TOX defines the degree of CD8+ T cell dysfunction in distinct phases of chronic HBV infection

Study cohort

Blood samples were obtained from patients with viral hepatitis recruited at the Department of Medicine II of the University Hospital Freiburg, Germany and at the Department of Gastroenterology, Hepatology and Endocrinology, Hannover Medical School, Hannover, Germany. Written informed consent was given by all donors prior to blood donation. Patients were classified into different clinical phases of cHBV infection according to European Association for the Study of the liver (EASL) guideline of 2017,¹⁷ which considers the presence of HBeAg, HBV DNA levels, transaminase levels (ALT and aspartat aminotransferase (AST)), and the presence or absence of liver inflammation. Nulcelos(t)ide analogues (NUC)-treated patients were included in the HBeAg- CHB patient cohort if they demarcate the characteristics of a CHB before initiation of antiviral therapy with NUC. Characteristics of included patients are detailed in online supplemental SI tables 1–9.

Peptides and tetramers

Peptides of immunodominant HLA-A*01:01-, HLA-A*02:01-, HLA-A*11:01-restricted HBV-derived epitopes (HBVcore₁₈: FLPSDFFPSV, HBVcore₃₀: LLDTASALY, HBVcore₁₄₁: STLPETTVVRR, HBVpol₄₅₅: GLSRYVARL), HLA-A*02:01- restricted HCV-derived epitopes (NS3₁₀₇₃: CINGVCWTV, NS3₁₄₀₆: KLSGLGLNAV) as well as CMV-derived epitope (pp65₄₉₅: NLVPMVATV), EBV-derived epitope (BMFL1₂₈₀: GLCTLVAML) and influenza virus (FLU)-derived epitope (M1₅₈: GILGFVFTL) were synthesised from Genaxxon, Germany. HLA class I epitope-specific tetramers were generated through conjugation of biotinylated peptide/HLA class I monomers with PE-conjugated streptavidin (ProZyme, USA) at a peptide/HLA I:streptavidin molar ratio of 5:1. Of note, targeted epitopes of virus-specific CD8+ T cells were previously analysed for viral sequence mutations as described in the online supplemental material. T cell responses of patients harbouring viral sequence mutations in the targeted epitope were excluded.

Peptide/HLA class I tetramer-based magentic bead enrichment

Enrichment of virus-specific CD8+ T cells was carried out as previously described^{18 19} and their ex vivo frequency was calculated according to Alanio et al.¹⁸ Afterwards, the enriched virus-specific CD8+ T cells were used for multiparametric flow cytometry analyses as described in the online supplemental material.

Expansion of virus-specific CD8+ T cells and assessment of CD8+ T-cell function

Functional analyses of virus-specific CD8+ T cells were performed after 14 days of in vitro expansion with epitope-specific peptides as described in the online supplemental material.

Statistics

The statistical analyses were conducted with GraphPad Prism V.8 (GraphPad Prism Software, USA) and R studio (R studio, USA). Bar charts show median values with IQR. Statistical tests used are depicted in the figure legends. Levels of significance are indicated as follows: *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.

Differential TOX expression in HBV-specific CD8+ T cells is linked to the clinical phase of infection

