Mutational landscape of the surface antigen of hepatitis B virus in patients with hepatocellular carcinoma
- Open Access
- 01.12.2025
- Research
Erschienen in:
Abstract
Introduction
HBV is a major health burden with more than 300 million people chronically infected worldwide [1]. The complications of chronic HBV are a leading cause of developing progressive liver diseases such as LC and HCC. The HBV-related deaths are currently estimated at 820,000 deaths annually [1, 2].
HBV is a DNA virus with a partially double-stranded genome and four overlapping open reading frames (ORFs) [3]. Due to this genome organization, the lack of proofreading during reverse transcription, and its high replication rate, the HBV genome accumulates more mutations than most DNA viruses [4]. This allows the virus to acquire advantageous mutations that affect different steps in the viral life cycle and pathogenesis. The surface antigen (HBsAg) of the virus is a region where multiple such mutations have been reported and were associated with progressive liver diseases [5, 6]. HBsAg is encoded by the pre-S/S gene, which contains three inframe start codons resulting in three isoforms: the small (S), the middle (S and pre-S2), and the large (S, pre-S2 and pre-S1) HBsAgs [7]. The large HBsAg (L-HBsAg) is 389–397 aa in length, while the middle HBsAg (M-HBsAg) and the small HBsAg (S-HBsAg) are 281 and 226 aa, respectively. The ratio of the isoforms in mature virions is typically 1:1:4 for S-HBsAg, M-HBsAg, and L-HBsAg, respectively. Biologically, HBsAg contains elements essential for hepatocyte recognition and viral morphogenesis [8, 9]. It also has an important role in eliciting and modulating immune response during chronic infection [10, 11]. Upon infecting liver cells (hepatocytes), HBsAg is released into the bloodstream [12]. During acute HBV infection, HBsAg levels surge, indicating active viral replication, while in chronic infection, HBsAg persists for at least six months, rendering chronic carriers infectious. The immune response against HBV involves the production of antibodies targeting HBsAg (anti-HBs), although the immune system often fails to clear the virus entirely in chronic infection [13]. Mechanistically, HBsAg facilitates viral entry into hepatocytes by binding to specific receptors, while also shielding the virus from immune recognition, thus impeding viral clearance [14]. Moreover, HBsAg is instrumental in HBV morphogenesis within hepatocytes, aiding in the assembly of new viral particles, which are subsequently released into the bloodstream. Clinically, persistent HBsAg presence signifies ongoing infection and potential transmission risk. Diagnosis relies on HBsAg testing, with a positive result confirming active HBV infection and contagiousness [15]. Notably, transient HBsAg positivity may occur following Hepatitis B vaccination but does not indicate infection. Treatment strategies aim to suppress viral replication and reduce HBsAg levels, while vaccination serves as a preventive measure against HBV infection [16]. Understanding the pivotal role of HBsAg informs diagnostic, therapeutic, and preventive approaches in HBV management.
Anzeige
In Saudi Arabia, HBV infection represents a notable health burden, with distinct epidemiological patterns and clinical outcomes compared to other regions [17]. HBV prevalence is estimated to be \(\sim\)2% in the country [18]. Despite considerable efforts to combat HBV, including vaccination programs and improved health-care infrastructure, the prevalence of chronic HBV infection remains relatively high in Saudi Arabia. Understanding the genetic diversity of HBV circulating in this population and its correlation with clinical outcomes is essential for optimizing diagnostic and therapeutic strategies. In this study, we screened for mutations in the pre-S/S regions of HBsAg (genotype D) in sequences retrospectively generated from samples collected from HBV DNA-positive patients in Saudi Arabia between 2007 and 2011. In the 304 full-length sequences generated, we compared the prevalence of mutations with an overall frequency exceeding 5% in four clinical groups: inactive (n = 180), active (n = 62), LC (n = 36), and HCC (n = 26). Additionally, evolutionary and structural analysis were conducted to infer the impact of the mutations that showed significant association with the HCC group.
Results
The detection of three mutations in HBsAg with significantly high frequency in HCC patients
The HBsAg sequences utilized in this study were previously obtained from four distinct clinical cohorts: inactive, active, LC, or HCC patients [19]. To determine the genotype, an alignment was constructed, encompassing all study sequences along with representative sequences for each genotype. The phylogenetic analysis revealed that the majority of sequences belonged to genotype D, consistent with the previous findings (Figure S1) [19]. After the exclusion of non-genotype D sequences, the remaining sequences (304) were analyzed for amino acid variations. To refine the selection of mutations with potential impact without dismissing functional ones, we selected a moderate threshold of 5%. Nineteen mutations with frequencies > 5% were identified. The mutations distributed across the pre-S and S regions included: T7A, T40P, L54M, I73M, S98T, N103D, F130L, S136N, V147A, H149P, and P160L in pre-S, and F8L, Q30K, I110L, S204R, S207N, I208T, S210R, and L213I in S. Among these mutations, N103D, Q30K, and I208T exhibited significantly higher frequency in HCC compared to other groups (Fig. 1). Moreover, the prevalence of Q30K in the LC group was notably lower than expected, with glutamine exclusively observed at this position in all LC group sequences (p-value 0.035). In the overlapping P ORF, the nucleotide changes causing N103D and Q30K in the S ORF result in K283R and A373E in the Spacer and the Reverse Transcriptase (RT) domains, respectively. Threonine was also observed at high frequency at site 373 due to another change in the same codon. The change causing the mutation I208T, however, is silent in the P ORF (H551).
