Distinct phenotype and function of circulating Vδ1+ and Vδ2+ γδT-cells in acute and chronic hepatitis B

Hepatitis B virus (HBV) persists with global and virus-specific T-cell dysfunction, without T-cell based correlates of outcomes. To determine if γδT-cells are altered in HBV infection relative to clinical status, we examined the frequency, phenotype and function of peripheral blood Vδ1+ and Vδ2+γδT-cells by multi-parameter cytometry in a clinically diverse North American cohort of chronic hepatitis B (CHB), acute hepatitis B (AHB) and uninfected control subjects. We show that circulating γδT-cells were comprised predominantly of CD3hiCD4- Vδ2+γδT-cells with frequencies that were 2–3 fold higher among Asian than non-Asian Americans and inversely correlated with age, but without differences between CHB, AHB and control subjects. However, compared to control subjects, CHB was associated with increased TbethiEomesdim phenotype in Vδ2+γδT-cells whereas AHB was associated with increased TbethiEomesdim phenotype in Vδ1+γδT-cells, with significant correlations between Tbet/Eomes expression in γδT-cells with their expression of NK and T-cell activation and regulatory markers. As for effector functions, IFNγ/TNF responses to phosphoantigens or PMA/Ionomycin in Vδ2+γδT-cells were weaker in AHB but preserved in CHB, without significant differences for Vδ1+γδT-cells. Furthermore, early IFNγ/TNF responses in Vδ2+ γδT-cells to brief PMA/Ionomycin stimulation correlated inversely with serum ALT but not HBV DNA. Accordingly, IFNγ/TNF responses in Vδ2+γδT-cells were weaker in patients with CHB with hepatitis flare compared to those without hepatitis flares, and this functional deficit persisted beyond clinical resolution of CHB flare. We conclude that circulating γδT-cells show distinct activation and differentiatiation in acute and chronic HBV infection as part of lymphoid stress surveillance with potential role in clinical outcomes.


Introduction
Hepatitis B virus (HBV) is an important human pathogen with a global impact in morbidity and mortality. As HBV is generally non-cytopathic, liver disease pathogenesis is largely immune-mediated with rapid progression to cirrhosis and cancer in some and minimal disease progression in others [1][2][3]. A critical role for conventional T-cells in viral clearance and liver disease has been shown in animal models and inferred in patients [4][5][6][7][8][9]. However, in patients with chronic hepatitis B (CHB), both HBV-specific and global T-cells are functionally suppressed due to continued antigenic stimulation, inflammation and the induction of multiple regulatory pathways [4][5][6][7][8][9][10][11][12]. Furthermore, there are no distinct T-cell based immune signatures for the dynamic clinical and virological phases of CHB [10]. These findings also raised the possiblity for alternate mechanisms beyond conventional T-cells in CHB pathogenesis.
Relevant for microbial pathogenesis, increased Vδ2 + γδT-cell frequency has been reported in infections with intracelluar pathogens such as mycobacterium tuberculosis, salmonella or malaria [17]. A protective role for γδT-cells in viral infections has been suggested by: increased Vδ2 + γδT-cells in elite HIV controllers [36], reduced Vδ2 + γδT-cell frequency associated with poor HIV-related outcomes [37][38][39], and Vδ1 + γδT-cell expansion associated with the resolution of cytomegalovirus (CMV) infection [40][41][42]. A role for γδT-cells in HBV immunobiology was first suggested in experimentally HBV-infected chimpanzees with hepatic induction of γδTCR-associated genes leading to viral clearance [43]. Findings in patients with CHB have been more conflicting [44][45][46][47][48], with reduced Vδ2 + γδT-cell frequencies in one study [44] but not another [48]. Both pathogenic and regulatory roles have been suggested by enhanced cytolytic potential reported in patients with HBV-associated acute-on-chronic liver failure [46] and CD8 T-cell exhaustion by inducing myeloid-derived suppressor cells in a mouse model [49].
Here, we compared the frequency, phenotype and function of circulating γδT-cells in human subjects with CHB relative to those with acute hepatitis B (AHB) and uninfected normal control subjects (NC), initially hypothesizing that their frequency and/or function will be reduced in CHB with increased disease activity. Contrary to our hypothesis, we found that γδT-cells are preserved in circulating frequencies in CHB and AHB, with distinct innate-like phenotype and/or effector function compared to uninfected control subjects. In particular, compared to uninfected controls, AHB subjects showed weaker IFNγ/TNF responses in Vδ2 + γδT-cells that improved with the resolution of AHB whereas CHB subjects showed with preserved IFNγ/TNF responses in Vδ2 + γδT-cells. Furthermore, in CHB, IFNγ/TNF responses to brief PMA/Ionomycin stimulation in Vδ2 + γδT-cells correlated inversely with serum alanine aminotransferase (ALT), with persistent deficit in patients with hepatitis ALT flares compared to those without flares. These findings provide new insights to γδT-cells with relevance to HBV pathogenesis.

Study subjects
Adults with CHB and acute hepatitis B (AHB) were enrolled into the Hepatitis B Research Network (HBRN) Adult Cohort Study sponsored by the National Institute for Diabetes and Digestive and Kidney Diseases (NIDDK), as previously described [50]. Subjects with hepatic decompensation, liver cancer, liver transplant, current hepatitis B antiviral therapy, known HIV co-infection and inability or unwillingness to attend follow-up visits were excluded. A subset of HBRN participants (215 CHB, 12 AHB) were recruited into the ancillary Immunology Study with additional informed consent for immunology blood draws from the following clinical centers: Toronto (University of Toronto), Dallas (University of Texas Southwestern), San Francisco (University of California San Francisco, California Pacific Medical Center); Richmond (Virginia Commonwealth University); Seattle (Virginia Mason Medical Center and University of Washington); Minnesota (University of Minnesota and Mayo Clinic), Boston (Beth Israel Deaconness Medical Center, Massachusetts General Hospital) and Chapel Hill (University of North Carolina), as previously described [10]. Among 215 CHB subjects initially enrolled into the Immunology Study, we included 189 in this study, excluding 10 who were found to be on antiviral therapy at the time of immunology blood draw and 16 without concurrent serum alanine aminotransferase (ALT) level.
Thirty-four HBsAg-negative normal control subjects (NC) were recruited from the HBRN Immunology Center in Philadelphia (University of Pennsylvania and Corporal Michael J. Crescenz Veterans Affairs Medical Center), Toronto and Dallas, including 29 with a history of prior HBV vaccination. NC subjects were recruited based on the absence of known liver disease, autoimmune disease, immunosuppression and active medical conditions that preclude large volume research blood draws. They were also negative for serum HBsAg, antibody to hepatitis C virus (anti-HCV) and antibody to human immunodeficiency virus (anti-HIV).
As shown in Table 1A, CHB, NC and AHB groups did not differ significantly in age or sex distribution, although the CHB group showed a marked Asian predominance consistent with the overall HBRN adult cohort [10,50]. Asians with CHB showed higher HBV DNA levels with higher prevalence of positive hepatitis B e antigen (HBeAg) status and infection with HBV genotypes B or C, compared to Non-Asians with CHB (Table 1B). As for liver biopsy

