Immunochemical characterization of antigenic domains on human interferon-beta: spatially distinct epitopes are associated with both antiviral and antiproliferative activities

Immunogenicity of an interferon-beta1a product

In order to determine whether Blastoferon®, a biosimilar interferon (IFN)- beta 1a formulation, shares epitopes with other known IFN-beta products, a series of neutralization bioassays were performed with a set of well-characterized anti-IFN- beta monoclonal antibodies and human sera (World Health Organization Reference Reagents). The bioassay was the interferon-induced inhibition of virus cytopathic effect on human cells in culture (EMC virus and A-549 cells). Computer-calculated results were reported as Tenfold Reduction Units (TRU)/ml. To further assess Blastoferon® immunogenicity, in vivo production of anti-IFN beta antibodies was determined in sera of patients included in the pharmacovigilance plan of Blastoferon® by the level of IFN- beta 1a binding antibodies (by enzyme immunoassay -EIA) and neutralizing antibodies (in the Wish-VSV system). The highly characterized neutralizing monoclonal antibodies A1 and A5 that bind to specific regions of the IFN- beta molecule reacted positively with the three beta 1a IFNs: Blastoferon®, Rebif®, and the IFN- beta WHO Second International Standard 00/572. As expected, the non-neutralizing monoclonal antibodies B4 and B7 did not neutralize any of the IFN- beta preparations. The commercially available monoclonal antibody B-02 reacted essentially equally with Rebif® and Blastoferon®. The WHO Reference Reagent human serum anti-IFN- beta polyclonal antibody neutralized all the IFN- beta products, whereas the WHO Reference Reagent human serum anti-IFN-alpha polyclonal antibody G037-501-572 appropriately failed to react with any of the IFN- beta products. On the basis of in vitro reactivity with known, well-characterized monoclonal and polyclonal antibody preparations, Blastoferon® shares immunological determinants with other human interferon- beta products, especially IFN- beta 1a. In vivo antibodies were detected by EIA in 72.9% of 37 chronically treated multiple sclerosis patients, whereas neutralizing antibodies were found in 8.1% of them. Blastoferon® appears to have immunological characteristics comparable to other IFN- beta 1a products.

Mapping of IFN-beta epitopes important for receptor binding and biologic activation: comparison of results achieved using antibody-based methods and alanine substitution mutagenesis

The epitopes important for receptor binding and activation of human interferon-beta1a (IFN-beta1a) were mapped with monoclonal antibodies (mAb), grouped on the basis of their specificity and ability to neutralize biologic activity, and alanine scanning mutagenesis (ASM). The binding properties of nine mAb were defined, using ASM-IFN-beta mutants having alanine substituted at targeted, surface-exposed residues. The results were correlated with the mAb neutralizing potency. Of six mAb that bound either at or adjacent to the IFNAR-2 receptor chain binding site defined by the ASM epitopes, only three had measurable neutralizing activity. Two of these inhibited IFN-beta/IFNAR-2 complex formation, suggesting that steric hindrance of receptor binding constitutes their mechanism of neutralization. However, two mAb that bound to sites remote from the IFNAR-2 binding site on IFN-beta also inhibited IFN-beta/IFNAR-2 complex formation and demonstrated potent neutralizing activity. Thus, neutralizing mAb may employ mechanisms other than steric blockade to inhibit directly the binding of receptor by cytokine, limiting their usefulness as tools to define precise receptor-ligand interaction sites.

Conformational and sequential epitopes on the human granulocyte-colony stimulating factor molecule (hG-CSF) and their role in binding to human placenta receptors