Deletions in the pre-S region were also identified in nine sequences (inactive=3, active=5, LC=0, and HCC=1) with no significant difference between the overall frequency and the frequency within each group (p = 0.54, p = 0.07, p = 0.6, and p = 0.56, respectively).
The mutational profile at the three sites was significantly associated with HCC in a multivariate analysis
To test the association between the three mutations and the clinical outcome, univariate and multivariate logistic regression analysis were conducted. Age, sex, Alanine aminotransferase (ALT), viral load, and Body Mass Index (BMI), were all previously associated with HCC [20‐22]. Thus, we included in the analysis age (years), sex, BMI, ALT (IU/l), HBV DNA (IU/ml), and the number of mutations (0, 1, 2, or 3 mutations) as independent variables, comparing individuals of each group with the individuals of the other groups combined. Other demographic factors and data, such as serology, other liver function indicators, and antiviral therapy were unavailable for the majority of the samples at the time of the study due to the old nature of the samples. A summary of the demographic data and tests for each clinical group is presented in Table 1.
Anzeige
Only the variables with significant associations in the univariate analysis were included in the multivariate analysis. For the inactive group, age (OR=0.97, 95% CI 0.95–0.98), sex (OR=0.40, 95% CI 0.22–0.69), and HBV DNA (OR=1–6\({\times }10^{-8}\), 95% CI 1–2\({\times }10^{-7}\) to 1–3\({\times }10^{-8}\)) showed significant association in the univariate analysis (Table S1). The multivariate analysis also showed a significant association with younger age (Table S2). Adjusted for age, there were significant associations with males (OR= 2.207, 95% CI 1.13–4.57) and higher HBV DNA (OR= 1+6\({\times }10^{-8}\), 95% CI 1+3\({\times }10^{-8}\) to 1+1\({\times }10^{-7}\)). Hence, the data suggests that in general the individuals in the inactive group are more likely to be younger, female, and with a lower viral load compared to individuals outside of this group. It also suggests that there are more younger males in the inactive group and the younger males in this group are likely to have higher viral loads than younger males outside of this group. For the active and the LC groups, age showed significant association in the univariate analysis (Table S3 & S4). While the active group showed a significant association with younger age (OR= 0.96, 95% CI 0.93–0.98), the LC group was associated with older age (OR= 1.06, 95% CI 1.03- 1.09). Except for the HCC group, none of the clinical groups had a significant association with the number of mutations. The univariate analysis for the HCC group showed significant association with age, BMI, sex, and No of mutations (Table 2). The HCC group are associated with older age, lower BMI, males, and a higher number of mutations at the three sites. In the multivariate analysis, sex could not be included due to quasi-perfect separation (only one out of the 26 HCC patients is female). Hence, the analysis was only adjusted for age and BMI which showed that the presence of the three mutations was significantly associated with HCC (OR=25.70, 95% CI 1.717–810.9) (Table 3).
Evidence of positive selection was detected at the three sites
To dissect the evolutionary dynamics driving HBsAg mutations, we used HyPhy (Hypothesis Testing using Phylogenies) software, which employs phylogenetic analysis to test for selection in multiple sequence alignment. Four models are available to detect selection at individual sites (codons): MEME (Mixed Effects Model of Evolution), FEL (Fixed Effects Likelihood), SLAC (Single-Likelihood Ancestor Counting), and FUBAR (Fast, Unconstrained Bayesian AppRoximation). After constructing a phylogeny comprising all 304 sequences, the four models were applied to test if any of the sites with mutation frequency >5% were subjected to positive selection in the S ORF or the overlapping region of the P ORF (Figure S2). Eight of the 19 sites showed a signature of positive selection in all four models in the S ORF. In the overlapping region of the P ORF, six sites showed a signature of positive selection in all four models. Importantly, sites 103 and 30 were not under positive selection in the S ORF in any of the models, while site 208 was positively selected in all models. In the P ORF, the sites corresponding to 103 and 30 showed positive selection in all models, while the site corresponding to 208 was not positively selected in any of the models (Table 4).
