Antigen-binding affinity and thermostability of chimeric mouse-chicken IgY and mouse-human IgG antibodies with identical variable domains

Constant (C)-region switching of heavy (H) and/or light (L) chains in antibodies (Abs) can affect their affinity and specificity, as demonstrated using mouse, human, and chimeric mouse-human (MH) Abs. However, the consequences of C-region switching between evolutionarily distinct mammalian and avian Abs remain unknown. To explore C-region switching in mouse-chicken (MC) Abs, we investigated antigen-binding parameters and thermal stability of chimeric MC-6C407 and MC-3D8 IgY Abs compared with parental mouse IgGs and chimeric MH Abs (MH-6C407 IgG and MH-3D8 IgG) bearing identical corresponding variable (V) regions. The two MC-IgYs exhibited differences in antigen-binding parameters and thermal stability from their parental mouse Abs. However, changes were similar to or less than those between chimeric MH Abs and their parental mouse Abs. The results demonstrate that mammalian and avian Abs share compatible V-C region interfaces, which may be conducive for the design and utilization of mammalian-avian chimeric Abs.

Both humans and mice express Abs with five classes of H chain (μ, δ, γ, ε, and α) and two classes of L chain (κ and λ) comprising different IgM, IgD, IgG, IgE, and IgA isotypes. By contrast, chickens express only three classes of H chain (υ, μ, and α) and a single type of L chain (λ) comprising IgY, IgM, and IgA isotypes. IgY, the major Ab in chickens, is present at high concentrations in serum and egg yolk, and is also transferred from hens to embryos via the egg yolk. IgYs have a molecular mass of ~180 kDa with two H (67-70 kDa each) and two L (25 kDa each) chains that are structurally similar to mammalian IgE comprising four C H domains that lack a hinge region 29 , but they are functionally similar to mammalian IgGs. IgY has two N-glycosylation sites, one in each of the Cυ2 and Cυ3 domains, whereas mammalian IgG has a single N-glycosylation site on the Cγ2 domain, and IgE has seven N-linked glycosylation sites spread across the Cε chain 30 .
Knowledge of how C-domain switching between mammalian and avian Abs affects Ab properties could potentially lead to biotechnology applications for mammalian-avian chimeric Abs. In the present study, we explored C → V allosteric signaling by replacing the C domain (C γ or C κ ) of mammalian IgG Ab with the corresponding C υ or C λ domain of avian IgY. To investigate how inter-species class switching contributes to Ab properties, we prepared three V domain-identical (parental mouse IgG, chimeric MH-IgG, and chimeric mouse-chicken [MC]-IgY) and two monoclonal (6C407 and 3D8) Abs. We analyzed their antigen-binding parameters and thermodynamic stability using surface plasmon resonance (SPR), intrinsic protein fluorescence, enzyme-linked immunosorbent assay (ELISA), and size-exclusion chromatography (SEC). Both the affinity and thermal stability of MC-6C407 and MC-3D8 IgYs were essentially comparable to those of their respective parental mouse IgGs. By contrast, MH-3D8 IgG differed markedly in terms of thermodynamic stability from its parental mouse 3D8 IgG.
From our results, we concluded that the C domains of chicken Abs are compatible with the V domains of mouse Abs, and can therefore transduce C → V allosteric signals that modulate the features of Abs without destroying them. Our results may assist the design and use of chimeric mammalian-avian Abs. To our knowledge, this is the first study to report the effects of C-region switching between mammalian and avian Abs.