Ethics statement
The study received approval by the institutional review board or equivalent committee(s) for each of the centers participating in patient recruitment as stated above and previously described [10]:

Clinical grouping of HBV-infected subjects
Clinical phase of each HBV-infected subject was assigned based on medical history and laboratory results as previously described as immune tolerant (IT), HBeAg + immune active (IA+), HBeAgimmune active (IA-) or inactive carrier (IC) status [10]. ALT levels were normalized by the upper limits of normal (ULN) for ALT (30 U/L for men and 20 U/L for women) as ALT/ULN ratio. ALT flare was defined by ALT/ULN �10 (ALT 300 U/L for males, 200 U/L for females). Subjects identified with an ALT flare were asked to return for additional immunology blood draw within 1-4 weeks of meeting criteria for ALT flare and again within 12-24 weeks from initial hepatitis flare or before starting antiviral therapy if not resolved. Acute hepatitis B (AHB) was defined by acute ALT elevation in the presence of HBsAg and IgM antibody to hepatitis B core antigen (IgM anti-HBc) without a previous history of HBsAg positivity. The Immunology Center personnel were blinded to clinical parameters while conducting immune assays.

Peripheral blood mononuclear cells (PBMC)
PBMC were isolated from blood drawn in lavender-top plasma tubes (ethylenediaminetetraacetic acid or EDTA as additive) using Ficoll-Histopaque (Sigma Chemical Co., St Louis, MO) density centrifugation and resuspended in complete media with 10% human male AB serum as described [10]. PBMC isolation was performed within 24 hours of blood draw.

Immunophenotyping
Cells were stained with fluorescent antibodies according to the manufacturer's instructions, acquired by FACSCanto (BD Biosciences, San Jose, CA) and analyzed with FlowJo (Tree Star Inc., San Carlos, CA). Magnetic beads coated with anti-mouse antibodies were used for compensation calculation (BD Biosciences, San Jose, CA) as described previously [51,52]. CD3 hi CD4 -T-cell frequency and phenotype in 189 CHB (including 39 with hepatitis flare), 34 NC and 12 AHB participants were examined by staining freshly isolated PBMC with a multi-parameter screening T-cell panel that included fluorescent antibodies for CD3, CD4, CD8, PD-1, CTLA-4, CD28 and CD127 [10]. More detailed γδT-cell analyses were performed with secondary FACS panels in 70 subjects with available cryopreserved PBMC (36 CHB, 27 NC and 7 AHB participants) with the inclusion of CD16, CD56, CD161, NKG2D, NKG2A, CD158a and/or CD94. Among the 36 CHB subjects, 14 with ALT flares (CHB-F) were first examined within 1-4 weeks of ALT flare. PBMCs from a second time point (at least 2 months from the initial time point) were also analyzed in 7 CHB-F and in 7 AHB subjects. Examination for γδTCR subtype (Vδ1, Vδ2, Vγ9), phenotype (T/NK markers), effector molecules (perforin, granzyme B) and transcription factor expression (Tbet, Eomes) was conducted directly ex-vivo. In select cases with available lymphocytes (11 CHB, 7 NC), PBMC were stained with metal-conjugated antibodies and acquired by CyTOF II as described [53,54], for markers including perforin, granzyme, PD1, CD38, HLA DR, Ki67, TIGIT and Tim3. As Vδ2 + γδTcells (but not Vδ1 + γδT-cells) co-expressed Vγ9 TCR with a distinct CD3 hi CD4phenotype, Vδ2 + γδT-cells were defined by the expression of Vδ2 or Vγ9 TCR as well as CD3 hi CD4phenotype based on available antibodies in our FACS panels.

Statistical analyses
Patient characteristics were compared using Pearson chi-square or Fisher's exact test for discrete variables such as sex, race and genotype, and Kruskal Wallis test for continuous variables such as HBV DNA and ALT. Immune measures between distinct patient groups were compared using Mann-Whitney U (for two groups) and Kruskal-Wallis test (for more than two groups). When comparing 3 or more groups, first comparison was made with Kruskal Wallis, followed by further comparisons between 2 groups by Mann Whitney U if the initial Kruskal Wallis yielded p-value < 0.05. Correlated samples (e.g. immune measures within the same cells) were compared using matched pair signed-rank test. Correlation between two immune measures was assessed using non-parametric Spearman correlation and the corresponding test. Proportions across two groups were compared by Fisher's exact test and across more than two categories by chi-square test or exact chi-square tests as appropriate. P values below 0.05 were considered statistically significant.
As for another host factor, age showed significant inverse associations with Vδ2 + and CD3 hi CD4 -(but not Vδ1 + ) γδT-cell frequencies (Fig 1G) consistent with lower γδT-cell frequencies reported in older persons [59,60]. There were no differences between males or females (median %CD3 hi CD4 -/CD3: males 2.4% vs females 2.7%, p = .80). Finally, there were no significant correlations between γδT-cell frequencies and serum HBV DNA or alanine aminotransferase (ALT) levels in CHB subjects (Fig 1H). Thus, circulating γδT-cell frequencies were not altered by HBV infection in our study, although Vδ2 + γδT-cell frequency was significantly impacted by host factors such as race/ethnicity and age.