To see whether certain sites experienced selective pressure exclusively within the viral population isolated from HCC patients, a phylogeny containing only the sequences from the HCC cohort was constructed for both ORFs (Figure S3). For this analysis, we applied FUBAR as it is able to identify pervasive (constant) positive selection across all branches and has more power in detecting weak positive selection (low values of \(\omega\) >1). Using this method, positive selection was detected in all three sites in the S ORF. The sites corresponding to 103 and 30 were also positively selected in the P ORF, while the site corresponding to 208 was negatively selected (Table 5). Positive selection in overlapping genes is often conflated with relaxed purifying selection. Sites evolving under relaxed purifying selection are expected to fall in a region enriched with non-synonymous mutations with strong negative selection in the overlapping ORF. Thus, to distinguish between positive selection and relaxed purifying selection, we identified all the positively and negatively selected sites in the S ORF and the overlapping region of the P ORF in the sequences isolated from the HCC cohort. Using the same method, FUBAR, the analysis revealed that site 103 mainly falls in a region enriched with negatively selected sites in the S ORF and a region enriched with positively selected sites in the overlapping P ORF (Fig. 2). Conversely, sites corresponding to 30 and 208 fall in regions enriched with negatively selected sites in the P ORF. This may indicate that, in the S ORF, site 103 evolved under positive selection while sites 30 and 208 evolved under relaxed purifying selection within the HCC cohort.
The regions spanning the mutations are important for viral morphogenesis and recognition by immunity
The functional domains within the pre-S/S region have been intensively studied which revealed their involvement in various stages of the viral life cycle. The polymerase domains overlapping with the S ORF were also described in previous studies. Notably, the regions housing the three mutations are involved in different steps in the viral life cycle. N103D is situated at the pre-S1-S2 junction (aa 81–127), a region implicated in the binding to the viral nucleoprotein and the cellular factor Hsc70 [8, 23, 24]. Moreover, this region encompasses critical B and T cell epitopes [25‐28]. Q30K and I208T are located within the first cytosolic domain (CYL-I) and the fourth trans-membrane domain (TM4), respectively. Both domains were found to be essential for viral secretion and assembly [29‐31].
In the P ORF, the site corresponding to 30 falls in a conserved region in the RT of polymerase previously reported to be essential for pgRNA packaging [32, 33]. The site corresponding to 208 also falls in a conserved region downstream of the YMDD motif critical for the RT activity and nucleotide binding [34, 35]. This region is a major target for antivirals where multiple antiviral resistant mutations were reported but not at this particular site [36‐38]. Conversely, the site corresponding to 103 resides within the spacer domain, which is known for its plasticity and tolerance for non-synonymous mutations [39]. Table 6 summarizes the regions affected by the mutations and their functions.
The structural analysis suggests an effect of N103D on the inter-molecule interactions of HBsAg
The relationship between structure and function is a fundamental aspect of viral envelope proteins, where mutations often lead to structural alterations and subsequent changes in function. To evaluate the structural impact of the identified mutations on HBsAg, we utilized the Iterative Threading ASSEmbly Refinement (I-TASSER) prediction tool to generate 3D structures based on two of the study sequences. One sequence (A) contained the wild-type (WT) residues at all three sites, while the other sequence (B) harboured all three mutations. Among the models generated for each sequence, the best model was selected based on its similarity to a reference structure (7tuk) regardless of the C-scores. Superimposition of all models with 7tuk highlighted model 5 as the best-fitted for both sequences A and B, exhibiting RMSD values of 6.07 Å and 5.9 Å across all atom pairs, respectively. Other models showed poor RMSD and hence were excluded. Structural studies of L-HBsAg have proposed a topology consisting of a flexible pre-S loop at the N-terminus and four transmembrane helices at the C-terminus (S protein) (Fig. 3a). These helices are connected by flexible loops, including two cytosolic (CYL-I and CYL-II) and one external loop (the major hydrophilic region). Due to their high flexibility, the structure and orientation of these loops are often poorly resolved [40]. Notably, the reference sequence 7tuk lacked these regions, thus, the pre-S and the connecting loops were omitted from subsequent comparisons. Superimposition of structures A and B revealed high structural similarity across the helical region, with an RMSD of 0.57 Å (Fig. 3b). To test the effect of the three mutations on different backgrounds, we introduced them to sequence A and reverted the three sites in sequence B to the WT. The resulting structures (referred to as Mut A and Mut B) exhibited high similarity to each other and the original structures, with RMDS <0.5 Å over the helical region, suggesting negligible impact of the mutations at least over this region (Figure S4).
The dimeric form of HBsAg has been proposed to play a significant role during the virus life cycle [40]. To explore whether any of the mutations could influence interactions at the dimeric interface, two molecules of each structure were superimposed onto a 7tuk dimer (Figure S5). In the case of Structure A dimer, interactions between residue 103 N in monomer 1 and residues 317 S (S 154) and 321 F (S 158) in monomer 2 were observed. Furthermore, interactions between residue 103 N in monomer 2 and residues 321 F (S 158) and 325 L (S 162) in monomer 1 were noted (Fig. 4a). Conversely, the Mut A dimer showed no interactions involving residue 103 with any other residue (Fig. 4b). For Structure B, an interaction was observed between residue 103D in monomer 1 and residue 381I (S 218) in monomer 2 (Fig. 4c). In Mut B, an interaction between residue 103 N in monomer 1 and residue 324 F (S 161) in monomer 2 was identified (Fig. 4d). Overall, our predicted models suggest that while the mutations may not significantly impact the core structure of HBsAg, they might influence oligomerization or interactions with other molecules.
