Methods used for the detection of BAb to cytokines include cytokine bioassays, immunometric assays and various blotting techniques. Serum BAb, however, do not always bind soluble polypeptides in a specific manner or with any appreciable affinity (1, 2). Thus, heterophilic BAb may give false positive results in ELISA consisting of two or more layers of Ab (3), and immunoblotting techniques and immunometric assays may show some degree of specificity, even though the binding of BAb to ligand is weak and topographically unassociated with the specific binding sites of the aAb; see figure below. For example, we and others have found that Western blotting and detection in direct ELISA are not suited for detection of IL-1a BAb because of strong, but non-specific immunoglobulin binding to IL-1a fixed to solid phases (4, 5).

Without previous knowledge of the binding characteristics, competition in these assays by surplus cytokine in solution may also not discriminate between the native conformation, different degrees of denaturation of the cytokines, or other factors as possible cause of the competitive activity.

Although ELISA techniques can be used for screening purposes, verification of specific cytokine BAb would require ligand binding to the Fab fragments of the BAb, using properly validated cytokine tracers, combined with saturation binding analyses and, at least in the case of natural aAb, demonstration of cross-binding to the endogenous cytokines (6).

Another potential problem is that assays for cytokine BAb often make use of unglycosylated, E.coli derived cytokines. Since many native cytokines are glycosylated, the recombinant forms may possess other or concealed antigenic sites that could jeopardise the interpretation of the assays; see figure below. Furthermore, even though glycosylated forms of some recombinant cytokines are available, their carbohydrate moieties may differ significantly from those of the wildtype cytokines. It is therefore possible that pre-existing BAb to certain polysaccharide structures may mimic a cytokine BAb by binding in vitro to a glycosylated recombinant cytokine (7, 8). These BAb may bind glycosylated cytokines both in direct binding assays, in solution and in solid-phase assays. Whenever binding to Ab (or aAb) has been detected with a recombinant cytokine, demonstration of cross-binding with the natural cytokine is always relevant (2, 9-11).

Example:

Only cases 1, 2 and 3 detect specific Ab to the native cytokine (case 3 only to aggregated forms). These cases should be discriminated from cases 4, 5 and 6 by competition experiments using excess (non-denatured) cytokine in solution. However, while Ab-binding is displaceable in cases 1 and 2 with the natural unaggregated cytokine, a non-glycosylated, non-denatured and unaggregated recombinant cytokine would displace binding only in case 1. In practice, the native cytokine used to capture Ab in case 2 will rarely be available, and a carbohydrate-specificity of an Ab will therefore go unnoticed. The binding in case 5 is displaceable only by the glycosylated recombinant cytokine and, possibly, by relevant saccharides. In case 6, it will be difficult or impossible to displace the binding with soluble cytokine. These 'Ab' are nonspecific and usually of low affinity; they may constitute a significant part of the so-called non-neutralising BAb in IFN-beta-treated MS patients; see Anti-interferon ab in Multiple Sclerosis.

The above principles also apply to the detection of Ab to soluble peptides by many other methods, particularly those involving solid-phase binding techniques.

Abbr.: CK: cytokine, E: enzyme.

Cited references:

1. Hansen MB, Svenson M, Bendtzen K. Serum-induced suppression of interferon (IFN) activity. Lack of evidence for the presence of specific autoantibodies to IFN-alpha in normal human sera. Clin Exp Immunol 1992; 88:559-62.

2. Svenson M, Hansen MB, Bendtzen K. Binding of cytokines to pharmaceutically prepared human immunoglobulin. J Clin Invest 1993; 92:2533-9.

3. Grassi J, Roberge CJ, Frobert Y, Pradelles P, Poubelle PE. Determination of IL1a, IL1b and IL2 in biological media using specific enzyme immunometric assays. Immunol Rev 1991; 119:125-45.

4. Suzuki H, Kamimura J, Ayabe T, Kashiwagi H. Demonstration of neutralising autoantibodies against IL-1a in sera from patients with rheumatoid arthritis. J Immunol 1990; 145:2140-6.

5. Suzuki H, Ayabe T, Kamimura J, Kashiwagi H. Anti-IL-1a autoantibodies in patients with rheumatic diseases and in healthy subjects. Clin Exp Immunol 1991; 85:407-12.

6. Svenson M, Herbrink P. Measurement of cytokine antibodies. Test development. Biotherapy 1997;10:87-92.

7. Sedmak JJ, Grossberg SE. High levels of circulating neutralising antibody in normal animals to recombinant mouse interferon-b produced in yeast. J Interferon Res 1989; 9, Suppl. 1:S61-S5.

8. Hansen MB, Svenson M, Diamant M, Ross C, Bendtzen K. Interleukin-6 (IL-6) autoantibodies and blood IL-6 measurements. Blood 1995; 85:1145.

9. Svenson M, Hansen MB, Bendtzen K. Distribution and characterization of autoantibodies to interleukin 1a in normal human sera. Scand J Immunol 1990; 32:695-701.

10. Hansen MB, Svenson M, Diamant M, Bendtzen K. High-affinity IgG autoantibodies to IL-6 in sera of normal individuals are competitive inhibitors of IL-6 in vitro. Cytokine 1993; 5:72-80.

11. Ross C, Svenson M, Hansen MB, Vejlsgaard GL, Bendtzen K. High avidity IFN-neutralising antibodies in pharmaceutically prepared human IgG. J Clin Invest 1995; 95:1974-8.