Circulating γδT-cells display an innate phenotype with the expression of both T and NK regulatory markers that are distinctly altered in AHB and CHB
Consistent with innate characteristics reported for γδT-cells [15], γδT-cells were more enriched in expression of NK markers compared to total CD3 + T-cells. For example, Vδ2 + γδT-cells expressed more CD56 and CD16 compared to total CD3 + T-cells, without significant differences between CHB, NC and AHB groups (Fig 2A, left panel). Vδ1 + γδT-cells also expressed more CD56 and CD16 than total CD3 + T-cells in AHB group compared to NC and CHB groups. As shown in Fig 2A (right panel), CD161, a C-type lectin expressed in NK cells and liver-homing T-cells with Th17/Th1 phenotype [61][62][63][64], was most enriched in Vδ2 + γδTcells compared to other subsets, but without significant differences between CHB, NC and AHB groups.
Innate phenotype of Vδ2 + γδT-cells was further confirmed in CD3 hi CD4 T-cells with significantly greater expression of NK-associated markers (e.g. CD56, NKG2A, CD94) but lower expression of T-cell associated markers (e.g. PD1, CD28), when compared to CD3 int CD4 -Tcells (Fig 2B, top panel). Furthermore, compared to NC subjects, CHB subjects showed significantly lower expression of PD-1 and CD158a (inhibitory killer immunoglobulin-like receptor 2DL1 or KIR2DL1) but greater expression of inhibitory NKG2A and CD94 in CD3 hi CD4 -Tcells, (Fig 2B, bottom panel). Further comparison of CHB and NC subjects with available lymphocytes confirmed reduced expression of PD1 as well as other activation and/or co-inhibitory markers including CD38, Ki67 and Tim3 in Vδ2 + γδT-cells from CHB compared to NC subjects, without such differences for Vδ1 + γδT-cells (Fig 2C). Collectively, these findings highlight distinct innate-like phenotype and activation of γδT-cells, with novel features of Vδ1 + γδT-cells in AHB (increased CD56 and CD16) and Vδ2 + γδT-cells in CHB (increased NKG2A/CD94 and reduced PD1, CD38, Ki67, Tim3 and CD158a,) compared to NC subjects.

Fig 1. Circulating γδT-cells are comprised predominantly of CD3 hi CD4 -Vδ2 + γδT-cells with frequencies that are impacted by host race/ethnicity and age but not HBV infection. A. Comparison of circulating γδT-cell frequencies relative to HBV infection. Bar graphs show median %γδT-cells in CD3
+ T-cells from 36 chronic hepatitis B (CHB), 27 uninfected normal controls (NC) and 7 acute hepatitis B (AHB) subjects determined by flow cytometry: %Vδ1 + (blue bar), %Vδ2 + (red bar) and % pan-γδTCR + (violet bar). Error bars indicate 25% and 75% interquartile ranges. Differences between %Vδ1 + and %Vδ2 + γδT-cells were examined by non-parametric Mann-Whitney U. P-values below 0.05 were considered statistically significant. B. Representative FACS appearance of γδT-cells. Multi-color FACS analysis of peripheral blood mononuclear cells (gated on CD3 + viable singlet lymphocytes) show that Vδ2 + cells co-express Vγ9 TCR and pan-γδTCR but not Vδ1 TCR. Overlay of Vδ1 + (orange box gate) and Vδ2 + cells (green box gate) onto lymphocytes (blue background) shows the CD3 hi CD4phenotype of Vδ2 + cells (green dots on the overlay) whereas Vδ1 + cells (orange dots on the overlay) show CD3 int CD4appearance. C. Enrichment of CD3 hi CD4 -T-cells for Vδ2 + γδT-cells. Gating strategy and distinct FACS appearance are shown for highly CD3-positive but CD4-negative cells (CD3 hi CD4 -) compared to CD3-intermediate CD4cells (CD3 int CD4 -). Upper panel shows pseudo-color plots of gated CD3 hi CD4 -T-cells which are markedly enriched for Vδ2 + γδT-cells but not Vδ1 + γδT-cells, αβTCR + conventional T-cells or CD1d-reactive NKT-cells. Bottom panel shows histogram overlays for CD3 hi CD4 -(red shade) and CD3 int CD4 -(black line) T-cells with enrichment of CD3 int CD4 -T-cells with αβTCR + CD8 + T-cells and not Vδ2 + γδT-cells. D. Correlations between circulating γδT-cell frequencies. Significant correlations were detected between %Vδ2 + γδTcells, %Vγ9 + γδT-cells, and %CD3 hi CD4 -T-cells, but not between %Vδ1 + γδT-cells and %Vδ2 + γδT-cells. E. Comparison of γδT-cell frequencies in CD3 + T-cell compartment between CHB, NC and AHB groups. Vδ1 + and Vδ2 + γδT-cell frequencies were compared between 36 CHB (29 Asians, 7 Non-Asians), 27 NC (12 Asians, 15 Non-Asians) and 7 AHB (7 Non-Asians) subjects with available cryopreserved PBMCs. CD3 hi CD4 -γδT-cell frequencies were compared in 189 CHB (158 Asians, 31 Non-Asians), 34 NC (16 Asians, 18 Non-Asians) and 12 AHB (12 Non-Asians) subjects in freshly isolated PBMCs. Frequency differences between CHB, NC and AHB groups were calculated by non-parametric Kruskal Wallis (k = 3). P-values below 0.05 were considered significant and shown in red font. F. Comparison of Vδ1 + , Vδ2 + and CD3 hi CD4 -γδT-cell frequencies between Asian and Non-Asian American subgroups with and without CHB. Initial comparisons between 4 subgroups (Asian+ CHB, Asian-CHB, Asian+ NC, Asian-NC) were made by non-parametric Kruskal Wallis (k = 4), followed by further 2-group comparisons by Mann Whitney U if initial 4-way comparison showed p-values below 0.05. As shown, Vδ2 + and CD3 hi CD4 -γδT-cell frequencies were greater in Asian Americans compared to Non-Asian Americans within CHB or NC groups, without significant differences between Asian American CHB versus Asian American NC subgroups. Horizontal red lines indicate median values. P-values below 0.05 were considered significant and shown in red font. G. Inverse association between age and circulating Vδ2 + and CD3 hi CD4 -γδT-cell frequencies. Comparisons between age in years (x-axis) and %γδT-cells (y-axis) are shown for NC Asian Americans, (red X), NC Non-Asian Americans (blue X), CHB Asian Americans (red diamond) and CHB Non-Asian Americans (blue diamond), with significant inverse associations between age and % Vδ2 + or %CD3 hi CD4 -γδT-cells but not Vδ1 + γδT-cells. Correlation coefficients and associated p-values for all subjects were determined by non-parametric Spearman rank order correlation test. H. Lack of significant correlations between serum levels of HBV DNA or ALT and circulating γδT-cell frequencies. Serum levels of HBV DNA or ALT on the y-axis are compared to circulating frequencies of Vδ1 + , Vδ2 + and CD3 hi CD4 -γδT-cells, with correlation coefficients and p-values by Spearman rank order correlation test.