N103D is associated with a reduction in the score of predicted B and T cell epitopes
In light of the structural analysis which indicated a potential impact of the N103D mutation on inter-molecular interactions, we sought to investigate its effect on recognition by humoral and cellular immunity. Utilizing the DiscoTope online tool, which predicts discontinuous epitopes based on 3D structures, we examined whether the mutation disrupts any predicted B cell epitopes. The analysis revealed three major epitopes: amino acids (aa) 24–46 (pre-S1), aa 83–110 (pre-S1-S2 junction), and aa 50–68 (S), consistently present in all produced structures (Figure S6). These epitopes have also been identified in previous epitope mapping studies of HBV genotype D and other genotypes [26, 27, 41, 42]. Notably, the “a” determinant (aa 124–147) in the S region was not detected by the prediction tool, although two structures exhibited positive scores at sites 114, 116, and 128. Importantly, the comparison of propensity scores for the pre-S1-S2 epitope at position 103 between Structure A and Mut A yielded scores of 2.388 and \(-\)1.079, respectively. Similarly, for Structure B and Mut B, the scores were 0.483 and 3.682, respectively. These findings suggest that the N103D mutation may exert a disruptive effect specifically on the pre-S1-S2 epitope.
Anzeige
Both CD4\(^{+}\) and CD8\(^{+}\) T cell responses play a major role in HBV infections and chronicity [43]. Since residue 103 is also located in a previously identified T cell epitope (see Table 6), we used the T cell epitope prediction tool TepiTool to test the effect of the mutation. A predicted MHC-class I epitope (aa 103–111) was the fifth highest-ranking peptide for Sequence A and Mut B (score 0.977). In contrast, the epitope was not among the top 20 highest-raking peptides for Sequence B or Mut A (score 0.696). For MHC class II epitope, none of the highest-ranking peptides spanned residue 103 in all the sequences. Similarly, positions 30 and 208 were not predicted to be among the highest-ranking peptides in any of the sequences. Thus, in addition to possible disruption of a B cell epitope, the results suggest that N103D may also disrupt a T cell epitope.
Tables
Table 1
Summary of the demographic data and tests of the clinical groups
Age | Sex | BMI | ALT | HBV DNA | |
|---|---|---|---|---|---|
(years) | No. (%) | (IU/l) | (IU/ml) | ||
Inactive | 39 | Male= 119 (66%) | 27.55 | 27.00 | 197.0 |
(n = 180) | (31.0–50.0) | Female= 61 (34%) | (24.77–31.83) | (19.00–41.00) | (19.00–1329) |
Active | 37 | Male= 51 (82.3%) | 28.07 | 61.00 | 1.1x10\(^{6}\) |
(n = 62) | (26–45.25) | Female= 11 (17.7%) | (21.47–31.93) | (40.5–100) | (52932–9.4x10\(^{7}\)) |
LC | 54 | Male= 27 (75%) | 26.15 | 38.50 | 1522 |
(n = 36) | (44–59.50) | Female= 9 (25%) | (22.35–29.36) | (27–66) | (57–5.3x10\(^{4}\)) |
HCC | 62 | Male= 25 (96.2%) | 23.98 | 56.50 | 12967 |
(n = 26) | (52.5–70) | Female= 1 (3.8%) | (21.62–27.35) | (39.50–79.25) | (481.8–7.4x10\(^{5}\)) |
Total | |||||
(n = 304) |
Table 2
Univariate analysis to determine the independent variables associated with HCC
Univariate logistic regression analysis | |||||||||
|---|---|---|---|---|---|---|---|---|---|
HCC | non-HCC | Odds ratio | |Z| | p value | C-statistic | p value | Likelihood | p value | |
(95% CI) | ratio test | ||||||||
Age | 62 | 40 | 1.146 | 5.653 | <0.0001 | 0.882 | <0.0001 | 53.86 | <0.0001 |
(years) | (52.5–70) | (31–50) | (1.097–1.207) | ||||||
ALT | 56.5 | 34 | 1.000 | 0.051 | 0.9592 | 0.697 | 0.0052 | 0.003 | 0.9582 |
(IU/l) | (39.50–79.25) | (21–58) | (0.996–1.001) | ||||||
BMI | 23.98 | 27.44 | 0.8790 | 2.531 | 0.0114 | 0.712 | 0.0047 | 7.227 | 0.0072 |
(21.62–27.35) | (23.88–31.23) | (0.791–0.967) | |||||||
Sex | Male = 25 (96.2%) | Male = 197 (70.9%) | 10.28 | 2.266 | 0.0235 | 0.627 | 0.0330 | 10.51 | 0.0012 |
Female = 1 (3.8%) | Female = 81 (29%) | (2.123 −185.0) | |||||||
No. of | 1 | 0 | 2.695 | 4.003 | <0.0001 | 0.658 | 0.0078 | 15.18 | <0.0001 |
mutations | (0–1.25) | (0–1) | (1.658 −4.430) | ||||||
HBV DNA | 12967 | 782 | 1–2\({\times }\)10−8 | 0.770 | 0.4415 | 0.619 | 0.1334 | 1.134 | 0.2870 |
(IU/ml) | (481.8–7.4x105) | (47.25 to 21788) | (1–1\({\times }\)10−7 to | ||||||
1+2\({\times }\)10−9) | |||||||||
Table 3
