Clinical associations and potential novel antigenic targets of autoantibodies directed against rods and rings in chronic hepatitis C infection

An estimated 180 million people worldwide are infected with hepatitis C virus (HCV), a known major cause of chronic liver disease [1, 2]. HCV infection is also associated with several immunological abnormalities, including the production of both organ specific and non-organ specific autoantibodies [3–5]. Organ specific autoantibodies include those directed against targets in pancreatic islet cells [6], thyroid [7–9], adrenal cortex [6] and gastric parietal cells [10]. Non-organ specific autoantibodies include anti-nuclear antibodies (ANA), anti-smooth muscle antibodies (ASMA), anti-mitochondrial antibodies (AMA), anti-neutrophil cytoplasmic antibodies (ANCA), and anti-liver/kidney microsomal antibodies (LKM) [11–14]. Although their clinical significance remains unclear, ANA have been reported in 4% to 63% of patients with chronic hepatitis C [11, 15–18]. Some studies have shown that ANA positivity is associated with stage and rate of fibrosis progression, serum transaminase concentrations and responsiveness to antiviral treatment [19–23]. Other reports have found no differences in these and other clinical parameters [17, 24–28]. Interferon (IFN) and ribavirin, cornerstones of the management of HCV infection, have immunomodulatory effects [29, 30] such as the production of autoantibodies [31].

In patients with chronic hepatitis C infection, a novel cytoplasmic autoantibody pattern (RR) characterized by rods (~3-10 μm in length) and rings (2–5 μm diameter) has been described on HEp-2 cells [32–34]. Inosine monophosphate dehydrogenase 2 (IMPDH2) and cytidine triphosphate synthase 1 (CTPS1) were identified as potential autoantibody targets localized to the RR structures [32, 33]. The objectives of this study were to determine the prevalence and clinical associations of RR autoantibodies in chronic hepatitis C patients, to examine the frequency of antibodies to IMPDH2 and CTPS1 and to identify other potential autoantibody targets by screening high density peptide and protein arrays.

HCV infection is associated with a wide spectrum of immune reactions, some of which are reflected by the presence of organ and non-organ specific autoantibodies. Possible mechanisms for the production of autoantibodies include molecular mimicry, an interaction of the HCV with B lymphocytes promoting B cell proliferation and activation, or direct infection of immunocytes by HCV [48–52]. In addition, patients receiving interferon and/or ribavirin therapy may have accelerated pre-existing autoimmune diseases, or de novo occurrence of autoimmune disorders or autoantibody production [26, 31, 53–58].

A novel autoantibody staining pattern has recently been reported in patients with HCV infection characterized by rods (~3-10 μm in length) and rings (2–5 μm diameter) localized to the cytoplasm of certain cell lines and expresed throughout the cell cycle [32–34]. Other studies have determined that this IIF pattern is associated with antibodies directed against IMPDH2 or CTPS1 [32, 33, 59]. In our study we confirmed that IMPDH2 reacts with a minority of HCV sera, a finding in keeping with reports by others [33, 59]. Although CTSP1 was localized to RR [33], it does not appear to be a primary target of human autoantibodies as none of our sera in this study or human sera in a previous study [33] reacted with the purified CTSP1 protein.

While the frequency of the reactivity to IMPDH2 in the present study is less than previously reported [32, 33, 59], it is clear from studies to date that other autoantibody targets remain to be identified. To address this possibility, we probed a commercially available protein and peptide microarray and identified a number of unique potential autoantibody targets (Table 2), where the Myc-associated zinc finger protein (MAZI) is of particular interest [39]. There is evidence that MAZI, which contains six C2H2-type zinc fingers, functions as a transcription factor with dual roles in transcription initiation and termination [40]. While the cellular localization has not been definitively determined, it is presumed to be primarily localized to the nucleus, although in brains of Alzheimer disease patients it is localized to plaque-like structures in the cytoplasm [60]. Of note, MAZI is expressed in kidney, liver and brain and it is a purine binding transcription factor. The latter feature is of particular interest because of its potential relation to inosine metabolism and IMPDH2 previously identified RR autoantibody targets [32, 33, 59].