Circulating γδT-cells display an innate phenotype with the expression of both T and NK regulatory markers that are distinctly altered in AHB and CHB. A.
Innate phenotype of Vδ1 + avδ/or Vδ2 + γδT-cells compared to total CD3 + T-cells. Bar graphs show median %CD56 + , %CD16 + and %CD161 + in total CD3 + T-cells (gray bars) relative to Vδ1 + γδT-cells (blue bars) or Vδ2 + γδT-cells (red bars) in CHB, NC and AHB groups, with error bars indicating 25% and 75% interquartile ranges. CD56 and CD16 expression levels were examined in 35 CHB, 18 NC and 7 AHB subjects, whereas CD161 expression was examined in 33 CHB, 27 NC and 4 AHB subjects. Expression levels between the cell subsets within individual subject were compared by matched pair signed-rank test. Comparisons between CHB, NC and AHB groups were made with Kruskal Wallis (k = 3). Histogram on the right bottom show overlay of CD3 (gray shade), Vδ1 + γδT-cells (blue line) or Vδ2 + γδT-cells (red line).

Tbet/Eomes expression is enriched in circulating Vδ1 + and Vδ2 + γδT-cells compared to total CD3 + T-cells, with differential hierarchy and phenotypes in CHB and AHB subjects
As Vδ2 + γδT-cells were enriched in the expression of CD161, a marker associated with Th17/ Th1 phenotype [61][62][63][64] }, we examined the expression of Th1 transcription factors Tbet and Eomes as well as Th17 transcription factor RORγt in Vδ2 + γδT-cells relative to those in Vδ1 + γδT-cells and CD3 + T-cells in CHB, NC and AHB subjects. In general, as shown in Fig 3A, Vδ1 + and Vδ2 + γδT-cells were more enriched in Tbet and/or Eomes expression than CD3 + Tcells, whereas RORγt was expressed in very few T-cells. As for reciprocal Tbet and Eomes expression associated with T-cell differentiation and/or exhaustion [65][66][67][68], both Tbet hi Eomes dim and Tbet dim Eomes hi populations were more prominent in γδT-cells than total CD3 + T-cells (Fig 3B). Both CHB and NC groups showed a hierarchy between T-cell subsets with the highest Tbet and/or Eomes expression as well as Tbet hi Eomes dim phenotype in Vδ2 + γδTcells followed by Vδ1 + γδT-cells and CD3 + T-cells. In AHB group, Vδ1 + γδT-cells were most enriched for Tbet + and Tbet hi Eomes dim cells, whereas Vδ2 + γδT-cells were most enriched in Tbet dim Eomes hi cells associated with T-cell exhaustion [65]. Thus, Tbet/Eomes expression patterns were distinctly altered in circulating Vδ1 + and Vδ2 + γδT-cells from AHB and CHB subjects compared to uninfected controls.
As shown in Fig 3C, AHB subjects showed the greatest %Tbet + and %Tbet hi Eomes dim cells in Vδ1 + γδT-cells and least %Tbet + with most Tbet dim Eomes hi cells in Vδ2 + γδT-cells, compared to NC and CHB subjects. By contrast, CHB subjects showed the highest %Tbet + and % Eomes + as well as %Tbet hi Eomes lo cells in Vδ2 + γδT-cells compared to NC and AHB subjects, without such differences for Vδ1 + γδT-cells. Our findings in Vδ2 + γδT-cells were not likely to reflect host race/ethnicity or age, since Tbet/Eomes expression patterns in Vδ2 + γδT-cells (e.g. CHB > NC for %Eomes + ; AHB < NC and CHB for %Tbet + and %Tbet dim Eomes hi ) persisted among Asian and Non-Asian Americans despite reduced statistical significance with smaller sample sizes, and did not correlate with age (S2A and S2B Fig).

B. Increased expression of NK but not T-cell markers in CD3 hi CD4 -T-compared to CD3 int CD4 -T-cells. (Top panel) Bar graphs show median % of cells expressing T-
cell markers (CD127, PD-1, CTLA-4 and CD28) and NK markers (CD56, NKG2D, NKG2A, CD158a, CD94) in CD3 hi CD4 -T-cells (red bars) and CD3 int CD4 -T-cells (white bars) in CHB subjects, with p-values calculated by matched pair signed-rank test. (Bottom panel) Bar graphs compare median % of cells expressing T/NK markers in CD3 hi CD4 -T-cells from CHB (gray bar), NC (white bar) and AHB (black bar) subjects. T-cell markers were measured in 189 CHB, 24 NC and 12 AHB subjects. NK markers were measured in 36 CHB, 17 NC and 7 AHB subjects. Error bars indicate 25% and 75% interquartile ranges. CHB, NC and AHB groups were compared by nonparametric Kruskal Wallis test (k = 3) with further comparisons between 2 groups by Mann Whitney U if the initial Kruskal Wallis test yielded p-values < 0.05. C. Reduced expression of T-cell activation and exhaustion markers in Vδ2 + γδT cells from CHB compared to NC subjects: Bar graphs compare 11 CHB (gray bars) and 7 NC subjects (white bars) for %Vδ1 + γδT cells (top) and %Vδ2 + γδT cells (bottom) that express various T-cell activation or exhaustion markers by CyTOF. Significant pvalues < 0.05 are highlighted in red font. ���� p < .0001; ��� p < 0.0001; �� p < .001. https://doi.org/10.1371/journal.ppat.1007715.g002 Circulating gammadelta T cells in acute and chronic hepatitis B expression in CHB and AHB that further associated with their expression of various NK/T-cell markers, but not with clinical or virological measures in CHB.

