Coevolution analysis finds amino-acids involved in immune evasion in Hepatitis

ALESSANDRA CARBONE; FRANCESCA NADALIN; ELIN TEPPA

Marseille

Jornada; JOBIM2018; 2018

IntroductionHBV is one of the smallest enveloped DNA viruses and the prototype member of the family Hepadnaviridae. The small HBV genome contains four overlapping open reading frames (ORF) that encode seven proteins, the viral polymerase (Pol), three surface proteins, two core proteins and the X protein. The gene of the surface proteins consists of a single ORF divided into three in-frame coding regions or domains called preS1, preS2 and S. The large protein (L) comprises the PreS1, preS2 and S domains; the middle surface protein M contains the Pres2 and S domain and the small surface protein S comprises only the small domain. All three HBV surface proteins are integral membrane proteins. They form the main antigen recognized by the immune system, responsible for the attachment of the virus to the hepatocytes and the epitope binding the neutralizing antibodies. Particularly, the S domain contains the major B cell epitope, known as the ?a? determinant. Mutations in and around the ?a? determinant may result in (1) escape of vaccine-induced immunity, (2) escaping anti-HBV immunoglobulin therapy and (3) cause diagnostic problems due to false negative results in serological tests (Lazarevic, 2014).The S ORF is completely overlapped with the polymerase ORF. The polymerase protein comprises four domains named Terminal Protein, Spacer, Reverse Transcriptase (RT) and RNaseH. The RT domain represents the target of the antiviral agents belonging to nucleotide/nucleoside analogs. The currently available drugs for chronic hepatitis B treatment are two immuno-modulators and five antiviral agents: lamivudine, telbivudine, entecavir, adefovir and tenofovir (Palumbo, 2008). The major limitation of long-term therapy is antiviral resistance.The main mutation associated with lamivudine resistance are M204I/V that appear to impair replication, the most common compensatory mutation is L180M, that restores the replicative capacity. In vitro studies have demonstrated that this mutation alone is insufficient to result in lamivudine resistance but it augments both viral replication and lamivudine resistance in the context of M204I/V (Bartholomeusz & Locarnini, 2006). Due to the overlapping between S and Pol ORFs, mutations arising in the RT domain cause the appearance of mutations in the surface proteins conferring to the virus the ability to evade the immune system (Croagh, Desmond, & Bell, 2015; Datta, Chatterjee, Veer, & Chakravarty, 2012).A coevolutionary analysis is an attractive approach to find out important positions and predict compensatory mutations associated with the primary resistance mutations. Coevolutionary analysis of HBV sequences represents a challenge due to the high conservation level of protein sequences and the high number of sequences available. Until now, there is no available method to compute coevolution in such data set. In one hand, a large panel of methods exists to compute coevolution in a large set of diverse sequences (de Juan, Pazos, & Valencia, 2013). On the other hand, a handful of methods exist to compute co-evolution signals on small sets of sequences such as CAPS (Fares & McNally, 2006) and BIS2 (Champeimont, Laine, Hu, Penin, & Carbone, 2016; Dib & Carbone, 2012; Oteri, Nadalin, Champeimont, & Carbone, 2017). The BIS2 method was specifically designed to identify clusters of coevolving positions in alignments with high conservation levels (such as viral genomes), or with a relatively low number of sequences (less than 50 sequences). It was successfully applied to reconstruct the protein-protein interaction network of the Hepatitis C Virus (Champeimont et al., 2016) and to identify a novel fusion mechanism in HCV (Douam et al., 2018). However, a novel strategy was needed to compute coevolution using BIS2 in all the available HBV protein sequences.Here, we present a coevolution analysis of mutations associated to drug resistance in the four major HBV genotypes A, B, C, and D. Here, we present a coevolution analysis of the Pol and L proteins of HBV by applying BIS2 iteratively on selected subsets of the set of HBV