The E. coli strain BL21 (DE3) was purchased from Sangon Biotech (China). The ABA-scFv gene sequence, referenced from Lu (2013), was synthesized by Sangon Biotech and cloned into the pET30a vector to create the expression plasmid pET30a-ABA-scFv. To establish the chaperone co-expression systems, BL21 (DE3) competent cells were first transformed with one of five molecular chaperone plasmids: pG-KJE8, pGro7, pKJE7, pG-Tf2, or pTf16 (Takara, Japan), as detailed in Table 1. After selection and cultivation of these host strains, the pET30a-ABA-scFv plasmid was transformed into each, resulting in five distinct co-expression strains.

The concentration of soluble scFv was determined by an indirect ELISA. This assay was specifically designed to quantify the yield of scFv variants presenting an accessible C-terminal His-tag. This approach ensures that the quantified protein is not only expressed but also correctly folded in a manner that allows for its detection and subsequent use in downstream immunoassays that rely on the His-tag for capture or detection. The use of highly specific monoclonal anti-His antibodies is a standard and reliable method for such quantification (Waugh, 2005; Gaberc-Porekar and Menart, 2001).

Firstly, 10 μL (OD600 diluted to 0.6) of each E. coli expression strain was inoculated into 10 mL of LB liquid medium containing 1 mM IPTG, 60 μg/L kanamycin (Kana), corresponding antibiotics for the molecular chaperone expression plasmids, and inducing agents (as shown in Table 1). The cultures were grown at 28 °C and 150 rpm until reaching the stationary phase, with four biological replicates for each treatment.

Upon reaching the stationary phase, cells were harvested by centrifugation. Biomass was measured directly as the wet weight of the bacterial pellet. To account for variations in cell density between cultures, the relative cell number was determined by quantifying the genomic copy number using absolute qPCR; this served as a normalization factor. This per-cell normalization is crucial for distinguishing the direct effects of chaperones on protein folding from their indirect effects on overall cell growth and biomass. The primers for the chromosomal reference were E. coli’s dxs monocopy gene (F: CGA​GAA​ACT​GGC​GAT​CCT​TA, R: CTT​CAT​CAA​GCG​GTT​TCA​CA) (Hernandez-Beltran et al., 2024). The qPCR cycling conditions were: 95 °C × 2.5 min; 50 cycles of (95 °C × 15 s, 55 °C × 30 s, 72 °C × 30 s) using a Real-time fluorescence quantitative PCR instrument (K9600, QuantReady, China), and the standard curve preparation followed the method by Shapiro et al. (2018).

The cells were centrifuged, and the wet weight of the bacterial pellet was measured. The bacterial protein extraction kit (CWBio, China) was used for cell lysis. The supernatant, containing the soluble expressed protein, was collected for analysis, while the pellet containing inclusion bodies was dissolved using the inclusion body solubilization buffer (Sangon Biotech, China) for separate quantification. The purified soluble protein from the supernatant was first verified for purity and correct molecular weight by SDS-PAGE. For further identity confirmation, a Western blot was performed. Proteins from the gel were transferred to a PVDF membrane, which was then blocked with 5% non-fat milk in TBST. The membrane was incubated overnight at 4 °C with a primary anti-His tag mouse monoclonal antibody (1:5,000 dilution, Sangon Biotech), followed by incubation for 1 h at room temperature with an HRP-conjugated goat anti-mouse IgG secondary antibody (1:10,000 dilution, Sangon Biotech). The signal was visualized using an ECL chemiluminescence kit. The protein content was measured using an enzyme-linked immunosorbent assay (ELISA) kit (Shanghai Yuanju Bio, China), and the absorbance was read at 450 nm using a microplate reader (Infinite E Plex, Tecan, Austria). All relevant operations followed the instructions provided with the respective kits. The final volumetric production of ABA-scFv was normalized against the initial culture volume and expressed in pmol per mL of the initial culture.