In a first set of experiments, we analysed the expression of TOX in HLA-A*02:01-restricted HBV-and HCV-specific CD8+ T cells in 46 chronically HBV-infected and 12 chronically HCV-infected patients. To increase the detection rate of virus-specific CD8+ T cells ex vivo, we performed peptide/HLA-A*02:01 tetramer-based enrichment as previously reported^{19 20} using two HBV epitope-specific and two HCV epitope-specific HLA-A*02:01 tetramers. As shown in representative plots and summarised for all analysed subjects in figure 1A, we found a significant TOX expression in HBV-specific CD8+ T cells that was, however, lower compared with HCV-specific CD8+ T cells. We next addressed whether the high variability in the TOX expression of HBV-specific CD8+ T cells (figure 1A) is associated with viral load by performing correlation analysis. As shown in figure 1B, we indeed found a moderate positive correlation between viral load and TOX expression suggesting that antigen levels may drive TOX expression in HBV-specific CD8+ T cells. Since viral loads and viral replication are also reflected by the different clinical phases of HBV infection, we also compared TOX expression in HBV-specific CD8+ T cells obtained from patients in different clinical phases of infection, specifically from acutely infected patients and patients with HBeAg+ or HBeAg- CHB, HBeAg- cHBV infection and acute-resolved versus chronic-resolved HBV infection. Characteristics of these patients are summarised in table 1. We observed a differential TOX expression in these patient groups with the highest expression in acutely infected patients and patients with HBeAg+ or HBeAg− CHB, followed by HBeAg- cHBV infection (figure 1C). TOX expression was lowest in subjects after acute-resolving infection (figure 1C), indicating that enhanced TOX expression is linked to antigen stimulation of HBV-specific CD8+ T cells. Interestingly, HBV-specific CD8+ T cells obtained after resolution of cHBV infection (defined by HBV surface antigen (HBsAg) seroconversion) still displayed high TOX expression (figure 1C). This is also shown in figure 1D in one patient who was followed longitudinally. Indeed, although expression of TOX decreased slightly after resolution of chronic infection, it clearly remained positive suggesting that the chronic imprint is maintained even after antigen elimination. To further address the impact of ongoing HBV viremia on TOX expression of HBV-specific CD8+ T cells, we longitudinally analysed the HBV-specific CD8+ T cell response in four patients after initiation of antiviral therapy with NUC. As shown in figure 1E for one representative CD8+ T cell response and for all four tested HBV-specific CD8+ T cell populations, TOX expression in HBV-specific CD8+ T cells was slightly decreasing after initiation of antiviral therapy in concert with a significant drop of viremia to undetectable levels. However, the expression was still clearly detectable and higher compared with acute-resolved infection (figure 1C) further indicating that TOX expression is maintained in chronically stimulated CD8+ T cells even after elimination of viral replication.

TOX expression is linked to phenotypic characteristics of T cell exhaustion

As TOX has been recently reported to be a marker for T cell exhaustion, we next comprehensively analysed the coexpression of TOX with other known T cell surface markers indicative of exhaustion in chronic infection (PD1, KLRG1 and CD57).^{15 21} As shown in figure 2A for one representative patient in a histogram and for up to 46 chronically HBV-infected patients analysed, we found a stronger expression of all three markers in TOX+ versus TOX- HBV-specific CD8+ T cells. Next, we analysed whether the expression of TOX is also linked to a transcription factor pattern that has previously been reported to characterise T cell exhaustion such as high EOMES and Helios expression.^{22 23} These analyses revealed a coexpression of TOX with both, EOMES^hi and Helios (figure 2B). We also found a higher expression of T cell factor 1 (TCF-1) in TOX+ versus TOX- HBV-specific CD8+ T cells (figure 2C). TCF-1 has previously been reported to be central for the maintenance of the virus-specific CD8+ T cell response in chronic viral infection.^24–26 In contrast, the antiapoptotic molecule BCL-2 was increased in TOX- versus TOX+ HBV-specific CD8+ T cells (figure 2C). This finding is in line with a higher survival capacity of not terminally exhausted TOX- HBV-specific CD8+ T cells. Next, we performed a comparative evaluation of the phenotypes of TOX+ versus TOX- HBV-specific CD8+ T cells. Specifically, we concatenated flow cytometry data obtained from TOX+ and TOX- HBV-specific CD8+ T cells and conducted t-distributed stochastic neighbour embedding (t-SNE) analysis (figure 2D). A distinct topographical clustering of TOX+ versus TOX- HBV-specific CD8+ T cells was observed based on markers indicative of exhaustion and memory formation (figure 2D). This supports the hypothesis that TOX expression segregates phenotypic distinct populations of HBV-specific CD8+ T cells with a polarisation of TOX+ HBV-specific CD8+ T cells towards terminal differentiation/more severe exhaustion. We, therefore, asked whether TOX is also differentially expressed in the distinct subsets of exhausted HBV-specific CD8+ T cells. We have previously reported that HBV-specific CD8+ T cells in chronic infection are not homogeneous but rather consist of heterogeneous subsets of primarily less differentiated progenitor/memory-like cells that are CD127+PD1+ and further differentiated/more severely exhausted CD127-PD1+ cells.²⁰ As shown in online supplemental SI figure 1, the CD127/PD1 subset distribution of HBV-specific CD8+ T cells differ in patients within different clinical phases of HBV infection. Indeed, we found a higher frequency of the CD127-PD1+ population in patients with HBeAg+ CHB and HBeAg− CHB compared with NUC-treated HBeAg− CHB and HBeAg− cHBV infection patients. Remarkably, this more severely exhausted population is not reduced in patients with chronic-resolved HBV infection. In comparison to CD127+PD1+ HBV-specific CD8+ T cells, a higher TOX expression was found within the terminally exhausted CD127-PD1+ HBV-specific CD8+ T cell population (figure 2E). Thus, we next asked whether the different subset distribution may explain differential TOX expression of HBV-specific CD8+ T cells in different clinical phases of infection. Indeed, we identified a clear correlation between CD127-PD1+ HBV-specific CD8+ T cells and the frequency of TOX-expressing HBV-specific CD8+ T cells in chronically HBV-infected patients, whereas an inverse correlation was found between the TOX expression and frequency of CD127+PD1+ HBV-specific CD8+ T cells (figure 2F). Combined, these results indicate that the level of TOX expression is associated with the degree of exhaustion in cHBV infection.

