Immunotherapy represents an attractive option for the treatment of chronic hepatitis B virus (HBV) infection. The HBV proteins polymerase (Pol) and HBx are of special interest for antigen-specific immunotherapy because they are essential for viral replication and have been associated with viral control (Pol) or are still expressed upon viral DNA integration (HBx). Here, we scored all currently described HBx- and Pol-derived epitope sequences for viral indispensability and conservation across all HBV genotypes. This yielded 7 HBx-derived and 26 Pol-derived reported epitopes with functional association and high conservation. We subsequently predicted novel HLA-binding peptides for 6 HLA supertypes prevalent in HBV-infected patients. Potential epitopes expected to be the least prone to immune escape were subjected to a state-of-the-art in vitro assay to validate their HLA-binding capacity. Using this method, a total of 13 HLA binders derived from HBx and 33 binders from Pol were identified across HLA types. Subsequently, we demonstrated interferon gamma (IFN-γ) production in response to 5 of the novel HBx-derived binders and 17 of the novel Pol-derived binders. In addition, we validated several infrequently described epitopes. Collectively, these results specify a set of highly potent T cell epitopes that represent a valuable resource for future HBV immunotherapy design.
Multiple HBV-derived T cell epitopes have been reported, which can be useful in a therapeutic vaccination strategy. However, these epitopes are largely restricted to HLA-A*02, which is not dominantly expressed in populations with high HBV prevalence. Thus, current epitopes are falling short in the development of a global immunotherapeutic approach. Therefore, we aimed to identify novel epitopes for 6 HLA supertypes most prevalent in the infected population. Moreover, established epitopes might not all be equally effective as they can be subject to different levels of immune escape. It is therefore important to identify targets that are crucial in viral replication and conserved in the majority of the infected population. Here, we applied a stringent selection procedure to compose a combined overview of existing and novel HBV-derived T cell epitopes most promising for viral eradication. This set of T cell epitopes now lays the basis for the development of globally effective HBV antigen-specific immunotherapies.
Chronic hepatitis B virus (CHB) infection affects roughly 250 million people worldwide and is a main cause of cirrhosis and hepatocellular carcinoma (HCC). Chronically infected patients can be treated with expensive viral replication inhibitors, but complete viral eradication as in hepatitis C virus infection is rare. In fact, curative treatment remains highly demanded since CHB infection is expected to remain a global health problem for many years.
Immunotherapy had already emerged in the 1990s as a promising option to treat CHB. T cell responses are considered essential for viral clearance but are scarce or exhausted in CHB patients. Still, they can be boosted or induced via several therapeutic strategies, e.g., vaccination or adoptive transfer of engineered hepatitis B virus (HBV)-specific T cells. However, these strategies are hampered by the lack of an HLA-broad epitope repertoire against which antiviral T cells can be directed. The majority of currently described epitopes are restricted to HLA-A*02, which is highly prevalent in Caucasians. Yet HLA-A*02 is expressed in only roughly 40% of the world’s population and is not dominant in Asian and African populations, whereas especially these populations show a high HBV prevalence. Thus, it is vital to identify non-HLA-A*02-restricted epitopes, especially for HLA types prevalent among Asians and Africans, such as HLA-A*24 or HLA-B*07.
Next to a lack of HLA diversity, the current HBV-derived epitope repertoire is skewed by the fact that many reports focus on dissecting T cell responses against the HBV surface antigen (HBsAg) or core antigen (HBcAg). However, the proteins X (HBx) and polymerase (Pol) also pose interesting targets, as both are vital for viral persistence and interfere with the antiviral immunity of the host. Furthermore, HBx is expressed only in infected hepatocytes and is involved in the development of HCC. The expression of HBx is likely to be retained upon HCC formation because of the productive integration of the HBx gene into the host genome. Collectively, this provides a rationale to target HBx in patients suffering from CHB as well as HBV-related HCC. Besides HBx, Pol represents an interesting immunotherapeutic target. Pol is more immunogenic than HBsAg in HBV transgenic mice, and high frequencies of Pol-specific T cells are associated with viral control after discontinuation of viral replication inhibitors in patients. This implies that Pol-specific T cells retain their function throughout the course of chronicity and can contribute to immune control in vivo. Others have explored strategies to predict HLA-I epitopes from Pol but focused exclusively on a single HLA type or assessed only a limited number of HBV sequences. Taken together, the identification of novel, non-HLA-A*02-restricted HLA-I epitopes derived from HBx and Pol would greatly benefit generic anti-HBV immunotherapy design.
