Identification of linear human B-cell epitopes of tick-borne encephalitis virus

Tick-borne encephalitis virus (TBEV) is one of the most important neurotropic arthropod-borne pathogens in Europe, causing nearly 3000 hospitalizations annually [1]. TBEV is endemic in many European countries, Russia and China. It is mainly transmitted by Ixodes ticks, but can also be transmitted via consumption of unpasteurized goat milk [2, 3]. There are three subtypes of TBEV circulating in geographically distinct areas; namely European, Siberian, and Far Eastern. The variation at amino acid level is up to 5.6% between subtypes and 2.2% within a subtype [4].

The flavivirus virion is approximately 50 nm in diameter and displays icosahedral symmetry. The virion consists of three structural proteins (capsid C, prM and envelope E), the RNA genome, and a lipid membrane derived from the host cell. The flavivirus genome is a single-stranded positive-sense RNA (approximately 11 kilobases), which is encapsidated by the C protein. The genome contains a single open reading frame, which encodes a polyprotein that is co- and post-translationally cleaved into ten proteins by viral and host proteases. The envelope protein E is a class II viral fusion protein. It consists of three distinct domains (I, II and III), and forms homodimers in a head to tail manner. In the virion, the homodimers further arrange into trimers parallel to each other. The other envelope protein, prM, is cleaved by furin during viral maturation and the pr moiety is released as a result of conformational changes. The seven nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5) are found in the infected cell. NS1 is the only nonstructural protein that is glycosylated and secreted outside the cell. NS3 and NS2B form the viral serine protease that is required for post-translational modification of the polyprotein [5]. NS5 is a multifunctional protein containing an N-terminal methyl transferase domain and a C-terminal RNA-dependent RNA polymerase domain [6, 7]. The other small nonstructural proteins (NS2A, NS4A and NS4B) are expected to function at least in the genome replication [8].

To date, the E and NS1 proteins are known to raise protective antibodies in infected humans, monkeys and mice [9]. PrM, does not elicit protective antibodies, but is probably required for the preservation of conformational epitopes of the E protein [10]. Previous studies suggest that infections with dengue (DENV), Japanese encephalitis (JEV) and West Nile (WNV) viruses can be differentiated by the antibody response to the prM protein [10, 11]. For DENV type 1, some of the dominant epitopes in E and NS1 proteins have been identified using protein fragmentation methods [9]. In addition, AnandaRao et al. characterized several immunodominat linear B-cell epitopes in C and NS4A proteins of DENV using multi-pin peptide synthesis strategy [12].

In the present study, we used a peptide-based approach to identify immunodominat linear B-cell epitopes from the entire TBEV genome, which have not been previously reported. We found TBEV-specific peptides in the E and NS5 proteins. The characterized epitopes showed potential in differentiating between other flavivirus infections, and between natural and vaccine-derived immunity to TBEV.

Currently, TBE diagnostics is based on assays measuring IgM and IgG antibody responses to purified viral particles or consisting of structural proteins, or in some cases recombinant virus-like particles [15, 16]. TBEV E-protein is highly cross-reactive among other members of the genus Flavivirus and thus the identification of TBEV-specific antigenic regions would be beneficial. The cross-reactivity between flaviviruses is mainly due to conformational epitopes, and therefore linear epitopes presented as peptides could represent a useful alternative for traditionally used purified virus or recombinant protein diagnostic antigens. Ideally, such antigens could enable differential diagnostics between flaviviruses, and perhaps also between natural TBEV infection and vaccine responses. On the other hand, the potential of the nonstructural proteins as antigens is currently not fully explored. Our study provides one holistic approach for screening antigenic areas with diagnostic potential. For mosquito-borne flaviviruses, Wong et al. have developed an NS5-based serological test for differentiating natural WNV-infection from other flavivirus infections (specifically Saint Louis encephalitis and dengue virus) [17].