Differential TOX expression in HBV pol₄₅₅-versus HBV core₁₈-specific CD8+ T cells

Previously, we and others have reported that HBV-specific CD8+ T cells targeting core versus polymerase epitopes significantly differ in frequency, phenotype and function.^{20 27} In particular, in comparison to HBV core-specific CD8+ T cells, a higher frequency of HBV polymerase-specific CD8+ T cells expressed markers indicative of more severe T cell exhaustion. Therefore, we compared TOX expression of HBV core₁₈-specific and HBV pol₄₅₅-specific CD8+ T cells in our cohort of chronically HBV-infected HBeAg+ and HBeAg– patients. As shown in figure 3A and in agreement with our previous study,²⁰ we were able to detect HBV core₁₈-specific CD8+ T cells in a significantly higher frequency compared with HBV pol₄₅₅-specific CD8+ T cells. More importantly, we found a significantly higher TOX expression in HBV pol₄₅₅-specific CD8+ T cells compared with HBV core₁₈-specific CD8+T cells (figure 3B). To rule out that the observed differences are only specific for a given epitope, we performed additional analyses with the HLA-A*01:01-restricted HBV core₃₀ and the HLA-A*11:01-restricted HBV core₁₄₁ epitopes but could not detect specific differences between the TOX expression of different HBV core epitope-specific CD8+T cells (online supplemental SI figure 2) Overall, these data are in line with our previous assumption that HBV core epitope-specific CD8+ T cell responses are less severely exhausted compared with HBV pol₄₅₅-specific CD8+ T cells.

TOX is associated with impaired functionality of HBV-specific CD8+ T cells

A hallmark of T cell exhaustion is impairment of effector functions. Therefore, we next assessed whether expression of TOX in HBV-specific CD8+ T cells is also linked to altered effector functions. Since the frequency of HBV-specific CD8+ T cells is too low for functional analyses ex vivo, we determined the expansion capacity in a 2-week peptide-specific stimulation in vitro (figure 4A). Of note, we have previously shown that ex vivo functionality correlated with the functional capacity of HBV-specific CD8+ T cells after in vitro expansion.²¹ As shown in figure 4B for one representative patient, we observed a significant expansion of HBV-specific CD8+ T cells on day 14 that were hardly detectable on baseline (day 0). Next, we correlated the expansion capacity (reflected by frequency and expansion index) of each HBV peptide-specific CD8+ T cell population with their respective TOX expression ex vivo. This analysis revealed an inverse correlation between TOX expression and the expansion capacity, for example, the higher TOX is expressed the lower is the expansion capacity of HBV-specific CD8+ T cell populations (figure 4B). We also found a similar inverse correlation between TOX expression and HBV-specific cytokine production (interferon-γ (IFN-γ) and tumour necrosis factor (TNF)) (figure 4C and online supplemental SI figure 3A) and degranulation as indicated by CD107a expression (figure 4D and online supplemental SI figure 3A). Thus, these results support that TOX not only marks cells with an exhaustion phenotype but is indeed also clearly associated with functional impairment.

Next, we asked whether these biological relevant associations of TOX can also be translated to the situation in vivo. For this, we compared the expression of TOX in HBV-specific CD8+ T cells in two different groups of chronically HBV-infected patients with low viral load; specifically, in HBeAg- patients who controlled viremia either by antiviral treatment (NUC-treated HBeAg- CHB cohort) or endogenously (HBeAg- cHBV infection cohort) (figure 4E and online supplemental SI figure 3B). In line with previous reports,⁴ we observed better effector functions, for example, expansion and TNF production, of HBV-specific CD8+ T cells in patients that endogenously controlled HBeAg- cHBV infection compared with patients with therapy-mediated viral control (figure 4F). This higher functionality of HBV-specific CD8+ T cells from patients with endogenously controlled HBeAg- cHBV infection was associated with a lower TOX expression compared with NUC-treated patients with HBeAg- CHB (figure 4G). These results clearly link lower TOX expression with endogenous control of HBV infection and support its biological relevance in the efficiency of HBV-specific CD8+ T cell responses.