In addition to a limited epitope repertoire, there is another hurdle in the development of HBV-directed immunotherapy. Established epitopes might not all be equally effective, as they can be subjected to different levels of viral mutagenesis and subsequent immune escape. Indeed, previous reports clearly demonstrate that HBV is subject to immune pressure and that mutation of epitope sequences leads to immune evasion or even HBV reactivation. Importantly, in the case of immune escape, responsive memory T cells may still linger in patients despite the loss of epitope presentation on target cells. Thus, prevalent detection of cognate T cells by itself offers no guarantee of clinical relevance. Long-term efficacy of generic immunotherapy can be expected to be dictated by epitope preservation across the patient population, in which infection is caused by different HBV genotypes. Moreover, amino acids conserved between genotypes are more likely to have functional importance to the virus. After all, mutation of functional sequences would lead to a loss of viral fitness, which drives subsequent negative selection. Indeed, amino acids essential for HBV replication are almost exclusively highly conserved. Taken together, T cell responses directed against conserved epitopes from functional protein domains would benefit the majority of patients while simultaneously hampering viral replication and immune escape.
Here, we have taken an effort to tackle the above-mentioned issues by integrating viral indispensability, genomic variation, HLA binding, and immunogenicity to identify the best HBx- and Pol-derived T cell epitopes for immunotherapy across 6 of the most prevalent HLA supertypes within the HBV-infected population. The results of this study pave the way for the development of globally effective HBV antigen-specific immunotherapies.
We first set out to rank reported epitopes for HBx and Pol based on protein conservation and function using a comprehensive database called Hepitopes. From Hepitopes, we extracted all unique epitopes identified in human hosts for HBx (n = 14) and Pol (n = 50) (Fig. 1, left), which were found to be largely HLA-A*02 restricted (see Fig. S1 in the supplemental material). To rank these epitopes based on conservation, all protein sequences for HBx and Pol were extracted from a large public HBV repository (HBVdb) (Fig. 1, right) and used to compute an overall consensus sequence across viral genotypes (see Materials and Methods). Based on this sequence, a conservation score (prevalence) was calculated (Fig. 2 and 3 for HBx and Pol, respectively). To additionally rank functional associations, we extended our analysis to described functional domains and amino acids that severely impaired viral function upon mutation (see Tables S1 and S2 in the supplemental material for details on reviews and studies used). These were additionally aligned to the consensus sequences of HBx and Pol (Fig. 2 and 3, respectively, arrows and bottom panels). As expected, functional domains and essential amino acids mostly aligned to highly conserved regions. Finally, we used the acquired information on conservation and functionality to rank all reported HBx- and Pol-derived epitopes (Tables 1 and 2, respectively).
Workflow of epitope ranking and discovery. All amino acid sequences from HBx and Pol were extracted from the publicly available database HBVdb and used to compute a consensus sequence and conservation score (top right) (see Materials and Methods). The resulting consensus sequences were subsequently used in NetMHCpan to predict sequences of HLA-binding peptides across the 6 most common HLA supertypes in the HBV-infected population. This yielded 251 and 1,655 HBx- and Pol-derived potential HLA binders, respectively. A total of 113 of these were selected for in vitro validation based on predicted HLA binding, conservation, and reported functional association within the viral protein. This yielded 13 HBx-derived and 33 Pol-derived validated HLA binders. All of these were subsequently tested for immunogenicity, in which 6 HBx- and 24 Pol-derived peptides elicited IFN-γ responses in an ELISA. In addition, all currently known Pol- and HBx-derived epitopes were extracted from the publicly available database Hepitopes (top left) and ranked according to conservation score and reported functional association. These findings are summarized in Table 4 (bottom center).
Alignment of reported and predicted CD8+ T cell targets based on protein conservation and function for HBx. The centered bar diagram depicts the length of the consensus sequence of the HBx protein (see Materials and Methods), in which the conservation score across viral genotypes is indicated by a color code (key) for each amino acid. Reported epitopes obtained from the Hepitopes database are aligned to this sequence and shown on top. Below this, potential novel binders predicted by NetMHCpan (9 to 11 amino acids) are depicted for each HLA supertype representative. The gray histogram represents the frequency of each amino acid within all predicted binders (8 to 14 amino acids long) over the protein sequence. Essential amino acids for which mutation leads to a loss of viral persistence are indicated by arrows matching the color of the conservation score. Functional domains are depicted at the bottom according to the nomenclature of HBVdb. References describing the experimental evidence for essential amino acids and functional domains are listed in Table S1 in the supplemental material.
Alignment of reported and predicted CD8+ T cell targets based on protein conservation and function for Pol.