Each panel includes V region-identical mouse IgG2a/κ, chimeric MH-IgG1/κ, and chimeric MC-IgY/λ Ab types. The predicted structures of the expressed proteins are shown in Fig. 1a. Amino acid sequences for the C regions of mouse IgG2a/κ, human IgG1/κ, and chicken IgY/λ are shown in Fig. 1b. Two original mouse Abs, 6C407 IgG2a/κ and 3D8 IgG2a/κ, were purified from the culture supernatant of their respective hybridoma cells using Protein L-agarose resin. The two chimeric MH-IgGs were purified from the culture supernatant of FreeStyle 293-F cells transfected with the specific KV10 vectors using Protein A-agarose resin at 7 days post-transfection, and the two chimeric MC-IgYs were purified using Protein L-agarose. SDS-PAGE analysis showed that mouse IgGs (6C407 and 3D8) and MH-IgGs (6C407 and 3D8) were of the expected sizes (~50 and ~25 kDa, respectively) under reducing conditions and >150 kDa under non-reducing conditions (Fig. 2a). Interestingly, MC-IgYs (6C407 and 3D8) yielded a single band of ~70 kDa for the υ H chain, and two distinct bands (~25 and ~27-28 kDa) for the λ L chain under reducing conditions. MC-IgYs also yielded two distinct bands (>150 and >200 kDa) under non-reducing conditions. This two-band appearance for IgY molecules has not been reported previously.
Next, we investigated whether chimeric MC-IgY proteins expressed in FreeStyle 293-F cells were glycosylated, based on the assumption that differences in glycosylation state may explain the two bands running at ~25 kDa under reducing conditions and the presence of two bands under non-reducing (native) PAGE. Coomassie Blue staining yielded protein bands with a slightly lower molecular mass following reaction with the deglycosylation enzyme cocktail (Fig. 2b, upper panel). In parallel, the fetuin positive control for deglycosylation shifted from 64 to 55 kDa following deglycosylation enzyme cocktail treatment. Signals corresponding to the H and L chains of MC-IgYs were not detected by PAS staining after reaction with the deglycosylation enzyme cocktail, whereas a signal was detected with unreacted MC-IgYs (Fig. 2b, lower panel). This indicates that both υ H and λ L chains in MC-IgYs were glycosylated. Thus, the presence of two bands in MC-IgYs could be explained by partial glycosylation of both Cυ H and Cλ L chains. This is in contrast with previous reports for natural IgY Abs purified from chicken serum in which IgYs contain two potential N-glycosylation sites located in Cυ2 and Cυ3 domains 30 . Class switching to chicken C regions affects antigen-binding parameters. Next, we measured binding parameters (k on , k off , and K D ) for the three-member panels of 6C407 and 3D8 Abs by SPR ( Fig. 3 and Table 1). The binding affinity of MH-6C407 was almost the same as that of parental mouse 6C407, whereas MC-6C407 displayed ~2-fold higher affinity than parental mouse 6C407. In the case of 3D8 Abs, the binding affinity of chimeric MC-3D8 was almost the same as that of parental mouse 3D8, while that of MH-3D8 was ~2-fold lower than that of parental mouse 3D8. These results indicate that mouse Abs can tolerate the replacement of their C regions (C γ and C κ ) by the corresponding avian C regions (C υ and C λ ) without destroying activity, consistent with the tolerance to switching of C regions between mammalian Abs, although antigen-binding kinetics are affected to some extent.
Structural stability of Abs under thermal stress. We compared the structural stability of the six Abs (three 6C407 and three 3D8 Abs) under thermal stress by analyzing intrinsic protein fluorescence using a Tycho NT.6 system. Parental mouse 6C407 and MC-6C407 yielded a single inflection temperature (Ti) value that is the temperature at which an unfolding transition occurs, calculated from the F350/F330 ratio of the fluorescence intensity at 350 vs. 330 nm, where tryptophan and tyrosine fluoresce in the unfolded and folded states, respectively. By contrast, three Ti values (38.6 °C, 42.1 °C, and 78.8 °C) were detected for MH-6C407, indicating three unfolding events (Fig. 4a,b). The major Ti values for the three 6C407 Abs were 84.6 °C (Ti 1), 78.8 °C (Ti 3), and 72.9 °C (Ti 1) for mouse 6C407, MH-6C407, and MC-6C407, respectively (Fig. 4a,b and Table 2). Thus, compared with mouse 6C407, both MH-and MC-6C407 underwent a leftward shift in Ti value to lower temperature ( Fig. 4a,b), and this shift to lower temperature was more pronounced in MC-6C407 than MH-6C407.