Multivariate analysis to determine the independent variables associated with HCC
Multivariate logistic regression analysis | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Odds ratio | |Z| | p value | VIF | C-statistic | p value | AIC | AIC | Pseudo | Reject null | |
(95% CI) | (no model) | (model) | R squared | hypothesis? (simpler model is correct) | ||||||
Age | 1.125 | 4.206 | <0.0001 | 1.060 | ||||||
(years) | (1.070–1.196) | |||||||||
Yes, p | ||||||||||
BMI | 0.821 | 2.561 | 0.0104 | 1.025 | 0.913 | <0.0001 | 116.7 | 78.27 | 0.39 | value |
(0.696–0.945) | <0.0001 | |||||||||
No. of | 2.197 | 1.062 | 0.2883 | 1.052 | ||||||
mutations (1) | (0.5034 −9.859) | |||||||||
No. of | 6.816 | 1.809 | 0.0705 | 1.036 | ||||||
mutations (2) | (0.717–54.28) | |||||||||
No. of | 25.70 | 2.167 | 0.0302 | 1.059 | ||||||
mutations (3) | (1.717 −810.9) | |||||||||
Table 4
Positively selected codons in the S ORF and the overlapping region of the P ORF inferred by MEME, FEL, SALC, and FUBAR methods in sequences isolated from all clinical cohorts
S ORF | Overlapping | ||||||||
|---|---|---|---|---|---|---|---|---|---|
Codon | MEME | FEL | SLAC | FUBAR | P ORF | MEME | FEL | SLAC | FUBAR |
Codon | |||||||||
pre-S | |||||||||
7 | \(\surd\) | 187 | \(\surd\) | \(\surd\) | \(\surd\) | ||||
40 | 220 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | ||||
54 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 234 | ||||
73 | 253 | ||||||||
98 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 278 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) |
103 | 283 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | ||||
130 | \(\surd\) | \(\surd\) | 310 | ||||||
136 | \(\surd\) | \(\surd\) | 316 | ||||||
147 | 327 | ||||||||
149 | \(\surd\) | 329 | |||||||
160 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 340 | ||||
S | |||||||||
8 | \(\surd\) | \(\surd\) | 351 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | ||
30 | 373 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | ||||
110 | \(\surd\) | \(\surd\) | \(\surd\) | 453 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | |
204 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 547 | ||||
207 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 550 | ||||
208 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 551 | ||||
210 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 553 | ||||
213 | \(\surd\) | \(\surd\) | \(\surd\) | \(\surd\) | 556 | ||||
Table 5
Positively and negatively selected codons in the S ORF and the overlapping region of the P ORF inferred by FUBAR method in sequences isolated from the HCC cohort
S ORF | Overlapping | ||
|---|---|---|---|
Codon | Selection | P ORF | Selection |
Codon | |||
pre-S | |||
7 | – | 187 | – |
40 | – | 220 | Positive |
54 | – | 234 | Positive |
73 | Positive | 253 | Negative |
98 | – | 278 | Positive |
103 | Positive | 283 | Positive |
130 | – | 310 | Positive |
136 | – | 316 | Negative |
147 | – | 327 | Negative |
149 | – | 329 | Negative |
160 | – | 340 | Negative |
S | |||
8 | – | 351 | – |
30 | Positive | 373 | Positive |
110 | – | 453 | – |
204 | – | 547 | – |
207 | Positive | 550 | Negative |
208 | Positive | 551 | Negative |
210 | – | 553 | – |
213 | – | 556 | – |
Table 6
Regions spanning N103D, Q30K, and I208T in the S ORF and their overlapping codons in the P ORF
Position in pre-S/S | Region | Function | Reference | Corresponding Position in polymerase | Region | Function | Reference |
|---|---|---|---|---|---|---|---|
103 | Pre-S1 C terminus | aa 94–117 B/T cell epitope, aa 81–105 cytosolic anchorage domain (heat shock protein Hsc70 binding site), aa 103–127 nucleocapsid binding site. | 283 | Spacer domain | Unknown | [39] | |
30 | S (first cytosolic domain CYL-I) | S antigen secretion and incorporation into nascent virions. | 373 | RT domain (N-Terminus) | Part of conserved region involved in pgRNA packaging. | ||
208 | S (TM4) | S antigen secretion. | [31] | 551 | Downstream of the YMDD motif (aa 538–541) | Unknown |
Anzeige
Fig. 1
Mutations associated with increased frequency in the HCC group. The graphs show the frequency of mutations N103D, Q30K, and I208T in the clinical groups (HCC, inactive, active, and LC) compared to the overall frequency. (*) indicates significance compared to the respective overall frequency (p-value <0.05 is considered significant)