The ABA conjugated antigen ABA-OVA was prepared according to Yan et al. (2022) using the dicyclohexylcarbodiimide (DCC) method. To prepare the ABA-OVA conjugate, 0.1 mmol of ABA was dissolved in 1.6 mL of N, N-dimethylformamide (DMF). Subsequently, 0.2 mL of 1.5 mol/L N-hydroxysuccinimide (NHS) in DMF was added, and the mixture was stirred for 15 min. Next, 2 mL of 0.1 mmol/L dicyclohexylcarbodiimide (DCC) in DMF was added to the solution, and the reaction was stirred overnight at room temperature. The mixture was centrifuged at 4,000 rpm for 10 min, and the resulting supernatant was collected and dialyzed against 0.1 M PBS for 3 days. Finally, the supernatant was freeze-dried. The resulting powder was re-dissolved in deionized water to a final concentration of 500 mg/L. Conjugation ratios were determined via intact isoform characterization of protein–hapten conjugates by LC-ESI-qTof mass spectrometry (Adamczyk et al., 1996) combined with spectral deconvolution using UniDec (Marty et al., 2015). OVA–ABA conjugates were desalted using 30 kDa ultrafiltration units (Briscale UF, Cobetter, China). Reversed-phase desalting was performed on a C4 column (ACQUITY UPLC Protein BEH C4, Waters, United States) using a binary gradient system, with solvent A consisting of 0.1% formic acid (FA) in water and solvent B consisting of 0.1% FA in acetonitrile. Proteins were eluted with a linear gradient from 5% to 95% B over 20 min at a flow rate of 0.3 mL/min. Intact protein masses were acquired on an ESI-QTOF instrument (6546 Q-TOF, Agilent Technologies, United States) under denaturing conditions, combined with spectral deconvolution using UniDec. Sample purity was validated by SDS-PAGE (Laemmli, 1970), which revealed a single dominant band at ∼45 kDa.The checkerboard ELISA procedure was conducted according to the method of Wang et al. (2021). For optimization, antigen coating dilutions of 1:50, 1:100, 1:200, 1:400, 1:800, 1:1,600, 1:3,200, and 1:6,400 (50 μL per well) were tested against antibody dilutions of 1:50–1:1,600 prepared from a 5 μg/mL stock using 0.1% PBS buffer containing 1% Tween-20 and 1 g/L gelatin (pH 7.5). In competitive binding wells, 50 μL of ABA at 100 ng/mL was added, and OD values of competitive binding and control wells were compared to identify the optimal antigen–antibody combination for each expression strain. Secondary antibody: HRP-conjugated Anti-HIS mouse monoclonal antibody, Sangon Biotech, cat#410002, dilution 1:2000. Subsequently, competitive ELISAs were performed under the optimized conditions. Serial ABA concentrations (0–1,000 ng/mL) were added to generate competitive binding curves, from which IC₅₀ values were calculated by four-parameter logistic regression. All experiments were conducted in triplicate.

For this analysis, the optimal coating antibody and antigen concentrations were used, as determined by the checkerboard titration described previously. The calculation method for IC₅₀ refers to Han et al. (2024). Six concentration gradients of 1,000 ng, 500 ng, 250 ng, 125 ng, 62.5 ng, and 31.25 ng were used to establish standard curves for each treatment. The OD value at IC₅₀ was calculated using Equation 1, and the ABA concentration at half-maximal inhibitory concentration (IC₅₀) was determined from the standard curve to evaluate its sensitivity. OD at  IC 50 = OD value of control well × 50 % (1)

The IC₅₀ values for common plant interfering factors (IAA, GA3, L-lysine, and benzoic acid) were determined, using the optimal coating combination. Their cross-reactivity rates were then calculated with Equation 2 to evaluate the specificity of ABA recognition. Cross − reactivity rate  % = IC 50   ABA   /   IC 50   interfering factor × 100 (2)

The purified ABA-scFv samples were mixed with KBr (w/w = 1/140), then vacuum freeze-dried and pressed into thin pellets. The infrared spectrum was recorded between 4,000 and 450 cm^-1 (Spectrum Two; PerkinElmer, Waltham, MA, United States). Each spectrum was obtained using a resolution of 1 cm^-1. This analysis was performed on proteins purified from three independent biological replicates, and representative spectra were presented. The spectrum absorption in the amide I region (1,600–1700 cm^-1) was analyzed using the Peakfit (Sea Solve Software Inc., Shanghai, China) (Vanga et al., 2020). The secondary structure of the ABAs-cFv protein was predicted using Phyre2 (https://www.sbg.bio.ic.ac.uk/phyre2/html) with default parameters (Kelley et al., 2015). The full-length amino-acid sequence of the expressed protein (VL–linker–VH; 239 aa, 25.6 kDa), including the C-terminal His-tag, was used for this analysis to ensure direct comparability with experimental data.

