comparison with results from UHPLC–MS/MS analysis validated that our scFv-based Dc-ELISA is a sensitive, specific, accurate, and cost-effective detection assay for FB1, FB2 and FB3 in naturally contaminated agricultural products. This scFv antibody can be used directly for immunoassays of fumonisin contamination in food/feed commodities. Acknowledgments This research was funded by grants from the National Basic Research Program of China (Grant 2013CB127801), the National Natural Science Foundation of China
(Grant 31272004), the Ministry of Science and Technology of China (BELSPO, grant S2012GR0016), and the Bureau of Science and Technology of Wuhan (Grant 2013020501010170). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version. References  W.C. Gelderblom, K. Jaskiewicz, W.F. Marasas, P.G. Thiel, R.M. Horak, R. Vleggaar, N.P. Kriek, Fumonisins–novel mycotoxins with cancer-promoting activity produced by Fusarium moniliforme, Appl. Environ. Microbiol. 54 (1988) 1806–1811.  J. Fotso, J.F. Leslie, J.S. Smith, Production of beauvericin, moniliformin, fusaproliferin, and fumonisins B1, B2, and B3 by fifteen ex-type strains of Fusarium species, Appl. Environ. Microbiol. 68 (2002) 5195–5197.  J.P. Rheeder, W.F. Marasas, H.F. Vismer, Production of fumonisin analogs by Fusarium species, Appl. Environ. Microbiol. 68 (2002) 2101–2105.  J.H. Wang, J.B. Zhang, H.P. Li, A.D. Gong, S. Xue, R.S. Agboola, Y.C. Liao, Pathogenicity of Fusarium temperatum isolated from maize in China, J. Phytopath. 162 (2014)147–157.  P.M. Scott, Recent research on fumonisins: a review, Food Addit. Contam. Part A Chem. Anal. Control. Expo. Risk Assess. 29 (2012) 242–248.
 S. Yazar, G.Z. Omurtag, Fumonisins, trichothecenes and zearalenone in cereals, Int. J. Mol. Sci. 9 (2008) 2062–2090.  J.M. Soriano, L. Gonzalez, A.I. Catala, Mechanism of action of sphingolipids and their metabolites in the toxicity of fumonisin B1, Prog. Lipid Res. 44 (2005) 345–356.  H. Stockmann-Juvala, K. Savolainen, A review of the toxic effects and mechanisms of action of fumonisin B1, Hum. Exp. Toxicol. 27 (2008) 799–809.  R.B. Myburg, M.F. Dutton, A.A. Chuturgoon, Cytotoxicity of fumonisin B1, diethylnitrosamine, and catechol on the SNO esophageal cancer cell line, Environ. Health Perspect. 110 (2002) 813–815.  J.P. Rheeder, W.F. Marasas, P.G. Thiel, E.W. Sydenham, G.S. Shephard, D.J. van Schalkwyk, Fusarium moniliforme and fumonisins in corn in relation to human esophageal cancer in Transkei, Phytopathology 82 (1992) 353–357.  G. Sun, S. Wang, X. Hu, J. Su, T. Huang, J. Yu, L. Tang, W. Gao, J.S. Wang, Fumonisin B1 contamination of home-grown corn in high-risk areas for esophageal and liver cancer in China, Food Addit. Contam. 24 (2007) 181–185.  Y. Ueno, K. Iijima, S.D. Wang, Y. Sugiura, M. Sekijima, T. Tanaka, C. Chen, S.Z. Yu, Fumonisins as a possible contributory risk factor for primary liver cancer: A 3-year study of corn harvested in Haimen, China, by HPLC and ELISA, Food Chem. Toxicol. 35 (1997) 1143–1150.
 M. Castegnaro, C.P. Wild, IARC activities in mycotoxin research, Nat. Toxins 3 (1995), 327–331, discussion 341.  N.W. Turner, S. Subrahmanyam, S.A. Piletsky, Analytical methods for determination of mycotoxins: a review, Anal. Chim. Acta 632 (2009) 168–180.  J.I. Azcona-Olivera, M.M. Abouzied, R.D. Plattner, W.P. Norred, J.J. Pestka, Generation of antibodies reactive with fumonisins B1, B2, and B3 by using cholera toxin as the carrier-adjuvant, Appl. Environ. Microbiol. 58 (1992) 169–173.  Y. Quan, Y. Zhang, S. Wang, N. Lee, I.R. Kennedy, A rapid and sensitive chemiluminescence enzyme-linked immunosorbent assay for the determination of fumonisin B1 in food samples, Anal. Chim. Acta 580 (2006) 1–8.  J.I. Azcona-Olivera, M.M. Abouzied, R.D. Plattner, J.J. Pestka, Production of monoclonal antibodies to the mycotoxins fumonisins B1, B2 and B3, J. Agric. Food Chem. 40 (1992) 531–534.  I. Barna-Vetró, E. Szabó, B. Fazekas, L. Solti, Development of a sensitive ELISA for the determination of fumonisin B1 in cereals, J. Agric. Food Chem. 48 (2000) 2821–2825.  S. Ling, J. Pang, J. Yu, R. Wang, L. Liu, Y. Ma, Y. Zhang, N. Jin, S. Wang, Preparation and identification of monoclonal antibody against fumonisin B1 and development of detection by Ic-ELISA, Toxicon 80 (2014) 64–72.