Circulating CD3 hi CD4 -Vδ2 + γδT-cells show greater effector capacity compared to Vδ1 + γδT-cells and/or total CD3 + T-cells in-vitro
We next compared the cytokine phenotype and effector capacity of Vδ1 + and Vδ2 + γδT-cells relative to total CD3 + T-cells by well-established intracellular cytokine staining protocol with brief 5 hours of PMA/ionomycin stimulation in-vitro [55][56][57]. As shown in Fig 4A, Vδ2 + γδTcells showed the highest IFNγ response to PMA/Ionomycin in CHB, NC and AHB groups, with a significant hierarchy (Vδ2 + γδT-cells > Vδ1 + γδT-cells > total CD3 + ) in CHB although not other groups. Vδ2 + γδT-cells also showed the highest IFNγ/TNF co-expression and TNF expression in CHB and NC but not AHB group. Greater effector capacity in Vδ2 + γδT-cells was confirmed in CD3 hi CD4 -T-cells compared to total CD3 + T-cells, by greater IFNγ, TNF and MIP1β expression as well as marginal CD107a mobilization (but without IL17 expression) (Fig 4B). Vδ2 + γδT-cells were also more enriched for cytolytic effector molecules perforin and granzyme B compared to Vδ1 + γδT-cells and/or CD3 + T-cells (Fig 4C). Thus, circulating CD3 hi CD4 -Vδ2 + γδT-cells showed greater effector capacity compared to other T-cell subsets, based on IFNγ/TNF responses to PMA/Ionomycin as well as perforin and granzyme B expression.

IFNγ/TNF responses to brief PMA/Ionomycin stimulation are greater and more multi-functional in CD3 hi CD4 -Vδ2 + γδT-cells from CHB compared to NC or AHB subjects
We further compared CHB, NC and AHB groups for IFNγ/TNF expression in γδT-cell subsets upon brief PMA/Ionomycin stimulation. As shown in Fig 5A, IFNγ and/or TNF response to PMA/Ionomycin did not differ between Vδ1 + γδT-cells from CHB, NC and AHB subjects. However, median %IFNγ + TNF + double-positive and %TNF + cells (but not %IFNγ + cells) were significantly higher by almost 2 fold in Vδ2 + γδT-cells from CHB compared to NC and AHB subjects. The findings were similar between CHB and NC subjects for %IFNγ + TNF + doublepositive and %TNF + cells as well as %MIP1β + cells in CD3 hi CD4 -T-cells, without differential

Tbet/Eomes expression is enriched in circulating Vδ1 + and Vδ2 + γδT-cells compared to total CD3+ T-cells with differential hierarchy and phenotypes in CHB and AHB subjects. A. (Left panel) Representative FACS plots showing Tbet, Eomes and RORγt expression in Vδ1 + γδT-cells and Vδ2 + γδT-cells in CD3-gated cells. (Right panel) Comparisons of Tbet, Eomes and RORγt expression between total CD3 + T-cells (white bars), Vδ1 + γδT-cells (blue bars) and
Vδ2 + γδT-cells (red bars). Tbet and Eomes expression in T-cell subsets was examined in 36 CHB, 27 NC and 7 AHB subjects, with further examination of RORγt in 19 CHB, 15 NC and 4 AHB subjects based on PBMC availability. Bar graphs show median values within each group, with error bars indicating 25% and 75% interquartile ranges and p-values calculated by matched pair signed-rank test comparing T-cell subsets within each subject. In CHB group, the highest Tbet and Eomes expression was detected in Vδ2 + γδT-cells followed by Vδ1 + γδT-cells and total CD3 + T-cells. In NC group, similar hierarchy was detected for Tbet expression whereas Eomes expression was elevated in both Vδ1 + and Vδ2 + γδT-cells compared to total CD3 + T-cells. In AHB, Tbet was most elevated in Vδ1 + γδT-cells whereas Eomes was similarly elevated in both Vδ1 + and Vδ2 + γδT-cells compared to total CD3 + T-cells. RORγt was detected in very few cells in all groups, although statistical comparison was not possible for AHB subjects due to insufficient sample size (n = 4). ���� p < 0.00001; ��� p < .0001; �� p < 0.001; � p < .01; #p = 0.008; n.s. sample size not sufficient for statistics for statistical comparison. B. Comparison of Tbet hi Eomes dim and Tbet dim Eomes hi phenotype between total CD3 + T-cells (white bars), Vδ1 + γδT-cells (blue bars) and Vδ2 + γδT-cells (red bars), with representative FACS density plot and gating strategy shown at the bottom for CD3 + T-cells (left) and Vδ2 + γδT-cells (right). ���� p < 0.00001; ��� p < .0001; �� p < 0.001; � p < .01; #p = 0.008. C. Comparison of Tbet/Eomes expression in circulating Vδ1 + and Vδ2 + γδT-cells from 36 CHB, 24 NC and 7 AHB subjects. P-values between 3 groups were determined by Kruskal Wallis test (k = 3), followed by further two-way comparisons by Mann Whitney U for initial p-values below 0.05. Circulating gammadelta T cells in acute and chronic hepatitis B CD107a mobilization or IL17 expression (Fig 5B). These differences in CD3 hi CD4 -Vδ2 + γδTcell function between CHB, NC and/or AHB groups were not due to race or age, as similar patterns persisted among Asian Americans for %IFNγ + TNF + double-positive and %TNF + cells (although not among Non-Asians with smaller sample sizes) and IFNγ/TNF responses did not correlate with age (S2C and S2D Fig).