The concentrations of the purified ABA-scFv samples were determined using a Nanodrop (ThermoFisher, United States) and then diluted to 0.1 mg/mL with PBS containing 0, 25, 50, and 100 ng/μL. All measurements were performed on samples from three independent biological replicates. Parameters of the circular dichroism spectrometer (Chirascan, Applied Photophysics, England) were adjusted (wavelength range: 190–250 nm, optical path difference: 1 mm). The zero point and reference background (the blank solvent) were calibrated, and a special cuvette was used to avoid bubbles. A stable temperature (usually 20–25 °C) was maintained during data acquisition, an appropriate scanning speed was set to obtain a clear spectrum, and absorption differences between the sample and reference spectra were recorded. Baseline correction and smoothing denoising were performed during data processing, and the secondary structure and conformational changes were analyzed by curve fitting or software CDNN (Applied Photophysics).

Analysis of variance (ANOVA) and multiple comparisons (Duncan’s method) were performed using SPSS (IBM, United States). Relevant charts and graphs were created using Origin (OriginLab, United States).

The growth and synthesis of single-chain antibodies in different molecular chaperone strains are shown in Figure 1. After co-expression with molecular chaperones, only the pGro7 and pKJE7 treatments significantly increased the bacterial cell number, while other treatments showed no significant difference compared to the control. However, in terms of biomass, co-expression of molecular chaperones reduced biomass for all strains, with the pG-TF2 treatment showing the most notable decrease. The pGro7 and pKJE7 strains exhibited higher cell numbers than the control. However, their final biomass was approximately 50% lower, indicating that increased cell proliferation did not correlate with higher protein synthesis under these conditions.

In all treatments, the control showed relatively higher protein synthesis efficiency. When interpreting the volumetric yields, it is crucial to consider the significant variations in cell number and biomass across the treatments, which indicate differing levels of metabolic load imposed by the chaperone systems. Regarding the synthesis of single-chain antibodies, the quantifiable yield of ABA-scFv, as determined by His-tag ELISA, was lower in all chaperone co-expression systems compared to the control. In terms of expression activity, although the majority of the expressed scFv remained in the insoluble fraction for all strains, co-expression with pTf16 significantly increased the proportion of detectable soluble scFv to 19.65%, compared to 14.20% in the control strain (Figure 1b), indicating that pTf16 was the most effective in assisting protein folding. SDS-PAGE and Western blot analysis of this purified soluble protein confirmed the presence of a single, distinct band at the expected molecular weight of ∼26 kDa, with no smaller fragments indicative of protein degradation observed (Supplementary Figure S3).

Prior to checkerboard titration analysis, the integrity of the OVA–ABA conjugates was validated by intact MS deconvolution (Supplementary Figure S2b), which showed a clear ladder of conjugated species separated by Δm ≈ 246 Da. The distribution centered at k = 6–10 with an average conjugation number of 8.21, indicating high conjugation efficiency (99.16%) with negligible unmodified OVA (<1%). In addition, SDS-PAGE analysis revealed a single dominant band at ∼45 kDa (Supplementary Figure S1), confirming the purity and integrity of the conjugates. These results demonstrated that the OVA–ABA conjugates were suitable for immunoassay development.

We therefore proceeded to screen the optimal antigen coating concentration and antibody dosage for each variant in the checkerboard titration assay. It is important to note that this individual optimization was a critical step in our experimental design. As different molecular chaperone systems can distinctly influence the folding efficiency, stability, and binding characteristics of the expressed scFv, establishing a unique, optimized assay condition for each is essential. The objective of this study was not to compare the absolute binding affinities of the scFv variants under a single, potentially suboptimal condition for all. Instead, the goal was to evaluate the maximum functional performance of each ‘chaperone-antibody system’ to determine its suitability for specific applications. The experimental results indicate that, in each treatment, higher coating concentrations or antibody dosages resulted in higher OD values in the control wells and, correspondingly, higher OD values in the competitive binding wells. This result suggests that within the working concentration range (100 ng/mL ABA), excessively high concentrations of coated antigens and antibodies weaken the competitive binding between free ABA (target ABA) and the antibody, thereby reducing sensitivity. On the other hand, excessively low concentrations of coated antigens and antibodies are insufficient to provide enough binding sites for the primary antibody and enzyme-labeled secondary antibody. Therefore, appropriately diluted coating antigens and antibodies are crucial for maintaining a relatively low OD value in the competitive binding wells while ensuring a significant difference from the control wells, thus improving antibody detection sensitivity. Finally, we selected the following conditions: For the control (CK): antigen coating dilution 100 times (50 mg/L), antibody dilution 800 times (6.25 μg/L). For pG-KJE8: antigen coating dilution 1,600 times (3.125 mg/L), antibody dilution 100 times (0.05 mg/L). For pGro7: antigen coating dilution 1,600 times (3.125 mg/L), antibody dilution 100 times (0.05 mg/L). For pKJE7: antigen coating dilution 3,200 times (1.5625 mg/L), antibody dilution 800 times (6.25 μg/L). For pG-Tf2: antigen coating dilution 100 times (50 mg/L), antibody dilution 50 times (0.1 mg/L). For pTf16: antigen coating dilution 1,600 times (3.125 mg/L), antibody dilution 800 times (6.25 μg/L).