The actin-related protein Arp1 (or centractin) is the major subunit of dynactin, a key component of the cytoplasmic dynein molecular motor [46]. Under certain conditions Arp1 has high homology to conventional actin, which has been shown to polymerize [46]. Arp1 is also predicted to bind ATP and another autoantibody target, the nuclear mitotic apparatus protein (NuMA) [61]. Likewise, the ankyrin repeat motif (ARM) identified as part of the sterile alpha motif domain containing 6 (ANKS6) protein is of interest. ARMs are typically comprised of 33 residues and are structurally represented as two alpha helices separated by loops [44, 45]. ARM is also one of the most common protein–protein interactions that mediate protein-protein interactions and several unique aspects of protein folding [44, 45]. Ankyrin repeats appear in virtually all organisms but are most abundant in eukaryotic cells where they are found in 6% of proteins of diverse function such as transcriptional initiators, cell cycle regulators, cytoskeleton, ion transporters, and signal transducers. The voltage-dependent anion channel 1 (VDAC1) localized to the outer mitochondrial membrane has been shown to control metabolic interactions between mitochondria and the rest of the cell [41]. VDAC1 has been implicated in the control of apoptosis, including via its interaction with the pro- and anti-apoptotic proteins [41, 42] and due to an abnormal interaction with amyloid beta and phosphorylated tau, is implicated in mitochondrial dysfunction in Alzheimer’s disease [43]. VDAC1 also contributes to the metabolic phenotype of cancer cells as reflected by its over-expression in many cancer types [41]. Whereas these candidate target autoantigens have common structural and functional properties (i.e. purine metabolism and protein folding, aggregation and polymerization), additional studies are needed to establish immunoassays and determine the prevalence of antibodies to these novel targets identified in our study of index RR sera.

In our study of various cell lines, commercially available HEp-2 substrates from INOVA Diagnostics and CHO cells maintained in our own tissue culture facility constitutively expressed RR. Others have also reported that the HEp-2 substrate provided by INOVA seems to be unique in demonstrating “out of the box” RR staining [34]. Of interest, a HEp-2 cell line obtained from Dr. Edward K.L. Chan, one of the first to report the RR pattern, produced the RR pattern but after 2 to 3 passages of these cells, RR became undetectable. This suggests that under certain growth or tissue culture conditions, RR expression can either be diminished or facilitated. However, when a variety of tissue culture protocols including various antibiotics, media and heat shock were used, we were unable to demonstrate RR formation. Since tissue culture and fixation protocols are considered a trade secret by manufacturers of HEp-2 diagnostic substrates, the reasons for this finding are not fully understood. As reported by others [32, 34], we confirmed that the RR pattern is restored after cells are treated with ribavirin. Of interest, ribavirin is likely not an obligatory reagent in all cells since we are the first to report that CHO cells, constitutively and without adding exogenous ribavirin, express similar immunoreactive RR structures.

In the present study, the 5% frequency of RR autoantibodies in our HCV cohort is lower than the 20-35% prevalence previously described in HCV sera [32, 34]. The finding of antibodies to RR appears to be relatively specific because we did not identify any RR autoantibodies in PBC or SLE sera. Retrospective chart review did not find an association between RR autoantibodies with clinical characteristics including age, gender, mode of HCV infection, prevalence of HCV genotype 1, serum ALT concentration, platelet count, severity of necroinflammation and fibrosis or the presence of either ANA or AMA. However, both uni- and multi-variate analysis showed that prior HCV IFN and ribavirin treatment was the only independent predictor of RR antibody positivity. Since none of the patients treated with IFN monotherapy had RR autoantibodies, RR autoantibodies seems to only be present in HCV patients treated with combination IFN and ribavirin therapy. This is supported by studies reported here and by others showing that RR are induced after treatment of cell lines with ribavirin, but not IFN [33]. These results also confirm previous reports that RR autoantibodies were significantly associated with prior IFN/ribavirin treatment [34]. Other studies have shown that anti-RR antibodies were not present at disease baseline, but appeared during IFN/ribavirin therapy and were more often detected in non-responder/relapsers than in responder patients [59]. Nevertheless, it is clear the antibodies to RR are seen in patient sera that have no obvious HCV infections or treatment with ribavirin or IFN [32], suggesting that other mediators are likely involved in induction of the B cell anti-RR response. As all of our patients treated for HCV achieved an SVR, we were unable to assess this in our study but is amenable to more effective analysis in multi-center studies of larger cohorts because it may be a biomarker for poor response to therapy. In addition, the effect of ‘triple therapy’ on the induction of RR autoantibodies, has yet to be investigated.

Limitations to our study include the small sample size of HCV treated patients and the majority of our cohort were not treated with IFN/ribavirin at the time of sera collection (convenience sample as described above). However, this may be an advantage in showing the importance of IFN/ribavirin as a triggering or modulating factor in the induction of these novel autoantibodies as it does support previous studies, that RR are primarily seen in IFN/ribavirin treated patients thus explaining our relatively low frequency of RR as compared to previous reports. The fact that all treated patients achieved an SVR, limited our ability to fully identify the relationship of RR autoantibodies to treatment outcomes. In addition, this is a retrospective study without longitudinal sera samples (i.e. pre, during, and post therapy).

In conclusion, a novel cytoplasmic autoantibody staining pattern, rods and rings (RR) has recently been reported in patients with HCV infection [32, 34]. We identified this pattern in 5% of our HCV cohort and found that prior IFN and ribavirin treatment was significantly associated with RR autoantibody positivity. Only a minority of our RR sera reacted with IMPDH2 and none reacted with CSP1, suggesting more effort is required in identifying the related target autoantigens. Further investigations are warranted to further determine the clinical, pathogenic, and prognostic significance of autoantibodies directed against RR.