Epitope Mapping of Neutralizing Monoclonal Antibodies to Human Interferon-γ Using Human-Bovine Interferon-γ Chimeras

Abstract

Our aim was to identify conformational epitopes, recognized by monoclonal antibodies (mAbs) made against human (h) interferon (IFN)-γ. Based on the mAbs' (n = 12) ability to simultaneously bind hIFN-γ in ELISA, 2 epitope clusters with 5 mAbs in each were defined; 2 mAbs recognized unique epitopes. Utilizing the mAbs' lack of reactivity with bovine (b) IFN-γ, epitopes were identified using 7 h/bIFN-γ chimeras where the helical regions (A-F) or the C terminus were substituted with bIFN-γ residues. Chimeras had a N-terminal peptide tag enabling the analysis of mAb recognition of chimeras in ELISA. The 2 mAb clusters mapped to region A and E, respectively; the epitopes of several mAbs also involved additional regions. MAbs in cluster A neutralized, to various degrees, IFN-γ-mediated activation of human cells, in line with the involvement of region A in the IFN-γ receptor interaction. MAbs mapping to region E displayed a stronger neutralizing capacity although this region has not been directly implicated in the receptor interaction. The results corroborate earlier studies and provide a detailed picture of the link between the epitope specificity and neutralizing capacity of mAbs. They further demonstrate the general use of peptide-tagged chimeric proteins as a powerful and straightforward method for efficient mapping of conformational epitopes.

FIG. 1.

Structure of IFN-γ and human-bovine IFN-γ chimeras. (A) Schematic drawing of the IFN-γ homodimer with helical regions shown as cylinders interconnected by nonhelical sequences shown as thin tubes. The respective monomers are indicated by dark and light gray with the pointed part of each helix pointing toward the C terminus. The figure was drawn using the program CN3D (www.ncbi.nlm.nih.gov/Structure/CN3D/cn3d.shtml) based on X-ray chrystallography data for bIFN-γ (Randal and Kossiakoff, 2000), which is very similar to hIFN-γ. (B) The aligned amino acid sequences of hIFN-γ and bIFN-γ are shown with helical regions boxed and denoted A through F. Chimeric constructs A-F and the nonhelical CT from bIFN-γ were based on hIFN-γ with the respective regions substituted with the corresponding bIFN-γ sequence. The sequence shown does not include the signal peptide. The alignment was made using Clustal W (www.ch.embnet.org/software/ClustalW.html; Larkin and others 2007). Symbols represent amino acids with identical (*), similar (:), or partly similar (.) side chains; below highly different amino acids, no symbol is shown. Amino acids shown by X-ray crystallography to interact with IFNGR1 are underlined (Thiel and others 2000) as is the KRKRS motif in CT implicated in receptor interaction (Döbeli and others 1988). IFN, interferon.

FIG. 2.

Epitope mapping by sandwich ELISA. All mAbs were analyzed in all possible combinations as capture and detection mAbs for reactivity with hIFN-γ in ELISA. (A) Example graphs are shown for mAbs MT126L, 7-B6-1, G23, and 11i used for capture in combination with each mAb in the panel as detection mAb. The analysis was made with 100 pg/mL of hIFN-γ and the background (0 pg/mL) was subtracted. When G23 and 11i was used as detection mAbs, 1,000 pg/mL of hIFN-γ was used since these mAbs yielded poor signals with 100 pg/mL when used as detection mAbs (indicated by gray bars). (B) The graph summarizing the functionality of mAb combinations is based on a cutoff definition for a positive signal. A signal was defined as positive for both mAbs (black boxes) if the absorbance value was >0.5 and if the signal was 2.5× higher than the signal obtained when using either of the mAbs in a homologous combination (i.e., the same mAb used for capture as well as detection). If only one mAb fulfilled the criteria the result is shown as >cutoff for one mAb (gray boxes). Combination of mAbs where none of the mAbs fulfilled the criteria are shown as <cutoff for both mAbs (white boxes). Data shown are the mean of 2 experiments.