IFNγ/TNF responses to pAg are preserved in Vδ2 + γδT cells from CHB (but not AHB) subjects and are associated with their expression of Tbet/ Eomes and NK markers but not PD1
While PMA and Ionomycin can provide robust pharmacological activation of multiple immune subsets [57] through protein kinase C and calcium signaling [69], phosphoantigens (pAg) such as zoledronate (Zol) or (E)-4-hydroxy-3-methylbut-2-enyl 4-diphosphate (HMBPP) specifically activate Vδ2 + γδT-cells as their natural ligands [17,58,70]. Based on control experiments in which IFNγ/TNF responses in Vδ2 + γδT-cells were greater with longer pAg stimulation for 23 hours compared to 8 hours (with opposite findings for PMA/Ionomycin) (S1 Fig), we examined the "late" cytokine responses in γδT-cells following stimulation for 23 hours with Zol, HMBPP and PMA/Ionomycin in-vitro. As shown in Fig 6A, IFNγ/TNF expression following Zol or HMBPP stimulation was detected in Vδ2 + γδT-cells (detected via Vγ9 TCR) but not Vδ1 + γδT cells or conventional CD3 + T-cells, with some downregulation of Vγ9 TCR upon activation in representative FACS plots on the top panel. Little to no IL-17 expression was detected in response to pAg or PMA/ionomycin, however (S5 Fig). Late IFNγ/ TNF responses to pAg in Vδ2 + γδT-cells were also weaker than those induced by PMA/Ionomycin (both late and early as shown in far right in bottom bar graphs in Fig 6A).
As shown in S6 Fig, there were significant correlations between IFNγ/TNF responses in Vδ2 + γδT-cells to various stimulation conditions: 1) between late responses to Zol, HMBPP and PMA/Ionomycin; 2) between late responses to Zol and to HMBPP; 3) between late and early responses to PMA/Ionomycin. These correlations supported technical consistency of our assays as well as shared pathway whereby pAg and PMA/Ionomycin activate Vδ2 + γδT-cells. However, correlations were largely lost between early PMA/Ionomycin and late pAg responses, reflecting different kinetics and strengths whereby Vδ2 + γδT-cells are activated invitro by PMA/Ionomycin and pAg.
As shown in Fig 6B, AHB subjects showed significantly weaker late IFNγ/TNF responses in Vδ2 + γδT-cells to both Zol and HMBPP (as well as IFNγ response to PMA/Ionomycin) compared to NC or CHB subjects, but without significant differences between CHB and NC subjects detected in early responses to PMA/Ionomycin (Fig 5A and 5B). However, late IFNγ/ TNF responses to pAg in Vδ2 + γδT-cells showed significant positive correlations with their NKG2A/CD94 expression (Fig 6C), similar to early IFNγ/TNF responses to PMA/Ionomycin (Fig 5C). Late IFNγ/TNF responses to pAg in Vδ2 + γδT-cells correlated positively with their Tbet/Eomes expression and inversely with CD158a expression, without significant

γδT-cells to PMA/Ionomycin stimulation and their expression of T/NK markers. Scatter plots show %IFNγ + , %IFNγ + /TNF + and %TNF + cells in Vδ2 + γδT-cells (following 5 hours of PMA/Ionomycin stimulation) on the y-axis, with x-axis showing percent expression of various T/NK markers as well as Tbet/Eomes. Correlation coefficient and p-values were determined by non-parametric Spearman rank order correlation test. For convenience, red font was used to indicate significantly positive correlations with p-values <0.05. D. Comparison between CHB, NC and AHB groups for multi-functionality of CD3 hi CD4 -T-cells following brief PMA/Ionomycin stimulation. CD3 hi CD4 -T-cells from CHB subjects (green bars) show greater co-expression of IFNγ, TNF and MIP1β following 5 hours of PMA/Ionomycin stimulation, compared to CD3 hi CD4 -T-cells from NC (white bars) and AHB (orange bars) subjects. Error bars indicate 25% and 75% interquartile ranges. P-values between 3 groups
were determined by Kruskal Wallis test (k = 3), followed by further two-way comparisons by Mann Whitney U for initial p-value below 0.05. P-values below 0.05 were considered significant and highlighted in red font for convenience. Gating strategy is shown in S4 Fig. https://doi.org/10.1371/journal.ppat.1007715.g005 Circulating gammadelta T cells in acute and chronic hepatitis B

Fig 6. IFNγ/TNF responses to pAg are preserved in Vδ2+ γδT cells from CHB (but not AHB) subjects and are associated with their expression of Tbet/Eomes and NK markers but not PD1. A. IFNγ/TNF responses to phosphoantigens in γδT-cells from CHB subjects. Representative FACS plots show IFNγ expression in γδT-cells
in CD3-gated cells. Anti-Vγ9 TCR was used to detect Vδ2 + (Vγ9 + ) γδT-cells. Bar graphs show %IFNγ + , %IFNγ + /TNF + and %TNF + cells in CD3 + T-cells (white bars), Vδ1 + γδT-cells (blue bars) and Vδ2 + γδT-cells (red bars) from 23 CHB subjects, with PBMC stimulated for 1 day (23 hours) in-vitro with phosphoantigens zoledronic acid associations with PD1 expression. Thus, late IFNγ/TNF responses to pAg in circulating Vδ2 + γδT-cells were weaker in AHB but preserved in CHB compared to uninfected controls, and correlated significantly with their expression of Tbet/Eomes and NK markers but not PD1.

Serum ALT but not HBV DNA levels in CHB correlates inversely with IFNγ/TNF responses in Vδ1 + and Vδ2 + γδT-cells to brief PMA/Ionomycin stimulation, but not to pAg stimulation
We then examined if IFNγ/TNF responses in γδT-cells correlate with clinical or virological parameters in CHB. As shown in Fig 7A, late IFNγ/TNF responses to pAg or PMA/Ionomycin in Vδ2 + γδT-cells did not correlate significantly with serum HBV DNA or ALT. By contrast, early IFNγ/TNF responses to PMA/Ionomycin in Vδ2 + as well as CD3 hi CD4 -γδT-cells showed significant inverse correlations with serum ALT (but not HBV DNA) (Fig 7B). Similar inverse correlation was detected between serum ALT and early IFNγ but not TNF expression in Vδ1 + γδT-cells following PMA/Ionomycin stimulation. Conversely, serum ALT correlated positively with %IFNγ -TNFdouble-negative Vδ1 + , Vδ2 + or CD3 hi CD4 -γδT-cells following brief PMA/ Ionomycin stimulation. As shown in Fig 7C, early MIP1β expression following PMA/Ionomycin stimulation (but not CD107a mobilization or IL17 expression) by CD3 hi CD4 -γδT-cells also correlated inversely with serum ALT, but not HBV DNA. Effect of hepatic fibrosis on γδT-cell phenotype and function could not be assessed as liver biopsy results were available in only 8/36 CHB subjects with detailed γδT-cell analyses, with only one showing cirrhosis. Thus, serum ALT was inversely associated with early IFNγ/TNF and MIP1β responses to PMA/Ionomycin in Vδ2 + γδT-cells (also IFNγ response in Vδ1 + γδT-cells) in-vitro, but not with late IFNγ/TNF responses to pAg or PMA/Ionomycin.