To further analyze the binding affinity of each antibody, the IC₅₀ of the antibodies was analyzed at the optimal antigen coating concentration and antibody dosage for each treatment, as shown in Figure 2 The standard curves of all treatments co-expressing and synthesizing ABA single-chain antibodies could be successfully fitted (R² ≥ 0.99). In terms of sensitivity and linear range (measured as the IC₂₀ to IC₈₀ interval), the antibodies exhibited different properties after assisted folding by various molecular chaperones.

The pKJE7-assisted scFv displayed the highest sensitivity, with an IC50 value 25.37% lower than that of the control; however, this was accompanied by a narrower linear detection range (IC₂₀-IC₈₀). In contrast, the pTf16 treatment significantly extended the detection linear range of the antibodies, but there was no significant improvement in sensitivity compared to the CK.

The cross-reactivity rates of different ABA single-chain antibodies were analyzed to verify the recognition specificity of ABA single-chain antibodies under different molecular chaperone-assisted syntheses (Table 2). The results showed that, except for the pG-Tf2 treatment, all molecular chaperone treatments significantly reduced the non-specific binding of ABA antibodies, particularly addressing the high cross-reactivity with indole-3-acetic acid (IAA) observed in the control. The co-expression with the pTf16 molecular chaperone resulted in the lowest cross-reactivity rate of ABA single-chain antibodies. Notably, the pG-Tf2 treatment significantly increased the cross-reactivity rates of the single-chain antibody with various substances, indicating that this treatment adversely affected the specificity of the ABA-scFv.

The impact of different molecular chaperones on the secondary structure of the ABA single-chain antibody was investigated using ATR-FTIR spectroscopy, with percentages of each structure quantified by fitting the amide I band. For comparison, the in silico structure prediction for the ABA-scFv indicated a conformation redominantly composed of β-sheet (54%) and disordered regions (35%), with a notable absence of α-helices (0%) (Figure 3).

Our experimental findings revealed significant structural alterations across the treatments (Figure 4). A key finding was in the β-sheet content, where the pKJE7 treatment produced a structure with 48.2% β-sheet, the most similar proportion to the predicted value of 54%. The pTf16 treatment, with a β-sheet content of 43.4%, ranked as the second most similar to the prediction. Another critical finding was in the α-helix content; the pTf16 system produced an scFv with 0% α-helix, perfectly matching the prediction, whereas the pKJE7 treatment resulted in 12.5% α-helix content, a notable deviation from the theoretical model.

Based on these findings, while no single chaperone perfectly replicated the predicted structure, pKJE7 was most effective in forming a native-like β-sheet core, whereas pTf16 was most effective in preventing the formation of non-native α-helical structures.

To gain mechanistic insight into the antigen recognition event, circular dichroism spectroscopy was used to monitor the conformational response of each scFv variant upon binding to ABA (Figure 5). The data revealed two fundamentally different modes of interaction depending on the chaperone system used.

First, the scFv produced by the pKJE7, pG-KJE8, and pTf16 systems exhibited a conformationally rigid response to ABA recognition. Their secondary structures remained stable across all tested ABA concentrations. For instance, the β-sheet content of the pKJE7 variant only fluctuated between 15.8% and 16.1%. This structural stability upon antigen binding suggests these chaperones produce scFv that exists in a pre-formed, ‘binding-ready’ state, consistent with what is expected for a lock-and-key mechanism.

In stark contrast, the scFv from the control (CK), pGro7, and pG-Tf2 systems displayed a conformationally flexible response, undergoing significant structural changes upon interaction with ABA. This was most evident in the pG-Tf2 variant, which showed a substantial decrease in β-sheet content from a baseline of 16.1%–11.1%. This structural rearrangement upon antigen binding suggests an ‘induced fit’ recognition mechanism, where the antibody must alter its conformation to achieve optimal binding.