Figures
Fig. 2
Positively and negatively selected codons in the S ORF and the overlapping region of the P ORF inferred by FUBAR method in sequences isolated from the HCC cohort. The positively selected sites are shown above the bars and the negatively selected sites are shown below. Sites 103, 30, and 208 in the S ORF and their corresponding sites in the P ORF are shown in red
Fig. 3
Structural prediction of S region for two of the study isolates. Two study sequences were used to generate the models: Sequence A, has the WT residues at 103 (pre-S), 30, and 208 while Sequence B has D, K, and T at these sites, respectively. Five models for each sequence were generated in I-TASSER prediction tool. The final models were selected based on the goodness-of-fit with a reference structure (7tuk HBV genotype E). The reference sequence was selected based on the highest- ranking structure in RCSB PDB. a Shows the structure of 7tuk monomer over the S region without the pre-S or the connecting loops. Helices 1–4 are shown (H1-H4). b Shows the superimposition of the generated structures (A and B) with the reference structure (7tuk). The images were created in UCSF Chimera
Fig. 4
The effect of N103D on L-HB sAg dimer interface interactions. Two molecules of each structure were superimposed on 7tuk dimer. The figure shows Structures A (a), Mut A (b), Structure B (c), and Mut B (d). The enlarged areas show the interactions of the residue at 103 in each of the structures (except Mut A). Predicted interactions are shown as black pseudo-bonds with atoms with \(\ge\)0.6 VDW overlap. Some loops were hidden for better visualization of the structures. The images were created in UCSF Chimera
Materials and methods
Patient cohort and samples
In this cross-sectional study, HBV DNA-positive samples collected between 2007 and 2011 were retrospectively analyzed. The samples were obtained from patients who were originally enrolled from three medical centres in Saudi Arabia: King Faisal Specialist Hospital and Research Center, King Khalid University Hospital, and Prince Sultan Military Medical City [19]. Written informed consent was obtained from all participating individuals, and approvals were obtained from the institutional review board of the participating hospitals in accordance with the Helsinki Declaration of 1975. The participants were selected based on clinical findings and categorized into four groups: inactive, active, LC, or HCC. The inactive group was defined as those who were positive for HBsAg, negative for HBeAg, normal ALT, HBV DNA <2000 (IU/ml), and normal liver fibro-scan test. This group was not subjected to antiviral therapy. The active group was defined as those who were positive for both HBsAg and HBeAg with high viral load. For the LC group, cirrhosis was confirmed either by liver biopsy, clinical, biochemical or radiological evidence of cirrhosis. Diagnosis of HCC was made by computed tomography and/or magnetic resonance imaging of the liver, according to the published guidelines for the diagnosis and management of HCC [44]. Patients with autoimmune, metabolic liver illnesses, or cases of co-infection with human immunodeficiency virus (HIV), hepatitis D virus (HDV), or hepatitis C virus (HCV) were excluded from this cohort. The final number of participants after the exclusion was 319.
Viral sequences and phylogenetic analysis
HBsAg sequences were generated as described previously [19]. The analysis was carried out in MEGA (version 11.0.13). The consensus sequences were deposited in GenBank (accession numbers: PV587505-PV587822). To generate phylogenetic trees, a model selection test was first run to determine the best-fitted model and the model with the lowest BIC score was applied. For all trees, a bootstrap of 1000 was applied. For amino acid variation analysis, a 5% threshold for mutation frequency was applied. This was based on previous studies which showed that functional mutations in HBsAg had a rate of >1%. While some studies set the threshold at 1% to scan for immune escape mutations in HBsAg, others used thresholds of 5% or even 10% [45‐47]. Thus, to refine the selection of mutations with potential impact without dismissing functional ones, we selected a moderate threshold of 5%.