TOX expression is not generally associated with CD8+ T-cell dysfunction

Next, we asked whether TOX selectively marks dysfunctional virus-specific CD8+ T cells in the context of an actively persisting viral infection such as cHBV infection or whether it is also expressed in other virus-specific CD8+ T cells in humans. We, therefore, first compared the expression of TOX in virus-specific CD8+ T cells targeting epitopes of actively persisting/cHBV (core₁₈, pol₄₅₅) versus latently persisting CMV (pp65₄₉₅) or EBV (BMLF-1₂₈₀) and self-limiting FLU (M₁₅₈). As depicted in figure 5A, TOX is expressed in similarly high frequencies of HBV-specific, CMV-specific and EBV-specific CD8+ T cells but only in a minor fraction of FLU-specific CD8+ T cells. Thus, these results support that TOX expression is associated with persistent/prolonged antigen recognition in general. Since TOX has been identified as master regulator within the differentiation programme of T cell exhaustion in the context of persisting antigen recognition,^{13 14 28} we next addressed the question whether the similar TOX expression in virus-specific CD8+ T cells targeting persisting viruses is associated with similar CD8+ T cell differentiation. For this, we tested TOX expression in naïve (T_N), central memory (T_CM), effector memory (T_EM) and terminally differentiated (T_EMRA) T-cell differentiation stages among HBV-specific, CMV-specific, EBV-specific and FLU-specific CD8+ T cells using CCR7 and CD45RA expression for discrimination (online supplemental SI figure 4A). Highest TOX expression was detectable among terminally differentiated T_EMRA stages (figure 5B) again indicating a role of repetitive/ongoing antigen recognition on the induction of TOX. Indeed, TOX expression also correlated with the frequency of T_EMRA cells in CMV-specific, EBV-specific and FLU-specific CD8+ T cells, however, this was not the case for HBV-specific CD8+ T cells (figure 5B). This clearly suggests that TOX expression in HBV-specific CD8+ T cells is not linked to terminally differentiated cells. Consequently, TOX expression in HBV-specific CD8+ T cells is not linked to the same differentiation stage compared with CMV-specific and EBV-specific CD8+ T cells and thus marks different characteristics although TOX expression is similarly high. This is also supported by coexpression analysis of TOX with other transcription factors that are relevant for T cell differentiation, for example, TCF-1, EOMES and T-bet (figure 5C and online supplemental SI figure 4B). In particular, in HBV-specific CD8+ T cells TOX and TCF-1 are highly coexpressed in contrast to CMV-specific, EBV-specific and FLU-specific CD8+ T cells while coexpression of TOX and T-bet or EOMES is reduced indicating different transcription factor networks of TOX in HBV-specific versus CMV-specific, EBV-specific and FLU-specific CD8+ T cells. This is supported in a more comprehensive way by t-SNE and by multidimensional scaling analysis of flow cytometric data from TOX+ HBV-specific, CMV-specific and EBV-specific CD8+ T cells, both showing distinct clustering of HBV-specific versus CMV-specific and EBV-specific CD8+ T cell populations (figure 5D). Of note, in contrast to CMV-specific and EBV-specific CD8+ T cells, HBV-specific CD8+ T cells were characterised by a cluster of markers that are associated with T cell exhaustion (online supplemental SI figure 4C). To further elaborate on the different characteristics marked by TOX expression, we next correlated the functional capacity with the TOX expression of CMV-specific, EBV-specific and FLU-specific CD8+ T cells. In contrast to HBV-specific CD8+ T cells (figure 4B–D), neither the expansion factor (figure 5E) nor cytokine production or degranulation (online supplemental SI figure 4D) were associated with TOX expression in CMV-specific, EBV-specific and FLU-specific CD8+ T cells suggesting that TOX expression indicates dysfunction of virus-specific CD8+ T cells only in actively persisting infections such as cHBV infection.

Patient consent for publication

Not required.

Ethics approval

In agreement with the federal guidelines, local ethics committee regulations of Albert-Ludwigs-University, Freiburg, Germany (HBUF 474/14) and Hannover Medical School (3374–2016) and the Declaration of Helsinki (1975).