In the 3D8 three-member panel, three Ti values were detected for parental mouse 3D8, compared with a single Ti value for the two chimeric 6C407 Abs. A leftward shift in major Ti value to lower temperature was detected for MH-3D8 (76.7 °C) and MC-3D8 (76.3 °C), compared with parental mouse 3D8 (83.7 °C), indicating a decrease in thermal stability for these Abs (Fig. 4c,d and Table 2). Interestingly, MH-3D8 displayed a dramatic increase in F350/F330 ratio at 85 °C, indicating the existence of a domain that retains structural stability over 90 °C, but this temperature could not be achieved using this measurement system. Taken together, the results suggest that replacement of the mouse C domain with a chicken C domain causes a decrease in thermal stability to a slightly greater extent than C-region switching between mouse and human Abs.
The change in F350/F330 ratio differed at the beginning and end of the thermal profile, and was greatest for the two chimeric MC Abs (0.1892 and 0.2487), followed by the two chimeric MH Abs (0.1473 and 0.2190), and the parental mouse Abs (0.1270 and 0.1031), as shown in Table 2. Accordingly, the structures of MC-chimeric Abs appear to be more susceptible to thermal stress than those of IgGs.

Functional stability and aggregation/degradation behavior of Abs under thermal stress.
We analyzed the antigen-binding activity and aggregation/degradation behavior by ELISA and SDS-PAGE, respectively, after heating Abs for 10 min to 4 h at 70-90 °C. The three 6C407 Abs displayed comparable functional www.nature.com/scientificreports www.nature.com/scientificreports/ stability, and all retained antigen-binding activity after heating at 60 °C for 4 h and at 70 °C for 10 min (Fig. 5). A loss in antigen-binding activity was first observed after heating at 70 °C for 30 min, and thermal stress at 80 °C for 10 min caused the loss of more than ~90% of activity (Fig. 5a,c,e).
Noncovalent aggregation and degradation behavior analyzed by reducing SDS-PAGE was similar among the three types of Abs, with all undergoing dramatic aggregation following heating at 80 °C and 90 °C for 2 h (Fig. 5b,d,f). Protein bands above 50 kDa, observed with increasing incubation time and temperature, are thought to be temporary aggregation intermediates undergoing protein degradation 31 .
The three types of 3D8 Abs displayed distinct differences in functional stability (Fig. 6). Parental mouse 3D8 lost ~90% of its antigen-binding activity after incubation at 60 °C for 2 h, whereas MC-3D8 retained activity after incubation at 60 °C for 2 h but lost ~90% of the activity after harsher thermal stress treatment at 70 °C for 10 min (Fig. 6a,e). Protein bands corresponding to mouse 3D8 IgG, MH-3D8, and MC-3D8 disappeared following thermal stress at 80 °C and 90 °C for 2 h (Fig. 6b,d,f). Interestingly, the activity of MH-3D8 was increased following thermal stress at 70 °C for 2 h, 70 °C for 4 h, and 80 °C for 10 min (Fig. 6c). However, inconsistently, the affinity of these heat-treated MH-3D8 molecules measured by SPR was ~10-fold lower than that of untreated MH-3D8 ( Fig. 7a and Table 3). This discrepancy may be caused by differences in experimental procedures; an anti-Fc Ab was used in ELISA but not in SPR experiments. In ELISA, the Fc structure may be altered within a specific window of thermal stress, and this may be better recognized by an anti-Fc Ab. The SEC profile showed single peaks corresponding to an apparent 150 kDa IgG (Fig. 7b), although slight aggregation was observed for heat-treated MH-3D8 in SDS-PAGE (Fig. 6d), indicating no aggregation or degradation. Therefore, it seems that a subtle change in MH-3D8 structure, rather than aggregation, is responsible for the enhanced DNA-binding activity within a specific window of thermal stress. This conclusion is supported by the emergence of an additional unfolding transition only for MH-3D8 at temperatures >90 °C (Fig. 4d). An Ab region corresponding to this shift in unfolding transition may be the Fc region that is not responsible for DNA binding.