Monoclonal antibodies (mAbs) named 8C2 and 6E3, directed against the recombinant human granulocyte colony-stimulating factor (hG-CSF), were used as probes to study the cytokine orientation on its binding to receptors from human placenta. Competition enzyme linked immunoabsorbent assays (ELISA) revealed that mAb 8C2 would be directed to a linear epitope, whereas mAb 6E3 would delimit a more assembled epitope. Gel-filtration high performance liquid chromatography (HPLC) of the immune complexes formed by incubating [(125)I]hG-CSF with each mAb showed that epitope 8C2, but not 6E3, was altered after cytokine iodination. In addition, mAb 6E3 completely inhibited [(125)I]hG-CSF binding to human placental microsomes. Although [(125)I]mAb 6E3 was unable to bind to preformed hG-CSF-receptor complexes, [(125)I]mAb 8C2 did recognize hG-CSF previously bound to receptors, suggesting that epitope 8C2 would remain accessible in the hG-CSF-receptor complex. To identify the cytokine region defined by mAbs, hG-CSF was digested with different proteolytic enzymes: Arg-C, Glu-C, trypsin and alpha chymotrypsin. Immunoreactivity of the resulting peptides was examined by Western blot and their sequences were established by Edman degradation. Results showed that mAb 6E3 would be directed to a conformation-dependent epitope located close to the hG-CSF binding domain and included into the sequence 1-122/123, whereas mAb 8C2 recognized the region 41-58, which represents a linear epitope left exposed after cytokine binding to receptors from human placenta.

The neutralization of interferons by antibody. I. Quantitative and theoretical analyses of the neutralization reaction in different bioassay systems

The highly specific ability of antibodies to inhibit the biologic activity of cytokines or other therapeutic proteins is widely used in research and a subject of increasing clinical importance. The need exists for a standardized approach to the reporting of neutralizing antibody potency soundly based on theoretical and practical considerations and tested by experimental data. Pursuant to the original studies of Kawade on the theoretical and functional aspects of neutralization of interferons (IFN), experimental data were obtained by different laboratories employing varied methodology to address two hypotheses concerning the nature of IFN neutralization reactions, based on a derived formula that allows expression of neutralizing power as the reduction of 10 laboratory units (LU)/ml to 1 LU/ml, the end point of most bioassays. Two hypotheses are posed: (1) antibody acts to neutralize a fixed amount of biologically active IFN molecules, or (2) antibody reduces IFN activity in a set ratio of added/residual biologically active IFN. The first, or fixed amount, hypothesis relates to the reactivity of high-affinity antibodies neutralizing equimolar amounts of antigen, whereas the second, or constant proportion, hypothesis postulates a reduction in the ratio of total added IFN to residual active IFN molecules, such as a low-affinity antibody might exhibit. Analyses of data of the neutralization of IFN-alpha and IFN-beta are presented, employing human polyclonal antibodies and murine monoclonal antibodies (mAb). The theoretical constructs of Kawade are extended in the Appendix and correlated with new experimental data in the text. The data clearly indicate that the low-antibody affinity, constant proportion hypothesis, rather than the high-antibody affinity, fixed amount hypothesis, is applicable, if the bioassay is sensitive to IFN. The findings presented here and in the following paper (pp. 743-755, this issue) taken together provide the basis for a standardized method of expression of neutralizing potency and substantiate the earlier operational 10/1 LU/ml approach recommended by the World Health Organization. The accompanying paper relates neutralization results to the sensitivity of the bioassay to IFN and describes the rationale for a recommended unit of antibody neutralization.

Systematic mutational mapping of sites on human interferon-beta-1a that are important for receptor binding and functional activity

A systematic mutational analysis of human interferon-beta-1a (IFN-beta) was performed to identify regions on the surface of the molecule that are important for receptor binding and for functional activity. The crystal structure of IFN-beta-1a was used to design a panel of 15 mutant proteins, in each of which a contiguous group of 2-8 surface residues was mutated, in most instances to alanine. The mutants were analyzed for activity in vitro in antiviral and in antiproliferation assays, and for their ability to bind to the type I IFN (ifnar1/ifnar2) receptor on Daudi cells and to a soluble ifnar2 fusion protein (ifnar2-Fc). Abolition of binding to ifnar2-Fc for mutants A2, AB1, AB2, and E established that the ifnar2 binding site on IFN-beta comprises parts of the A helix, the AB loop, and the E helix. Mutations in these areas, which together define a contiguous patch of the IFN-beta surface, also resulted in reduced affinity for binding to the receptor on cells and in reductions in activity of 5-50-fold in functional assays. A second receptor interaction site, concluded to be the ifnar1 binding site, was identified on the opposite face of the molecule. Mutations in this region, which encompasses parts of the B, C, and D helices and the DE loop, resulted in disparate effects on receptor binding and on functional activity. Analysis of antiproliferation activity as a function of the level of receptor occupancy allowed mutational effects on receptor activation to be distinguished from effects on receptor binding. The results suggest that the binding energy from interaction of IFN-beta with ifnar2 serves mainly to stabilize the bound IFN/receptor complex, whereas the binding energy generated by interaction of certain regions of IFN-beta with ifnar1 is not fully expressed in the observed affinity of binding but instead serves to selectively stabilize activated states of the receptor.