 W.K. Min, Y.J. Cho, J.B. Park, Y.H. Bae, E.J. Kim, K. Park, Y.C. Park, J.H. Seo, Production and characterization of monoclonal antibody and its recombinant single chain variable fragment specific for a food-born mycotoxin, fumonisin B1, Bioprocess Biosyst. Eng. 33 (2010) 109–115.  F.Y. Yu, F.S. Chu, Production and characterization of monoclonal antibodies against fumonisin B1, Food Agric. Immunol. 11 (1999) 297–306.  H.R. Hoogenboom, A.D. Griffiths, K.S. Johnson, D.J. Chiswell, P. Hudson, G. Winter, Multi-subunit proteins on the surface of filamentous phage: methodologies for displaying antibody (Fab) heavy and light chains, Nucleic Acids Res. 19 (1991) 4133–4137.  Z.Q. Hu, H.P. Li, J.B. Zhang, T. Huang, J.L. Liu, S. Xue, A.B. Wu, Y.C. Liao, A phage-displayed chicken single-chain antibody fused to alkaline phosphatase detects Fusarium pathogens and their presence in cereal grains, Anal. Chim. Acta 764 (2013) 84–92.  J.L. Liu, Z.Q. Hu, S. Xing, S. Xue, H.P. Li, J.B. Zhang, Y.C. Liao, Attainment of 15-fold higher affinity of a Fusarium-specific single-chain antibody by directed molecular evolution coupled to phage display, Mol. Biotechnol. 52 (2012) 111–122.  N.G. Nossal, L.A. Heppel, The release of enzymes by osmotic shock from Escherichia coli in exponential phase, J. Biol. Chem. 241 (1966) 3055–3062.  G. Rojas, A. Pupo, S. Gómez, U. Krengel, E. Moreno, Engineering the binding
site of an antibody against N-glycolyl GM3: from functional mapping to novel anti-ganglioside specificities, ACS Chem Biol. 8 (2013) 376–386.  D. Peschen, H.P. Li, R. Fischer, F. Kreuzaler, Y.C. Liao, Fusion proteins comprising a Fusarium-specific antibody linked to antifungal peptides protect plants against a fungal pathogen, Nat. Biotechnol. 22 (2004) 732–738.  S. Xue, H.P. Li, J.B. Zhang, J.L. Liu, Z.Q. Hu, A.D. Gong, T. Huang, Y.C. Liao, Chicken single-chain antibody fused to alkaline phosphatase detects Aspergillus pathogens and their presence in natural samples by direct sandwich enzyme-linked immunosorbent assay, Anal. Chem. 19 (2013) 10992–10999.  K.W. Lee, B.U. Hur, S.Y. Song, H.J. Choi, S.H. Shin, S.H. Cha, Methods for rapid identification of a functional single-chain variable fragment using alkaline phosphatase fusion, BMB Rep. 42 (2009) 731–736.  P.J. Nicholls, V.G. Johnson, M.D. Blanford, S.M. Andrew, An improved method for generating single-chain antibodies from hybridomas, J. Immunol. Methods 165 (1993) 81–91.  G.H. Choi, D.H. Lee, W.K. Min, Y.J. Cho, D.H. Kweon, D.H. Son, K. Park, J.H. Seo, Cloning, expression, and characterization of single-chain variable fragment antibody against mycotoxin deoxynivalenol in recombinant Escherichia coli, Protein Expr. Purif. 35 (2004) 84–92.  H.R. Zhou, J.J. Pestka, L.P. Hart, Molecular cloning and expression of recombinant phage antibody against fumonisin B1, J. Food Prot. 59 (1996)
1208–1212.  B. Lauer, I. Ottleben, H.J. Jacobsen, T. Reinard, Production of a single-chain variable fragment antibody against fumonisin B1, J. Agric. Food Chem. 53 (2005) 899–904.  L. Zou, Y. Xu, Y. Li, Q. He, B. Chen, D. Wang, Development of a single-chain variable fragment antibody-based enzyme-linked immunosorbent assay for determination of fumonisin B1 in corn samples, J. Sci. Food Agric. 94 (2014) 1865–1871.  Z.Q. Hu, J.L. Liu, H.P. Li, S. Xing, S. Xue, J.B. Zhang, J.H. Wang, G. Nölke, Y.C. Liao, Generation of a highly reactive chicken-derived single-chain variable fragment against Fusarium verticillioides by phage display, Int. J. Mol. Sci. 13 (2012) 7038−7056.  Z. Han, Y.P. Ren, J.F. Zhu, Z.X. Cai, Y. Chen, L.J. Luan, Y.J. Wu, Multianalysis of 35 mycotoxins in traditional Chinese medicines by ultra-high-performance liquid chromatography-tandem mass spectrometry coupled with accelerated solvent extraction, J. Agric. Food Chem. 60 (2012) 8233–8247.  X. Li, P. Li, Q. Zhang, Y. Li, W. Zhang, X. Ding, Molecular characterization of monoclonal antibodies against aflatoxins: a possible explanation for the highest sensitivity, Anal. Chem. 84 (2012) 5229–5235.  N.M. O'Boyle, M. Banck, C.A. James, C. Morley, T. Vandermeersch, G.R. Hutchison, Open Babel: an open chemical toolbox, J. Cheminform. 3 (2011) 33.
