Over the last years increasing numbers of autochthonous cases, zoonotic spread and chronic infections has led to a better understanding of viral hepatitis E . The prevalence of anti-HEV in Central Europe may reach up to 15% . It is unclear whether this high prevalence is caused by a high number of undiagnosed cases of subclinical HEV infections or by a high false positive rate of unreliable serologic HEV antibody assays. Despite the increasing epidemiological and clinical relevance of the HEV and demanding research needs , available serological tests are limited to the detection of antibodies against so-called "traditional" antigens ORF2 and ORF3 .
In this study we describe a novel seroreactivity for ORF1 domains and extensively characterized it for their diagnostic potential in direct comparison to commercial available ORF2 and ORF3 antigens. The sequence of the potential ORF1 antigen was determined by the probably most native structure namely defined functional domains. Other studies of putative ORF1 epitopes are based on computer analysis of hydrophobicity and secondary structure [18, 20] or overlapping decamers . A common disadvantage of these artificially predicted epitopes is the presence of only linear epitopes and therefore conformational antigens might be missed . ORF1 epitopes found in earlier studies were within the regions of Met [18, 19], Plp , V , × [18, 19], Hel , Rdrp [18–20]. According to Zhao et al. the major antigenic epitopes of HEV are located in the ORF2 and ORF3. Four ORF1 antigens were excluded early in the study after initial experiments showed inadequate immunoreactivities of ORF1 antigens . Kaur et al. used two serum pools to screen for reactive ORF1 polypeptides but did not investigate the diagnostic impact with sensitivity and specificity levels of the distinct ORF1 proteins . Overall these data revealed that the diagnostic impact of ORF1 antigens is low. Only Qi et al. could show that the inclusion of an Rdrp epitope in a serologic assay increases the test performance for an unknown reason but did not focus on the performance of other ORF1 epitopes . In our study we also calculated if the additional presence of an ORF1 antigen could complement ORF2 and ORF3 antigens in a diagnostic test. It became clear that the two ORF1 antigens included in the diagnostic test according to backwards regression analysis (Y in IgG detection and V in IgM detection) could be exchanged by antigens already used in commercially test without significant change of the diagnostic performance.
For the first time we are able to describe immunogenic properties of the Y protein, a protein with homologies to non-structural proteins of the Rubella virus and beet necrotic yellow vein virus . This finding is in line with results from all other ORF1 regions and completes the list now consisting of all ORF1 domains exhibiting a moderate immunogenicity. This is easily explained by the fact that all parts of the ORF1 polyprotein are expressed in the same amount in the cytoplasm of infected cells, processed and presented to the immune system equally. Nevertheless, Y shows in our study the highest IgG and IgM sensitivities among all investigated ORF1 domains.
Within the X domain we measured higher sensitivities and specificities compared to a peptide from this region detected by Zhao et al.  who investigated its seroreactivity exclusively in swine. In our study values from the X-protein in human HEV diagnostics even showed a better performance than commercial O2MGT1 in detection of IgG and IgM. Interestingly, it was shown, that X-protein impacts viral pathogenicity of a murine Corona Virus (MHV A59) [27, 28].
In contrast to the predictions of Zhao et al.  and Kaur et al.  we could not show any antigenic properties within the methyltransferase domain. Nevertheless, this does not necessarily exclude the possibility of a diagnostic value of ORF1 antigens under different conditions like for example investigations in endemic regions with a larger patient population. By changing the methodical approach of protein expression, further studies should investigate the seroreactivity of the remaining ORF1 domains not expressed in this study as for example the polymerase domain, which comprises several linear epitopes.
Logistic regression as well as ROC analysis confirmed that the ORF2 C-terminus has the highest diagnostic potential of all until now investigated HEV antigens. It is well known that immunogenicity of ORF3 is also determined by the C-terminus . Our data confirm that the N-terminus of ORF3 is not suitable for detection of IgG and IgM antibodies. In contrast the reactivity of the C-terminus correlates very well with the full length ORF3 protein and showed an even better AUC in ROC analysis. A differentiated view on N- and C-terminal parts of ORF3 seems to play an important role not only in serologic diagnostics but also in the newly postulated involvement of ORF3 in viral egress [17, 27, 30]. For ORF3 the prediction of a transmembrane domain was included in our study to optimize mimicry of naturally occurring antigen conformations and probably explains the predominant seroreactivity of the C-terminal subdomain. Of course this hypothetic secondary structure needs to be confirmed in additional molecular studies. In general the finding of different seroreactivities of morphologic ORF3 subdomains supports the approach to investigate functional ORF1 subdomains separately.
In this study we followed our method for serum profiling experiments  achieving an even higher expression level of 60% and a purification efficiency of 100%. As HEV genome size is restricted to about 7,200 nt the complete sequence could be synthesized in vitro thus avoiding the necessity of handling infectious virions during test development and pre-selection of specific viral antigens. In addition the recomDOT system applied in this study represents a different and independent assay-format as compared to the already commercially available LIA and ELISA, where the "Mikrogen antigens" are already in use.
Our statistical analysis of genotype 1 ORF2 (O2C vs. O2CGT1) and ORF3 (O3 vs. O3GT1) seroreactivities (IgG and IgM) showed, that respective proteins reacted in the same way with patient serum unrelated to the production method. Correlation coefficients were more than 0.9 between homologous proteins and specificity as well as sensitivity levels appeared to be almost equal. Therefore the described pipeline was proven to be an adequate method to produce proteins of proper quality and characteristics for serum profiling experiments. Furthermore results indicate, that existing commercial HEV assays seem to measure HEV prevalence properly .
Germany is a country of low HEV endemicity. Studies could show that genotype 1 infections are commonly associated with traveling in endemic countries and autochthonous genotype 3 infections probably acquired from domestic livestock . As PCR results frequently remain negative and thus no sequence data are obtained genotype during HEV infections often remains unclear. "Serologic genotyping" would easily allow to answer epidemiologic questions when the source of infections is unknown or to estimate the potentially genotype dependent outcome of HEV infections during pregnancy . Comparison of measured signal intensities of genotype 1 and genotype 3 antigens with defined patient sera indicate that there might be a genotype specific reactivity of ORF2 antigens. Further studies with a larger population need to be performed to confirm this finding. Eventually specific sequences of ORF1 also hold the potential for "serologic genotyping" and therefore should be investigated addressing this question.