Identification of a linear epitope of interferon-alpha2b recognized by neutralizing monoclonal antibodies

Four monoclonal antibodies (mAbs) directed against the recombinant human interferon-alpha2b (IFN-alpha2b) were used as probes to study the interaction of the IFN molecule to its receptors. The [125I]IFN-alpha2b binding to immobilized mAbs was completely inhibited by IFN-alpha2b and IFN-alpha2a but neither IFNbeta nor IFNgamma showed any effect. Gel-filtration HPLC of the immune complexes formed by incubating [125I]IFN-alpha2b with paired mAbs revealed the lack of simultaneous binding of two different antibodies to the tracer, suggesting that all mAbs recognize the same IFN antigenic domain. Furthermore, the mAbs were also able to neutralize the IFN-alpha2b anti-viral and anti-proliferative activities as well as [125I]IFN-alpha2b binding to WISH cell-membranes. As [125I]mAbs did not recognize IFN exposed epitopes in the IFN:receptor complexes, mAb induction of a conformational change in the IFN binding domain impairing its binding to receptors was considered unlikely. In order to identify the IFN region recognized by mAbs, IFN-alpha2b was digested with different proteolytic enzymes. Immunoreactivity of the resulting peptides was examined by Western blot and their sequences were established by Edman degradation after blotting to poly(vinylidene difluoride) membranes. Data obtained indicated that the smallest immunoreactive region recognized by mAbs consisted of residues 107-132 or 107-146. As this zone includes the sequence 123-140, which has been involved in the binding to receptors, and our mAbs did not show an allosteric behaviour, it is concluded that they are directed to overlapping epitopes located close to or even included in the IFN binding domain.

Quantitative liquid-phase chemiluminescence ELISA: detection of subtle epitope differences in HuIFN-beta

We have developed a new liquid-phase, chemiluminescence-enhanced, inhibition ELISA (LP-CEI-ELISA) to explore the binding sites recognized by two neutralizing monoclonal antibodies (mAb) against recombinant human IFN-(beta)ser (rHuIFN-(beta)ser). In this assay, the initial antigen-antibody reaction occurs in solution under more physiologic conditions than in a standard solid-phase ELISA. Subsequently, the reaction mixture is applied to a membrane that is exposed to a second, peroxidase-labeled mAb, chemiluminescent reagents are added, and the membrane is photographically recorded. Competitive inhibition of binding of a second, labeled mAb by the first mAb decreases the signal detected. Two well-characterized mAb A1 and A7, have been shown to recognize distinct epitopes on rHuIFN-(beta)ser and to neutralize its antiviral and antiproliferative activity (Proc. Natl. Acad. Sci. USA 88, 4040-4044, 1991). In conventional solid-phase ELISA, mAb A1 does not inhibit the binding of A7 to rHuIFN-(beta)ser, but we observed partial inhibition in the new liquid-phase assay. In contrast, A7 did not inhibit the binding of A1, consistent with the solid-phase ELISA results. This observation suggests that in the LP-CEI-ELISA, A1 and A7 may recognize epitopes differently than in solid-phase assays. Thus, the LP-CEI-ELISA, which is simple, sensitive, and quantifiable, appears also to be able to detect subtle, conformational differences of epitopes not evident in a standard solid-phase ELISA.