 R.J. Rosenfeld, D.S. Goodsell, R.A. Musah, G.M. Morris, D.B. Goodin, A.J. Olson, Automated docking of ligands to an artificial active site: augmenting crystallographic analysis with computer modeling, J. Comput. Aided Mol. Des. 17 (2003) 525–536.  H.J. Kim, M.R. McCoy, Z. Majkova, J.E. Dechant, S.J. Gee, S. Tabares-da Rosa, G.G. González-Sapienza, B.D. Hammock, Isolation of alpaca anti-hapten heavy chain single domain antibodies for development of sensitive immunoassay, Anal. Chem. 84 (2012) 1165–1171.  C.R. Bever, Z. Majkova, R. Radhakrishnan, I. Suni, M. McCoy, Y. Wang, J. Dechant, S. Gee, B.D. Hammock, Development and utilization of camelid VHH antibodies from alpaca for 2,2',4,4'-tetrabrominated diphenyl ether detection, Anal. Chem. 86 (2014) 7875–7882.  M.J. Rosok, M. Eghtedarzadeh-Kondri, K. Young, J. Bajorath, S. Glaser, D. Yelton, Analysis of BR96 binding sites for antigen and anti-idiotype by codon-based scanning mutagenesis, J. Immunol. 160 (1998) 2353–2359.
Legends of Figures and Tables
Fig. 1. Chemical structures of FB1, FB2, and FB3. Fig. 2. Phage ELISA of randomly selected individual clones against FB1–BSA.
Ninety-five clones were selected from the third round of panning; eighty-one clones had a positive reaction toward FB1–BSA with various signal intensities. Seventeen clones labeled with green stars were all derived from the 1C11 mAb, whereas the remaining three antibodies labeled with red circles were derived from 2C5. Fig. 3. Soluble scFv-ELISA analysis of selected phage-displayed recombinant antibodies. B1, F8, F10, G3, G9, and H2 scFv antibodies were derived from the 1C11 mAb, whereas A11, G8, and H7 scFvs were derived from the 2C5 mAb. These antibodies were analyzed in the presence or absence (control) of the FB1–BSA antigens. The y-axis represents the OD405nm values minus the control values. Values represent the means ±SD of triplicate assays. The different letters A, B, C, D, and E above the columns represent a significant difference at P < 0.01. Fig. 4. Binding analysis of the H2 scFv antibody and the parent 1C11 mAb using surface plasmon resonance on a BIAcore T200 system. The purified H2 scFv (A) and 1C11 mAb (B) were injected at the indicated concentrations over a sensor chip coated with FB1–BSA conjugate. RU, resonance unit. Fig. 5. SDS-PAGE of the purified H2 antibody and inhibition analysis of the H2 scFv antibody on free fumonisins FB1, FB2, and FB3 by Dc-ELISA. The values represent means ± SD of triplicate assays. M, protein molecular weight standards. Fig. 6. Correlation analysis of the results measured using the developed Dc-ELISA and UHPLC–MS/MS analyses. Each sample was measured in duplicate by the scFv-ELISA and both replicates are shown in the figure. Fig. 7. Molecular docking model of the H2 scFv antibody with fumonisins generated
using AutoDock 4.0. The heavy and light chains, the linker, and fumonisin B1 are shown in yellow, blue, red, and green, respectively. Amino acid residues involved in the interactions with the fumonisins are labeled, and the hydrogen bonds are shown as yellow broken lines adjacent to the respective amino acids and numbers. Fig. 8. Analyses of electrostatic matching (B) and hydrophobic interactions (C) between the H2 scFv antibody and the three fumonisins. Negative charges and hydrophobic domains are displayed in red, whereas positive charges and hydrophilic domains are shown in blue. The FB1, FB2, and FB3 molecules are shown in green, pink, and yellow, respectively.
Table 1. Accuracy of Dc-ELISA evaluated by FB1, FB2 and FB3 recovery from maize samples. Spiked concentration (ng mL–1)