FIG. 3.

Epitope mapping by competitive ELISA. Each mAb, used as a capture mAb, was allowed to bind biotinylated hIFN-γ that had been premixed with 0.2 μg/mL of a competitive mAb. The concentration of hIFN-γ used for each capture mAb was determined by their IC50 value (half the maximal absorbance value) when mAbs were allowed to bind a serial dilution of hIFN-γ in the absence of any competitive mAb (Table 1). (A) The capacity of a competitive mAb to inhibit the binding of hIFN-γ by the capture mAb is shown in graphs for MT126L, 7-B6-1, G23, and 11i. (B) The graph summarizing the competitive ELISA data shows the inhibition for each combination of capture and competitive mAb as 100%–81% (black boxes), 80%–60% (gray boxes), and <60% (white boxes). Data shown are the mean of 2 experiments.

FIG. 4.

Identification of epitope regions using chimeric human-bovine IFN-γ. Seven hIFN-γ constructs, each with one helical region (A–F) or the C terminus (CT) substituted with the corresponding bIFN-γ region, were compared with WT hIFN-γ for binding to all mAbs. MAbs were used as capture mAbs and allowed to bind serial dilutions of the chimeras and WT hIFN-γ followed by detection of bound protein using an anti-tag mAb. (A) Example graphs are shown for MT126L, 7-B6-1, G23 and 11i. bIFN-γ was not recognized by any mAb (not shown). (B) Summary of the chimeras causing a decreased binding by the respective mAbs. The loss of binding is shown as a complete loss (*), >75% (black boxes), 50%–75% (gray boxes), and <50% (white boxes). The percentual loss of binding was calculated by comparing the concentration of each chimera yielding half the maximal absorbance value against the concentration of WT IFN-γ yielding half the maximal absorbance value. Data shown are the mean of 2 experiments.

FIG. 5.

IFN-γ-mediated activation of human cells with human or bovine wild-type IFN-γ or human-bovine IFN-γ chimeras. (A) HEK BLUE™ cells were incubated with serial dilutions of human IFN-γ WT (hWT), human-bovine IFN-γ chimeras, or bovine IFN-γ WT (bWT). After 20 h incubation, ALP secreted into the supernatant in response to IFN-γ activation was measured in an ELISA reader at 650 nm. bIFN-γ did not activate the cells. (B) Human aortic endothelial cells, responding to hIFN-γ by expression of MHC class II, were incubated with 0.5 and 5 ng/mL of hWT, human-bovine IFN-γ chimeras or bWT. After 48 h incubation, MHC class II expression was analyzed by flow cytometry and 10,000 events were acquired. bIFN-γ did not activate the cells.

FIG. 6.

Ability of mAbs to neutralize IFN-γ-mediated activation of human cells. (A) HEK BLUE™ cells were incubated with 100 pg/mL hIFN-γ, with or without serial dilutions of mAbs to hIFN-γ. IFN-γ-induced enzyme secretion was measured after 20 h. The addition of anti-IFN-γ mAbs in the absence of hIFN-γ had no effect on the cells (not shown). Addition of isotype control mAbs to IFN-γ-activated cells had no neutralizing effect (not shown). Data shown are one of 3 representative experiments. (B) Neutralization of IFN-γ-induced activation of human aortic endothelial cells. Endothelial cells were incubated with 1 ng/mL of recombinant hIFN-γ, with or without serial dilutions of selected mAbs to hIFN-γ. IFN-γ-mediated MHC class II expression was analyzed by flow cytometry after 48 h. The graph to the left shows unstimulated cells (hatched line) and IFN-γ-stimulated cells without neutralizing mAb (solid line, gray field). The graphs to the right show neutralization with 4 different mAbs at 3 concentrations. The addition of anti-IFN-γ mAbs in the absence of hIFN-γ had no effect on the cells (not shown). Addition of isotype control mAbs to IFN-γ-activated cells had no neutralizing effect (not shown). Data shown are one of 2 experiments with reproducible results.