CHB with ALT flare is associated with weaker early IFNγ/TNF responses to brief PMA/Ionomycin stimulation in Vδ2 + γδT-cells, compared to CHB without ALT flare
ALT flares in CHB are clinically relevant events associated with necroinflammatory changes in the liver [71][72][73]. Given the inverse associations between serum ALT and early IFNγ/TNF responses in γδT-cells to PMA/Ionomycin stimulation in CHB, CHB subjects with and without a recent hepatitis flare (ALT/ULN �10 within a month of immune analyses) were further compared for IFNγ/TNF responses in γδT-cells. As shown in Fig 8A, early IFNγ/TNF responses in Vδ2 + (but not Vδ1 + ) γδT-cells to PMA/Ionomycin were significantly greater in CHB Non-Flare (NF) compared to CHB Flare (F), NC and/or AHB subjects. Conversely, IFNγ -TNFdouble-negative cells were more enriched in Vδ2 + γδT-cells from CHB F compared to CHB NF and NC subjects.

Comparisons between CHB, NC and AHB subjects for IFNγ/TNF responses in Vδ2 + γδT-cells to 1 day stimulation with phosphoantigens and PMA/Ionomycin stimulation.
Bar graphs show %IFNγ + , %IFNγ + /TNF + and %TNF + cells in Vδ2 + γδT-cells following 1 day stimulation in-vitro with phosphoantigens zoledronic acid (zol) or (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP) in addition to PMA/Ionomycin. Based on available cryopreserved PBMC, 23 CHB, 11 NC and 7 AHB subjects were included in this analysis. Error bars indicate 25% and 75% interquartile ranges. P-values between 3 groups were determined by Kruskal Wallis test (k = 3), followed by further two-way comparisons by Mann Whitney U for initial p-value below 0.05. P-values below 0.05 were considered significant and highlighted in red font for convenience. C. Correlations between IFNγ/TNF responses in Vδ2 + γδT-cells to zoledronic acid and their expression of T/NK markers. Scatter plots show % IFNγ + , %IFNγ + /TNF + and %TNF + cells in Vδ2 + γδT-cells (following 23 hours of stimulation) on the y-axis, with x-axis showing percent expression of various T/NK markers and Tbet/Eomes, combining results from 23 CHB, 11 NC and 7 AHB subjects. Correlation coefficient and p-values were determined by non-parametric Spearman rank order correlation test. For convenience, red font was used to indicate significantly positive correlations with p-values <0.05 whereas blue font was used to indicate significantly negative correlations with p-values <0.05. https://doi.org/10.1371/journal.ppat.1007715.g006 Circulating gammadelta T cells in acute and chronic hepatitis B Further comparisons between clinically defined CHB phenotype subgroups showed significantly greater early IFNγ/TNF responses to PMA/Ionomycin in Vδ2 + γδT-cells from immune tolerant (IT) or inactive carrier (IC) groups with normal ALT, compared to HBeAg+ immune active (IA+) or HBeAg-immune active (IA-) groups with elevated ALT (Fig 8B). As for late IFNγ/TNF responses in Vδ2 + γδT-cells to pAg or PMA/Ionomycin stimulation, comparisons between CHB IT, IA or IC subgroups could not be made due to small sample sizes precluding meaningful statistical comparisons (e.g. only 2 IT and 3 IC), although no significant differences were detected between CHB subgroups with and without ALT flares (Fig 8C) or with and without circulating HBeAg (Fig 8D).

Early IFNγ/TNF responses in Vδ2 + γδT-cells to brief PMA/Ionomycin stimulation improve with the resolution of AHB but not CHB Flare
To determine if clinical resolution of ALT flare in CHB is associated with changes in Vδ2 + γδT-cells, 7 CHB F subjects with available cryopreserved PBMC were examined at a second time point (T2) at least 2 months (9-41 weeks) after the initial evaluation (T1) for frequency, phenotype and early IFNγ responses to PMA/Ionomycin (Fig 9A). Seven AHB subjects with available PBMC at similar time frame were also examined for comparison (Fig 9B). As expected, ALT levels declined after the initial evaluation in all subjects, whereas HBV DNA levels declined in most but not all subjects. However, the overall circulating frequency and Tbet expression in Vδ2 + γδT-cells did not differ significantly between T1 and T2 in CHB Flare or AHB subjects. Furthermore, early IFNγ responses to PMA/Ionomycin in Vδ2 + γδT-cells did not change significantly between T1 and T2 for CHB Flare subjects (median %IFNγ + : T1 46.9% vs T2 25.9%, p = .13) and in fact declined in 5/7 subjects. By contrast, IFNγ expression in Vδ2 + γδT-cells increased in 6/7 AHB subjects (median %IFNγ+: T1 49.5% vs T2 60.2%, p = .029). Conversely, %IFNγ -TNFαdouble-negative Vδ2 + γδT-cells declined significantly in AHB but not CHB Flare subjects.
We further examined the dynamic evolution in serum ALT (red diamond) and HBV DNA (green diamond) relative to %IFNγ + /Vδ2 + γδT-cells (blue circle) in CHB Flare and AHB subjects (Fig 9C). As shown, resolution of ALT flare was not associated with increased %IFNγ + / Vδ2 + γδT-cells in most CHB Flare subjects, including CHB-87 who achieved prompt reduction in ALT and HBV DNA on antiviral therapy (gray shade). By contrast, %IFNγ + /Vδ2 + γδTcells increased in most AHB subjects as acute ALT elevation and viremia resolved, with the rank order correlation test. For convenience, red font was used to indicate significant positive correlations with p-values <0.05 whereas blue font was used to indicate significant inverse correlations with p-values <0.05. C. Serum ALT (but not HBV DNA) levels in CHB correlate with early MIP1β response in CD3 hi CD4 -T-cells to PMA/Ionomycin stimulation. Scatter plots show %MIP1β + , %IL17 + , %CD107a + CD3 hi CD4 -T-cells on the x-axis, with y-axis showing concurrent levels of HBV DNA (log IU/ml) and ALT/ULN from the day of immune sample collection. Among 36 CHB subjects with available PBMC for early cytokine responses to PMA/Ionomycin, 36 had concurrent ALT values and 30 had concurrent HBV DNA levels. Correlation coefficient and p-values were determined by non-parametric Spearman rank order correlation test. For convenience, red font was used to indicate significantly positive correlations with p-values <0.05 whereas blue font was used to indicate significantly negative correlations with p-values <0.05. https://doi.org/10.1371/journal.ppat.1007715.g007 Circulating gammadelta T cells in acute and chronic hepatitis B exception of AHB-79 who initiated antiviral therapy with prolonged viremia above 4 log and ALT elevation beyond 24 weeks.
Collectively, we show that γδT-cells are preserved in circulating frequency but altered with distinct innate phenotype and effector function in acute and chronic HBV infection with relevance to clinical status.

