E. coli and the Etiology of Human PBC: Anti-mitochondrial Antibodies and Spreading Determinants.

The increased frequency of urinary tract infections in PBC patients and the cross-reactivity between the lipoyl domains (LD) of human pyruvate dehydrogenase complex (hPDC-E2) and Escherichia coli PDC-E2 (ePDC-E2) have long suggested a role of E. coli in causality of PBC. This issue however has remained speculative. We hypothesized that by generating specific constructs of human and E. coli PDC-E2, we would be able to assess the specificity of autoantibody responses and define whether exposure to E. coli in susceptible hosts is the basis for the AMA response. Importantly, the reactivity of hPDC-E2LD affinity purified antibodies against hPDC-E2LD could only be removed by prior absorption with hPDC-E2LD and not ePDC-E2, suggesting the presence of unique human PDC-E2 epitopes distinct from E. coli PDC-E2. To identify the autoepitope(s) present in hPDC-E2LD, a more detailed study using a variety of PDC-E2 constructs were tested including the effect of lipoic acid (LA) on ePDC-E2 conformation and AMA recognition. Individual recombinant ePDCE2 LD domains LD1, LD2 nor LD3 did not react with either AMA or antibodies to LA (anti-LA) but in contrast, anti-LA was readily reactive against purified recombinant LD1, LD2 and LD3 expressed in tandem (LP) ; such reactivity increased when LP was pre-cultured with LA. Moreover, when the 3 LD (LD1, LD2, LD3) domains were expressed in tandem in pET28a, or when LD1 was expressed in another plasmid pGEX, they were lipoylated and reactive to PBC sera. In conclusion, our data are consistent with an exposure to E. coli that elicits specific antibody to ePDC-E2 resulting in determinant spreading and the classic autoantibody to hPDC-E2LD. We argue this is the first step to development of human PBC.