Discussion
In this study, we demonstrate that chimeric MC-IgY Abs allow modulation of antigen-binding parameters and thermal stability, as has been described for chimeric MH-IgGs. Both V and C domains are responsible for Ab structure and function, as demonstrated recently by analyzing different mammalian (mouse, human, and chimeric MH) Abs with identical V regions. The current study further supports this notion through C-region switching between mammalian and avian Abs with identical V regions. This study also expands our understanding of C-region switching between Abs of evolutionarily distant species, and the findings may have implications for improving Ab engineering, and thus possible utilization of Abs in biotechnology. www.nature.com/scientificreports www.nature.com/scientificreports/ Most previous studies on the effects of class switching have focused on C H switching of Abs within a species, while C L regions are not switched. V region-identical mouse IgG1, IgG2a, IgG2b, and IgG3 Ab subclasses exhibited differences in affinity and thermodynamics. Moreover, the affinity and thermodynamics between IgG2a and its Fab form (IgG2a-Fab), as well as between IgG3 and its Fab form (IgG3-Fab), are known to differ 10 . Structural differences imposed by the C H chain of Abs have been proven using small-angle X-ray scattering measurements 14 . V region-identical human IgA1/κ, IgG1/κ, and even Fab fragments derived from these Abs exhibit significantly different affinities 23 . Thermodynamic analysis of the binding of IgGs and their Fabs showed that Fc Chimeric MC (IgY/λ) (5.11 ± 0.01) × 10 5 (6.32 ± 0.03) × 10 −4 1.24 × 10 −9

3D8
Mouse IgG2a/κ (6.04 ± 0.15) × 10 6 (1. 36  www.nature.com/scientificreports www.nature.com/scientificreports/ regions contribute strongly to affinity by inducing conformational changes in V regions through an allosteric mechanism 32 . Even mutation at two amino acids in IgA Fc (C H α3 domain) resulted in a ~20-fold reduction in antigen-binding affinity compared with that of wild-type IgA 33 . These studies demonstrate that conformational changes in the V region can be dictated by C H domains (such as hinge and Fc regions), and are not limited by the C H 1 domain. On the other hand, studies focusing on C L switching were performed using the Fab format rather than the full-size Ig format 34,35 . C κ -to-C λ chain switching in catalytic Abs can alter structure and function 34 , and the thermodynamic stability of human Fabs can differ depending on the type of C L chain (C κ or C λ ) present in the Ab 35 . Class switching and the structural effects of V regions can differ depending on which class of C H or C L is present, and whether full-size IgG or Fab forms are studied. Since the chimeric MC-IgY Abs in the present study were produced by replacing both mouse C γ and C κ domains with chicken C υ and C λ domains, changes in the antigen-binding parameters and thermal stability of MC-IgY Abs are likely due to the properties of C υ and C λ chains that mediate an intrinsic C → V allosteric signal.
Two chimeric MC Abs (MC-6C407 and MC-3D8 IgYs) displayed only subtle differences in antigen-binding kinetics and thermal stability, compared with parental mouse Abs. This suggests that the C regions (C υ and C λ ) of chicken IgYs are compatible with the V regions of mouse Abs, even though birds and mammals diverged from their common ancestor >300 million years ago 36 . This compatibility of V-C region interfaces between mouse   Table 2. Summary of the thermal shift analysis. Initial ratio: the 350 nm/330 nm fluorescence ratio at the start of the experiment (at 35 °C). ∆Ratio: difference between the ratio at the beginning and at the end of the thermal profile. Ti, the inflection temperature at which an unfolding transition occurred. and chicken Abs is supported by a report showing that chimeric chicken-human Abs composed of chicken V domains and human C domains exhibit similar binding affinity to parental chicken Abs 37 . Furthermore, V-C region compatibility between these species has been observed in a chickenized single-chain variable fragment (scFv) Ab harboring complementarity-determining regions (CDRs) of the mouse Ab and framework regions (FRs) of a chicken Ab 38 . IgYs do not show cross-reactivity to mammalian IgGs, i.e., Abs raised against IgYs do not bind to mammalian IgGs due to immunological differences between IgYs and mammalian IgGs. Also, the C region of IgYs does not bind to mammalian Fc receptors. This feature makes chicken IgY Abs particularly valuable in experimental research and in immunodiagnostic assays, such as immunohistochemistry, Western blotting, flow cytometry, and ELISA, all of which require low signal-to-noise ratios. These properties of IgYs, which mainly stem from the C region of IgY, could also be shared by chimeric MC-IgYs. The current results suggest that Ab engineering using the C region of chicken IgYs may expand the biotechnological applications of chimeric MC-IgYs. For example, indirect sandwich ELISA usually requires two Abs derived from different species that recognize different epitopes, and one of two Abs can be obtained by replacing the C region of a mammalian IgG with the C region of a chicken IgY following the production of two monoclonal IgG Abs from a single mammalian host. In other words, construction of chimeric MC-IgYs from mammalian IgGs could eliminate the inconvenience of using antibodies from different mammalian species in laboratory assays.