Anzeige
Positive selection analysis
Hyphy (Hypothesis Testing using Phylogenies) was used for the codon-based selection analysis of the S and the P ORFs. It is an open-source software employed to infer selection utilizing techniques in phylogenetics, molecular evolution, and machine learning [48]. Four models are available to detect selection at individual sites (codons): MEME (Mixed Effects Model of Evolution), FEL (Fixed Effects Likelihood), SLAC (Single-Likelihood Ancestor Counting), and FUBAR (Fast, Unconstrained Bayesian AppRoximation). MEME employs a mixed-effects maximum likelihood approach to detect both episodic and pervasive selection that acts on individual sites in a subset of lineages (a proportion of branches) [49]. FUBAR, SLAC, and FEL all assume that the selection act on individual sites is constant along the entire phylogeny [50, 51]. Positive selection at each site is detected when the positive selection component \(\beta\) is > the synonymous substitution rate \(\alpha\) (or \(\omega\) >1). For all methods, except FUBAR, p-value < 0.05 was considered significant. For FUBAR, a posterior probability of > 0.9 is strongly suggestive of positive selection.
Structural analysis
The protein models of the study sequences were generated in I_TASSER (Iterative Threading ASSEmbly Refinement) prediction tool. I_TASSER identifies structural templates from the PDB by multiple threading approach LOMETS (Local Meta-Threading Server, version 3) [52]. For each sequence, five models were generated. The final model was selected based on the structural similarity (goodness-of-fit) with a reference structure regardless of the C- or TM-scores. 7tuk was used as a reference structure as it was the highest-ranking protein structure in the RCSB PDB. It is the structure of HBsAg of genotype E determined by Cryo-EM at a resolution of 6.3 Å [40]. Structures comparison, visualization, and generation of images were done in UCSF Chimera (version 1.12).
B and T cell epitope prediction
DiscoTope tool was used to predict discontinuous B cell epitope of the protein structures in PDB format. The default threshold value of this version is \(-\)7.7 which corresponds to a specificity of 75%. This method involves solvent-accessible surface area and contact distance calculations reporting a propensity score for each residue [53]. For T cell epitope prediction, TepiTool was used to predict peptides binding to MHC class I and II [54]. The 27 most frequent alleles were included in the analysis for both classes.
Statistical analysis
Statistical analysis was carried out in GraphPad Prism (version 10.1.1). The overall frequency of each mutation was compared to the frequency in each of the clinical groups using Fisher’s exact test. Simple and multiple logistic regression tests were used for the univariate and the multivariate analysis, respectively. For the logistic regression analysis, each clinical group was compared with the other clinical groups combined. To determine significant associations in the univariate analysis, the Odds Ratio (OR) and c-static (the area under the curve) were reported for each variable. Variables with significant associations were included in the multivariate analysis. Model diagnostic tests were run to determine the goodness-of-fit of the selected model. These tests included: overall c-static, AIC (Akaike’s information criterion), Tjur’s R squared, and hypothesis testing (G squared). Multiple collinearities between covariates were assessed using the variance inflation factor (VIF). Multiple collinearities were defined as positive when the VIF was \(\ge\)5. For all tests, p-value <0.05 was considered significant.
Anzeige
Ethical approval declarations
The study was conducted in accordance with the Declaration of Helsinki (1975), and approved by the Institutional Review Board of King Faisal Specialist Hospital and Research Centre (project number: 2150008, date of approval 6 September 2015). Informed consent was obtained from all subjects involved in the study.
Discussion
HBV is a leading cause of HCC accounting for up to 80% of all cases [55, 56]. HBV is postulated to initiate carcinogenic processes within hepatocytes through a multifaceted array of mechanisms [57]. These mechanisms encompass a spectrum of complex interactions between viral components and host cellular machinery, ultimately culminating in the dysregulation of cellular processes pivotal for maintaining genomic stability and homeostasis. HBV oncogenic potential is thought to stem from its ability to disrupt various cellular pathways involved in proliferation, apoptosis, DNA repair, and immune surveillance [58]. Mutations in the pre-S/S gene were frequently observed in HCC patients [59, 60]. While the role of variations in other HBV genes in HCC is more well-established, the role of the variations in the S gene is poorly understood [61]. Some studies suggested that mutations in this region trigger oncogenic events through mechanisms such as the induction of HBsAg retention or immune escape [6, 62‐64]. The accumulation of HBsAg mutants in the endoplasmic reticulum (ER) was found to cause ER stress and oxidative DNA damage, a mechanism that was suggested to initiate oncogenesis [65]. It was also reported that the accumulated mutants in the ER can activate growth factors, such as the vascular endothelial growth factor-A, inducing oncogenic events [66].
In this cross-sectional study, we have identified three mutations (N103D, Q30K, and I208T) within HBsAg exhibiting significantly higher than expected frequencies in sequences isolated from HCC patients. Deletions in the pre-S region in nine of the analyzed sequences were also identified. While previous studies of other HBV genotypes found a significant association between deletions in this region and HCC, such association was not observed in the current study [67]. This is consistent with previous studies that found no association between deletions in the pre-S region and severe liver diseases in the context of genotype D infections [68, 69].