In this study, we describe the mapping of linear B-cell IgG epitopes in the TBEV proteome. Initial mapping using TBEV-seronegative and acute TBEV-seropositive serum pools revealed several regions containing linear epitopes throughout the TBEV proteome. Further analysis of the epitope regions using individual serum samples revealed that most of the epitopes were not recognized by all of the positive samples. After data analysis, we selected 11 peptides in total from E, NS1, NS2B, NS3, NS4B and NS5 proteins for more thorough characterization. In addition, we addressed the cross-reactivity of TBEV seropositive sera against the corresponding regions from other flaviviruses. Eventually, we identified three TBEV-specific epitope regions that reacted with the majority of TBEV seropositive sera. Two of the identified peptides are in the E protein and one in the NS5 protein region. Of these, peptide E-85/59 showed the highest specificity and potential in serological differentiation between natural and vaccine-derived immunity. Holzmann et al. used a similar type of approach to study linear epitopes in TBEV E protein utilizing mouse monoclonal antibodies. Their results demonstrated that only few of the MAbs reacted with synthetic peptides from TBEV E protein. Curiously, these MAbs bound only TBEV E protein only if directly coated on microtitre plate or otherwise denatured [18]. However, in our study we were able to demonstrate several linear epitopes throughout TBEV E protein. Peptide E-85/59, the region identified by Holzmann et al. as an epitope for TBEV MAb, was recognized only by sera of infected individuals and not by sera of vaccinated individuals.The space-filling model of the E protein (Figure 2) demonstrates that the identified peptides lie on the surface of the folded protein. Furthermore, both E-74/38 and E-85/59 that are located in domains I and II, respectively, actually form or are parts of a conformational epitope (Figure 2C and D). Neither of these peptides was reactive with sera of TBEV-immunized individuals, thus suggesting that the conformational epitopes at these regions are affected by the fixation used for virus inactivation. Furthermore, we observed that the sera of TBEV-immunized individuals recognized several epitopes in the E protein that were less reactive with sera of naturally infected TBEV-seropositive individuals. This probably highlights the fact that the natural B-cell response towards TBEV, and perhaps towards flaviviruses in general, is based on conformational epitopes. Although vaccination using formalin-fixed inactivated virions is effective, stronger responses would obviously be obtained with replication in cells or with native proteins.

We generated a homology model for TBEV NS5 to visualize the epitopes found in this region. The peptides NS5-521/198, NS5-531/223 and NS5-532/250 identified in this study, are located in the palm domain in the TBEV NS5 structure model. Of these peptides, NS5-521/198 is identical to the respective region in Omsk hemorrhagic virus NS5. The parent peptide reacted (positive with 6/6 of TBE patient and 3/20 of seronegative samples) rather similarly with the OMSKV peptide (positive with 4/6 of TBE patient and 3/20 of seronegative samples). Thus it seems evident that OMSKV-seropositive serum would also recognize this peptide epitope. However, since a couple of epitopes are located in the palm domain of TBEV NS5, one could envision using this region as recombinant antigen for flavivirus diagnostics.

Throughout our study we used densitometry to quantify the signal intensities of the individual peptides on the SPOT array. However, since the detection of SPOT binding is based on an enzymatic reaction, many factors contribute to the signal intensity (incubation and exposure times, membrane regeneration, antibody dilutions, ECL reagents, etc.). Quantification can therefore be done only inside the array and not between individual serum samples. Due to this, we evaluated each membrane individually, and considered peptides positive if the intensity was above a certain assay-specific threshold value. However, this method brings out a challenge in selecting the true-positive peptides, and vice versa, it is likely to miss some potential spots under the arbitrary threshold. As exemplified by a slight discrepancy in the results obtained with DENV2 (3/6 were positive) and DENV3 (4/6 were positive) peptides having identical sequences. This discrepancy is directly explained by the selected threshold values and the factors described above. Additionally, the regeneration steps required for re-use of the SPOT peptide array membranes are not ideal for high throughput screening. Thus we will, in the near future, synthesize some of the promising peptides found in this study in soluble form, and use them as antigens in an enzyme-linked immunosorbent assay (ELISA) format. ELISA will allow us to further test these peptides with larger serum panels and at different dilutions.

In conclusion, we have for the first time characterized the linear human B-cell epitopes of TBEV using TBE-patient and cross-reacting flavivirus sera. The TBE patients’ sera detect peptide targets particularly on the surface of E and NS5 proteins, and two of these peptides seem promising in distinguishing between TBEV and other flavivirus infections or TBEV vaccine derived immunity.