ra recognized the propeptide and 11 sera recognized the light chain. These were inconsistent with Suzuki’s study,, which reported the negative binding to the light chain. However, our MedChemExpress CSP-1103 results were consistent with the Bruner’s study, which demonstrated that the major epitope sites of MPO-ANCA existed on both the propeptide and the light chain. In Bruner’s study, the antigenic sequences identified on light chain included amino acids 213222. In a recent study with a large sample size by Suzuki et al, the light chain was recombined in two 18362028 parts divided by amino acid 214 and 215, which were among the aa 213222 in Bruner’s study. The epitope sites on light chain might be just missed. In an early study by Suzuki’s group, the light chain was recombined as a whole, but only 4 patients were analyzed in the study. In the present study, the light chain was recombined as a whole, and the results suggested that the reactivity to light chain was not rare in Chinese patients with AAV. In addition, the binding to light chain of MPO-ANCA was associated with more severe renal dysfunction and systemic disease activity in AAV patients. Based on the threedimensional model of MPO, the light chain is largely `hidden’ in the groove between the two MPO monomers and is poorly `accessible’ to antibody binding,. Once the light chain was recognized, it might suggest that the epitope spreading was much more extensive, which may explain the severity in patients with MPO-ANCA reactivity to the light chain. In addition, patients with antibodies to the light chain reacted to higher number of the recombinant fragments than those without, which suggests an intra-molecular epitope spreading might happen. In the current study, the epitope sites recognized by MPOANCA from AAV patients without serum anti-GBM antibodies displayed a more diverse repertoire, while the epitope sites recognized by MPO-ANCA from AAV patients with coexisting serum anti-GBM antibodies were much less diverse and mainly focused on the N-terminus of the heavy chain. The result indicated that the epitope specificities of MPO-ANCA were different between these two disease phenotypes, i.e. AAV patients with and without anti- GBM antibodies. It was also found in the current study that the recognition frequency to H1 fragment tends to be higher in AAV patients with anti-GBM antibodies than those without anti-GBM antibodies. In addition, the similarity of amino acid sequences 107112 of a3NC1 and amino acid sequences 373378 of MPO in H1 fragment raised the question that the H1 fragment might play a role in the development of the two antibodies, which still needs further investigation. Whether the same epitopes are targeted during initial onset and relapse of AAV is an important issue and is not fully clear yet. In the present study, we tested sequential serum samples in 10 patients with AAV, at different stages of the disease, i.e., initial onset, remission and relapse. To the best of our knowledge, this is the first study to map the possible 
changing or epitope spreading of MPO-ANCA in AAV patients during the relapse of the disease. It was found that in the sera from relapse which recognized linear fragments, the reactivity to linear fragments was similar to that of initial onset. This may suggest a role of “immunological memory”for inducing relapse. In 22286128 addition, although there was no fixed antibody-binding changing pattern in the three different stages of AAV, the number of recognized epitope sites tends to de