The thermodynamic stability of Abs is associated with the subclass of human IgGs rather than the V domains 39 . However, all three V region-identical types of 6C407 Abs (mouse 6C407, MH-6C407, and MC-6C407) were similar in terms of thermal stability evaluated by antigen-binding activity (Fig. 5), but there were remarkable differences between the three V region-identical types of 3D8 Abs (Fig. 6). This suggests that both V and C domains of Abs can influence their thermal stability. Interestingly, analysis of thermal stability of Abs based on antigen-binding activity was not consistent with analysis based on monitoring fluorescence. The fluorometric method indicated that the conformational structure of the two chimeric MC-IgYs (MC-6C407 and MC-3D8) was more susceptible to thermal stress than that of their corresponding parental mouse IgGs and MH-IgGs ( Fig. 4 and Table 2), demonstrating that MC-IgYs are thermodynamically less stable. This tendency has been reported in a previous study in which the thermal stability of a chicken IgY measured by antigen-binding activity was similar www.nature.com/scientificreports www.nature.com/scientificreports/ to that of cow, goat, and pig IgGs, whereas the conformational stability of the chicken IgY measured using a fluorometric method was lower 40 .
Only correctly folded and assembled Abs that pass through the endoplasmic reticulum (ER) quality-control system are secreted from Ab-producing cells 41 . Under the ER quality-control system, typical IgGs assemble first as H chain dimers that are incompletely folded and associated with the molecular chaperone of binding immunoglobulin protein (BiP) in the ER until they associate with cognate chains. Complete folding of C H 1 occurs only after interaction with the folded C L domain 42,43 . The ER quality-control machinery, first identified in B lineage cells, is now known to function in Chinese hamster ovary (CHO) cells used for therapeutic Ab production 44 . Recently, however, it was found that this machinery does not work on some engineered IgG Abs such as humanized IgGs, as evidenced by secretion of dimeric H chain-only IgGs not associated with L chains in CHO cells 45 . Thus, evading this machinery may be dependent on the intrinsic properties of the V H domain because engineered V H domains of IgGs confer complete folding on C H 1 domains to evade ER retention 45 . We predict that this machinery may also function in FreeStyle 293-F cells used for Ab production in the present work. Chimeric MH-IgG and MC-IgY Abs were efficiently secreted in an assembled form from FreeStyle 293-F cells (Fig. 2a), yielding ~5 mg antibody per 100 ml cell culture. Therefore, these chimeric Abs might pass the ER quality-control system, suggesting that mouse V H domains confer incomplete folding on chicken C υ 1 and human C γ 1 domains until they are associated with chicken C λ and human C κ domains, respectively, in the ER.  Table 3. Binding kinetics and affinity a of Ig proteins for their antigens b . a The K D values were calculated by analyzing at least five data sets using different protein concentrations. b The single-stranded oligodeoxynucleotide labeled with biotin at the 5′ end [bio-ss-(dN) 40 ].
Size-exclusion chromatography (SEC). SEC analyses of purified Abs were performed using a DGU-20A3 UFLC system (Shimadzu) fitted with a TSK G3000SWXL column (7.8 × 300 mm; Toso Haas). Ab proteins were diluted with PBS to 1 mg/ml, and 30 μl of diluted Ab solution was injected onto the column. Where necessary, the protein was heated for 10 min, 2 h, or 4 h at 70 °C prior to injection. The mobile phase was 100 mM HEPES/85 mM HNaSO 4 (pH 6.8), and the flow rate was 1 ml/min. Chromatograms were obtained by monitoring the absorbance at 280 nm.