sequences, where the set is defined for each viral genotype separately. In the RT Pol domain, we found high coevolution signals at positions involved in drug resistance. In the surface protein we found high coevolution in 6 out of 7 positions involved in vaccine escape mutations; in 6 out of 15 involved in immune globulin escape; 3 out of 4 known to be connected with Lamivudine resistance and 10 out of 19 reported as ?diagnostic escape? mutants. Understanding the relationship between mutations in Pol and L proteins may provide valuable information to improve diagnostic procedures and to create more efficient therapeutic protocols. Material and MethodsData setSequences of L and Pol proteins from genotypes A, B, C, and D were retrieved from HBVdb (Hayer et al., 2013). We filtered out incomplete sequences (i.e. truncated proteins) and we retained entries when both protein sequences were available in the same genome. We ended up with 972, 1809, 2006 and 955 sequences belonging to genotypes A, B, C, and D respectively. The average identity in the data set is ~96% for both proteins. A multiple sequence alignment was built for each genotype for L and Pol proteins using clustal omega (Sievers & Higgins, 2018). The resulting 8 alignments were used as input for the coevolution analysis.Coevolution methodAn iterative strategy was used to compute the BIS2 method in a large number of highly conserved sequences. In the first step, the phylogenetic tree T is predicted from the aligned sequences (BioNJ). In the second step, the sequences corresponding to each subtree T' of T are used to perform BIS2 prediction. In the third step, statistically significant clusters are selected (P-value ≤ 0.005). In this last step, appropriate criteria are applied to remove the redundancy of the coevolution clusters whenever they are drawn from non-disjoint trees, based on the significance and coevolution pattern and on the number of elements in the clusters.We considered, for further analysis, the 10 coevolving clusters with the best p-value for each genotype. The clusters of coevolving residues are sorted in increasing order by p-value and ranked from 1 to 10.Results and discussionTop 10 coevolving clusters in the L proteinWe found that positions related to immune escape are present in the top 10 coevolving clusters of the L surface protein in the analyzed genotypes. Namely, we identified:position 130 showing a high coevolution signal in genotypes A, B and C. Mutations at position 130 are related to immune globulin and diagnostic escape.6 out of 7 positions responsible of vaccine escape.6 out of 15 positions related to immune globulin escape. due to the overlapping between S and P ORFs, there are mutations in the S ORF that emerge in connection with lamivudine resistance. We found high coevolution in 3 out of 4 of positions. high coevolution signal in 10 of 19 positions related to diagnostic escape.The complete description of the top 10 coevolving clusters at the L protein mutant positions associated with immune or diagnostic escape is summarized in additional file Table 1. Top 10 coevolving clusters in the RT domain of the Pol proteinIn the RT Pol domain, the highest signal of coevolution was found at position 204 in all genotypes showing the amino acid variations M204I/V that correspond to the most common drug resistance mutation.In genotype A residues, L180 and M204 have the highest signal of coevolution, they coevolve with S109 and N248. The coevolution between L180 and M204 was also found more than once among the top 10 clusters of genotypes C and D, and it corresponds to the most common compensatory mutation. This double mutation is known to confer resistance to lamivudine and telbivudine and to reduce the susceptibility to entecavir and adefovir agents.In genotype C, L180 and M204 appear twice in the top 10 clusters, and they do not coevolve with any other position.In genotype D, positions 180 and 204 appear in two clusters, respectively comprised of positions 180, 204 and 229, and positions 80, 91, 180, 204, 253, 266 and 315. The variations L229W/V and L80I/V are associated with telbivudine resistance, whereas the remaining positions have not been previously reported as important positions. Also, the mutations