Influence of the washing buffer composition on the sensitivity of an enzyme-linked immunosorbent assay using mycobacterial glycolipids as capture antigens.

Immunogenic glycolipids from the cell wall of Mycobacterium tuberculosis are potential capture antigens in enzyme-linked immunosorbent assays (ELISAs) for the serodiagnostis of tuberculosis. Typically, washing steps in ELISAs are performed with buffers containing a detergent. However, Tween-20, the most commonly added detergent, was reported to be able to remove the coating of certain glycolipid antigens from microtitre wells. In order to determine the influence of the washing buffer composition on the results, we measured serum immunoglobulin G (IgG) against three mycobacterial glycolipids by ELISA, conducting three separate experiments with three different buffers: Tris-buffered saline (TBS), TBS plus 0.02% Tween-20 (TBS-Tween), or TBS plus 0.3% bovine serum albumin (TBS-BSA). The capture antigens applied were lipoarabinomannan with the basic arabinose-containing motif (AraLAM), the mannose-capped version of lipoarabinomannan (ManLAM), and trehalose-6,6'-dimycolate (cord factor). All ELISAs achieved acceptable specificities around 95%. The sensitivities, however, varied widely, depending upon the sort of washing buffer used. In 38 patients with sputum smear-positive pulmonary tuberculosis and control groups of 79 patients with non-tuberculosis lung disease and 92 healthy volunteers, the anti-cord factor ELISA achieved 100%, 31.6%, and 60.5% with TBS, TBS-Tween, and TBS-BSA, respectively. Corresponding sensitivity values for AraLAM were 39.5%, 26.3%, and 23.7%, and for ManLAM 94.7%, 65.8%, and 55.3%. We conclude that Tween-20 or BSA should be omitted from the washing buffer in ELISAs, when the capture antigen is of lipid nature.