The expression of potency of neutralizing antibodies for interferons and other cytokines

The occurrence of antibody formation in patients administered biologically active human proteins as biotherapy for different diseases emphasizes the importance of establishing a common method of reporting neutralizing antibody levels for such cytokines. For quantitative neutralization bioassays, the preferred expression of the neutralizing potency of an antiserum is a titer, that is, the dilution of serum that reduces 10 Laboratory Units (LU)/ml of the cytokine to 1 Laboratory Unit/ml, the endpoint of most bioassays. This 10-to-1 LU/ml expression, which has been recommended by the World Health Organization for recording the results of interferon neutralization by the constant interferon method (with varying dilutions of serum) can also be used with the constant antibody method (with varying concentrations of interferon). For various reasons, interferon doses in International Units (IU)/ml should not be used for the neutralization test. Should the interferon concentration vary, intentionally or otherwise, from the intended dose of 10 LU/ml, a simple calculation allows expression of the neutralizing potency as the recommended reduction of 10-to-1 LU/ml as follows: the titer to be reported is the reciprocal of the antibody dilution (achieving the endpoint), multiplied by the interferon concentration (measured in that day's titration) minus one, divided by 9. This index of neutralization is the preferred method to represent the neutralizing potency of polyclonal and monoclonal antibodies and should make the results from different laboratories more readily interpretable and enable comparison.

A homology model of human interferon alpha-2

An atomic coordinate five alpha-helix three-dimensional model is presented for human interferon alpha-2 (HuIFN alpha 2). The HuIFN alpha 2 structure was constructed from murine interferon beta (MuIFN beta) by homology modeling using the STEREO and IMPACT programs. The HuIFN alpha 2 model is consistent with its known biochemical and biophysical properties including epitope mapping. Lysine residues predicted to be buried in the model were primarily unreactive with succinimidyl-7-amino-4-methylcoumarin-3-acetic acid (AMCA-NHS), a lysine modification agent, as shown by mass spectrometric analysis of tryptic digests. N-terminal sequence analysis of polypeptides generated by limited digestion of HuIFN alpha 2 with endoproteinase Lys-C demonstrated rapid cleavage at K31, which is consistent with the presence of this residue in a loop in the proposed HuIFN alpha 2 model. Based on this model structure potential receptor binding sites are identified.

Antibodies that neutralize human beta interferon biologic activity recognize a linear epitope: analysis by synthetic peptide mapping

The location of biologically relevant epitopes on recombinant human beta interferon in which Ser-17 replaces Cys-17 (rh[Ser17]IFN-beta) was evaluated by testing the immunoreactivity of antibodies against 159 sequential, overlapping octamer peptides. Three monoclonal antibodies (mAbs) that neutralize rh[Ser17]IFN-beta biologic activity, designated A1, A5, and A7, bound to peptides spanning only residues 39-48, whereas nonneutralizing mAb bound less specifically at multiple sites near the amino terminus. The immunoreactivity of peptides spanning residues 40-47 that contained a series of single amino acid substitutions suggested that residues 41-43 (Pro-Glu-Glu) and 46 (Gln) are important for the binding of neutralizing mAbs. The reactivity of mAbs to larger synthetic peptides containing rh[Ser17]IFN-beta sequences from residue 32 through residue 56 was evaluated. All mAbs except A7 reacted with synthetic peptides representing rh[Ser17]IFN-beta residues 32-47, 40-56, and 32-56, but only mAbs A1 and A5 bound to the core peptide composed of residues 40-47. Peptide 32-56 effectively blocked the binding of mAbs A1 and A5 to rh[Ser17]IFN-beta and markedly inhibited their neutralizing activity. Biologic activity of the peptides was undetectable. Rabbit antisera raised against peptides 32-47 and 40-56 recognized rh[Ser17]IFN-beta but did not neutralize its antiviral activity. Thus, structure-function analysis by peptide mapping has permitted the identification of a linear epitope recognized by neutralizing antibody on a biologically active cytokine. We conclude that the region spanning residues 32-56 is of major importance in the expression of the biologic activity of human IFN-beta.