L80V/I have been reported as a compensatory mutation of the primary mutation M204V/I that confers lamivudine resistance. It is worth mentioning that the cluster above, containing the variation L80I/V related to lamivudine resistance, also contains seven positions that belong to Spacer domain in Pol (Pol positions: 178, 208, 235, 286, 290, 294 and 301). This result suggests that important variation related to drug resistance could be outside the RT domain.In genotype B, position 204 coevolves with positions 80 and 830. As mentioned before, covariation of positions 204 and 80 is associated with drug resistance whereas position 830, which belongs to RNaseH Pol domain, has not been previously reported as an important position. The analysis of the top 10 coevolving clusters at the RT domain mutant positions associated with drug resistance is summarized in additional file Table 2. ConclusionGiven the amount of data available for HBV, this virus is a good model to study coevolutionary signals due to biological functions in contrast to structural contacts. The high level of conservation of HBV makes the coevolutionary analysis challenging or even impossible for state-of-the-art covariation methods. In this work, we propose an iterative approach for the BIS2 method that is able to predict clusters of coevolving residues at the L and Pol HBV proteins.Analyzing the L protein we found high coevolution signals at positions that were reported as responsible of vaccine escape, immune globulin escape, and diagnostic escape. Regarding the Pol protein we found high coevolution at positions responsible for antiviral resistance, including the known compensatory mutations at positions 204 and 180. The coevolving clusters that contain positions related to drug resistance also include other positions that had not been previously reported as important. Further analyses are needed to evaluate the effect of variation on those positions, it is likely that they may play a role in drug resistance or as compensatory mutation to restore viral fitness.ReferencesBartholomeusz, A., & Locarnini, S. A. (2006). Antiviral drug resistance: clinical consequences and molecular aspects. Seminars in Liver Disease, 26(2), 162?170.Champeimont, R., Laine, E., Hu, S.-W., Penin, F., & Carbone, A. (2016). Coevolution analysis of Hepatitis C virus genome to identify the structural and functional dependency network of viral proteins. Scientific Reports, 6, 26401.Croagh, C. M., Desmond, P. V., & Bell, S. J. (2015). Genotypes and viral variants in chronic hepatitis B: A review of epidemiology and clinical relevance. World Journal of Hepatology, 7(3), 289?303.Datta, S., Chatterjee, S., Veer, V., & Chakravarty, R. (2012). Molecular biology of the hepatitis B virus for clinicians. Journal of Clinical and Experimental Hepatology, 2(4), 353?365.de Juan, D., Pazos, F., & Valencia, A. (2013). Emerging methods in protein co-evolution. Nature Reviews. Genetics, 14(4), 249?261.Dib, L., & Carbone, A. (2012). Protein fragments: functional and structural roles of their coevolution networks. PloS One, 7(11), e48124.Douam, F., Fusil, F., Enguehard, M., Dib, L., Nadalin, F., Schwaller, L., ? Lavillette, D. (2018). A protein coevolution method uncovers critical features of the Hepatitis C Virus fusion mechanism. PLoS Pathogens, 14(3), e1006908.Fares, M. A., & McNally, D. (2006). CAPS: coevolution analysis using protein sequences. Bioinformatics , 22(22), 2821?2822.Hayer, J., Jadeau, F., Deléage, G., Kay, A., Zoulim, F., & Combet, C. (2013). HBVdb: a knowledge database for Hepatitis B Virus. Nucleic Acids Research, 41(Database issue), D566?D570.Lazarevic, I. (2014). Clinical implications of hepatitis B virus mutations: recent advances. World Journal of Gastroenterology: WJG, 20(24), 7653?7664.Oteri, F., Nadalin, F., Champeimont, R., & Carbone, A. (2017). BIS2Analyzer: a server for co-evolution analysis of conserved protein families. Nucleic Acids Research, 45(W1), W307?W314.Palumbo, E. (2008). New drugs for chronic hepatitis B: a review. American Journal of Therapeutics, 15(2), 167?172.Sievers, F., & Higgins, D. G. (2018). Clustal Omega for making accurate alignments of many protein sequences. Protein Science: A Publication of the Protein Society, 27(1), 135?145.