In the multivariate analysis performed here, combining the three mutations showed a significant association with HCC group but not the non-HCC group in the tested cohort independent of age. However, it is important to acknowledge the limitations of our multivariate analysis, particularly the absence of variables that could potentially have a detrimental effect such as variations in other genes, antiviral therapy data, and other viral markers including HBsAg titers and HBV core-related antigen. The exclusion of these data may have introduced “omitted-variable” biases, potentially leading to an overestimation of the odds ratio associated with the identified mutations. Data for other liver function indicators and enzymes such as AST, ALP, and GGT, were also unavailable. High levels of these enzymes, together with ALT, were previously associated with HCC [20]. Nonetheless, ALT is commonly accepted as a surrogate for other liver enzymes [70, 71]. Another factor that may have affected the statistical power of our analysis, is the small size of the HCC group. Therefore, future studies incorporating comprehensive clinical, biochemical, and molecular data and larger sample size are warranted to provide a more nuanced understanding of the factors associated with HCC in HBV-infected individuals.
Anzeige
Additionally, due to the design of the study it is unclear whether the mutations contributed to the risk of developing HCC as it is unknown which event preceded the other. Longitudinal studies with sufficient follow-up time are required to establish causality. If these mutations or some of them emerged following HCC, it is intriguing to know whether or not they confer viral fitness advantage or are associated with unfavorable prognosis in HCC patients.
Our evolutionary and structural analysis indicated that the mutations may confer advantageous properties and influence the functionality of the HBsAg protein. Notably, all three mutation sites showed signatures of positive selection specifically within the HCC cohort, indicating potential selective pressures driving their prevalence in this clinical context. It is important to note, however, that our analysis also indicated that some of these sites might be under relaxed purifying selection due to strong selection in the overlapping P ORF. Nonetheless, it is difficult to determine which gene is deriving the selection at a certain site based on the applied analysis. Additionally, it is possible that a mutation emerging as a result of the function of one gene can also affect the function of the overlapping gene. For example, it was previously reported that an anti-viral resistant mutation in the polymerase also disrupted an HBsAg epitope and caused immune evasion [72, 73]. Further studies are required to decipher the role of genetic variations affecting overlapping genes on the evolution of HBV and their potential contribution to progressive liver diseases.
The Structural analysis also revealed that position 103, in particular, exhibits high accessibility, implying that mutations at this site could perturb immune responses as it resides within previously reported B- and T-cell epitopes [25‐28]. Indeed, the N103D mutation was previously observed at a significantly higher frequency in patients with HBV mono-infections compared to those co-infected with HIV, suggesting a potential role in immune evasion [74]. Additionally, the mutation may interfere with the interactions with viral/cellular factors. Both the viral nucleocapcid binding site and the host factor Hsc70 binding site overlap with the region harboring this mutation [8, 23, 24].
The other two mutations, Q30K and I208T, are located in regions that were found to affect the secretion and assembly of viral particles and both were previously detected in patients with occult hepatitis B infections (OBI) [75, 76]. Notably, position 30 exclusively harboured glutamine residues in all LC sequences analyzed in our study. This observation raises intriguing possibilities regarding the association between specific amino acids at this position and disease progression. It is plausible that Q30K was only detected in LC patients who subsequently progressed to HCC as some HCC patients may have pre-existing liver cirrhosis. This may suggest that Q30K emerged before HCC development and may increase the risk for HCC in LC compared to non-LC patients.
While our study suggests a potential functional impact of the mutations N103D, Q30K, and I208T, it is essential to acknowledge that the true impact, including their impact on the polymerase function, can only be definitively confirmed through empirical validation and comprehensive analysis. Secondly, the structural analysis conducted in our study relied on predicted models, which may not accurately represent the true conformation of the protein, especially in regions characterized by flexibility, such as loop regions. Thus, caution should be exercised when extrapolating structural implications based solely on predicted models. Lastly, the sequences analyzed in our study were derived from dated samples, potentially limiting the generalizability of our findings to the current landscape of HBV strains circulating in the region. Therefore, future studies incorporating more recent samples and addressing these limitations are warranted to further elucidate the potential role of HBsAg mutations in HCC development or prognosis accurately.
Conclusion
In conclusion, this study explored the variation in HBsAg in four clinical groups and identified mutations significantly associated with progressive liver diseases. Evolutionary and structural analyses further underscored the functional significance of these mutations, indicating potential selective pressures and alterations in protein interactions. However, it is crucial to interpret these findings cautiously, recognizing the inherent limitations of the study design and methodology.
Acknowledgements
The authors would like to acknowledge the support of the administration of the research centre at King Faisal Specialist Hospital and Research Centre.
Declarations
Ethics approval and consent to participate
The study was conducted in accordance with the Declaration of Helsinki (1975), and approved by the Institutional Review Board of King Faisal Specialist Hospital and Research Centre (project number: 2150008, date of approval 6 September 2015). Informed consent was obtained from all subjects involved in the study.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
