The strong link between ankylosing spondylitis (AS) and HLA-B27 has been well established. Any aetiological agent or mechanism implicated in AS must provide an explanation for the link with HLA-B27. An amino acid sequence homology, QTDRED, found in the variable region of B*2705, (residues 72–77) and the KP2 component of Klebsiella pneumoniae nitrogenase enzyme (residues 188–193) has been reported. Furthermore, AS patients were shown to have increased concentrations of antibodies to a homologous sequence of both B*2705 3 and the KP2 component ofK pneumoniae nitrogenase reductase although some workers have been unable to confirm these results. Antibody aYnity is often lower with peptide sequences compared with binding by the native protein and this could be because of conformational changes between peptide and the native protein molecule, thereby accounting for these diVerences in reactivity. In the light of these conflicting findings, this study was undertaken to measure antibodies to the native KP2 component of nitrogenase reductase enzyme of K pneumoniae, to determine if it has a part to play in the development of disease, and whether antinitrogenase antibodies are correlated with C reactive protein (CRP) concentrations. Serum samples were obtained from AS patients attending the AS Research Clinic at the Middlesex Hospital, and rheumatoid arthritis (RA) patients attending the Rheumatology Department at the Lister Hospital, Stevenage. Healthy control samples were supplied by the Blood Transfusion Service, London. The diagnosis of AS was according to the New York criteria and that of RA by the American Rheumatism Association. In the study, IgG, IgA, and IgM immunoglobulin antibodies were measured in 200 subjects against K pneumoniae nitrogenase reductase. The groups examined were as follows: 100 patients with AS (71 male, 29 female, mean age 47 years); 50 RA patients (15 male, 35 female, mean age 57 years), and 50 healthy control subjects (30 male, 20 female, mean age 44 years). The KP2 component of Klebsiella nitrogenase,which contains the QTDRED sequence, was provided by Dr Martin Buck, of the Nitrogen Fixation Laboratory in Sussex University and purified as previously described. Antibody levels against the KP2 component of K pneumoniae nitrogenase reductase were measured by enzyme linked immunosorbent assay (ELISA) as previously described. All assays were carried out under code, so that the status of each serum sample under investigation was not known to the tester. CRP concentrations were determined by the single radial immunodiVusion method of Mancini et al and the results expressed as mg/l of serum. The mean optical density (OD) units of IgG, IgA, and IgM immunoglobulin antibodies against the KP2 component of K pneumoniae nitrogenase reductase in diVerent groups were compared with Student’s t test. The association between the anti-KP2 component antibodies and CRP was analysed using Pearson’s correlation analysis. We found that the AS patients had significantly increased antibody concentrations of the IgG (mean (SEM)) (0.32 (0.01), OD units) (t = 6.12; p<0.001) and IgA class (0.42 (0.02), OD units) (t = 8.84; p<0.001) to KP2 component of K pneumoniae nitrogenase reductase compared with the healthy control group. In addition, AS patients had significantly increased antibody concentrations of the IgG (0.32 (0.01), OD units) (t = 4.33; p<0.001) and IgA class (0.42 (0.02), OD units) (t = 7.97; p<0.001) to the KP2 component of K pneumoniae nitrogenase reductase compared with the RA patients. However, there was no significant increase in IgM antibody concentrations against the nitrogenase reductase protein (table 1). The RA patients showed no significant antibody increases in IgG, IgA or IgM classes with respect to control subjects. The mean CRP concentrations were 21.6 mg/l (SEM) (2.4) in the AS patients, 39.1 mg/l (3.9) in the RA group, and 6.9 mg/l (1.7) in the healthy control group. There was a significant correlation between IgG (r = +0.385, p<0.001) (fig 1A) and IgA (r = +0.442, p<0.001) (fig 1B) antibody concentrations against the KP2 component of Klebsiella nitrogenase reductase and CRP values respectively. However, there was no correlation between IgG and IgA antibody concentrations against the KP2 component of K pneumoniae nitrogenase reductase and CRP concentrations in the RA patients tested. These findings suggest that AS patients may have been exposed to the KP2 component of K pneumoniae nitrogenase reductase protein, which contains an amino acid sequence homology with HLA-B27. Further studies are required to show that antibodies against the K pneumoniae homologous peptide, which are present in AS patients, are cytotoxic for HLA-B27 positive but not HLA-B27 negative cells, and assess whether anti-Klebsiella treatment, can reduce the severity of relapses and modify the progression of the disease.