Discussion
While T-cells play a critical role in disease pathogenesis and viral clearance in acute HBV infection, their role in chronic HBV infection is less clear due to functional impairment with the induction of multiple immune regulatory pathways [4][5][6][7][8][9][10][11][12]. As γδT-cells are non-conventional T-cells that participate in lymphoid stress surveillance [29,30] and microbial pathogenesis [17], we asked if γδT-cells contribute to HBV pathogenesis. To this end, we examined the frequency, phenotype and effector function of circulating γδT-cells in a North American cohort of HBV-infected and uninfected subjects. Our findings show that γδT-cells are preserved in circulating frequency regardless of HBV infection, with CD3 hi CD4 -Vδ2 + γδT-cells as the predominant subset. We further show distinct phenotypic and functional characteristics of γδT-cells in acute and chronic HBV infection with potential pathogenetic relevance.
Contrary to altered γδT-cell frequencies reported in HIV, CMV and several other intracellular pathogens such as mycobacterium tuberculosis [17,[36][37][38][39], circulating γδT-cell frequencies did not differ significantly between HBV-infected and uninfected subjects in our study. Our findings differ from reports associating γδT-cell frequencies with clinical status of CHB [44][45][46][47][48]. For example, reduced Vδ2 + γδT-cell frequency was associated with severe HBV-associated liver disease in several studies of Chinese subjects with CHB [44-46]. Conversely, asymptomatic HBV carriers with low HBV DNA and ALT showed increased circulating Vδ1 + and Vδ2 + γδT-cell frequencies compared to controls in a study from Ireland with mixed African, Caucasian and Asian subjects [48]. Unlike previous studies, we examined North Americans well-preserved liver function without decompensation [10,50]. Although most of our CHB subjects consisted of Asian Americans, our findings persisted when Asian and non-Asian Americans were examined separately. We also confirmed circulating γδT-cell

Fig 8. CHB with ALT flare is associated with weaker early IFNγ/TNF responses to brief PMA/Ionomycin stimulation in Vδ2 + γδT-cells, compared to CHB without ALT flare. A. Early IFNγ/TNF responses to brief PMA/Ionomycin stimulation is greater in Vδ2 + γδT-cells (but not Vδ1 + γδT-cells) from CHB Non-Flare (NF) subjects compared to CHB Flare (F), NC or AHB subjects. Bar graphs compare 22 CHB Non-Flare (NF), 14 CHB Flare (F), 24 uninfected control (NC) and 7 AHB (A) subjects for median percentage of cells with and without IFNγ/TNF expression in Vδ1 + γδT-cells (top panel) and Vδ2 + γδT-cells (bottom panel) upon 5 hours of PMA/
Ionomycin stimulation in vitro. Right panel shows characteristic FACS contour plots for IFNγ and TNF expression in Vδ2 + γδT-cells upon PMA/Ionomycin stimulation from CHB-Non-Flare and CHB-Flare subjects. Error bars indicate 25% and 75% interquartile ranges. P-values between 4 subgroups were determined by Kruskal Wallis test (k = 4), followed by further two-way comparisons by Mann Whitney U for initial p-value below 0.05. P-values below 0.05 were considered significant and highlighted in red font for convenience. B. Early IFNγ/TNF response to brief PMA/Ionomycin stimulation is greater in Vδ2 + γδT-cells from clinical CHB phenotype groups with lower ALT values. Bar graphs compare 5 immune tolerant (IT), 18 HBeAg+ immune active (IA+), 8 HBeAg-immune active (IA-) and 5 inactive carrier (IC) subjects with CHB, showing median percentage of cells with and without IFNγ/TNF expression in Vδ2 + γδT-cells following 5 hours of PMA/Ionomycin stimulation in vitro. Error bars indicate 25% and 75% interquartile ranges. P-values between 4 subgroups were determined by Kruskal Wallis test (k = 4), followed by further two-way comparisons by Mann Whitney U for initial p-value below 0.05. P-values below 0.05 were considered significant and highlighted in red font for convenience. C/D. IFNγ/ TNF response to 1 day of pAg stimulation does not differ between CHB Non-Flare (NF) subjects compared to CHB Flare (F) or between HBeAg+ or HBeAg-CHB subjects. Bar graphs compare 14 CHB Non-Flare (NF) and 9 CHB Flare (F) subjects, as well as 15 HBeAg+ CHB and 8 HBeAg-CHB subjects, with median percentage of cells with and without IFNγ/TNF expression in Vδ2 + γδT-cells following 1 day (23 hours) stimulation with phosphoantigens zoledronic acid (zol) or (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate (HMBPP) in addition to PMA/Ionomycin (P/I). P-values between 2 subgroups were determined by Mann Whitney U with p-values below 0.05 considered significant. E. Lack of differential Tbet and Eomes expression between Vδ2 + γδT-cells from CHB Non-Flare and CHB Flare subjects. Bar graphs compare median %Tbet+ and %Eomes+ cells in Vδ2 + γδT-cells between patient groups, without significant differences between CHB Non-Flare and CHB Flare subjects. F. Lack of differential T/NK marker expression in CD3 hi CD4 -Vδ2 + γδT-cells from CHB Non-Flare and CHB Flare subjects. Bar graphs compare median % of cells expressing T/NK markers in CD3 hi CD4or Vδ2 + γδT-cells from CHB Non-Flare (CHB NF, white bar) and CHB Flare (CHB F, red bar) subjects. Expression of Tcell markers (CD127, PD1, CTLA4 and CD28) in CD3 hi CD4 -T-cells were measured in 150 CHB NF and 39 CHB F subjects. Expression of NK markers (CD56, NKG2D, NKG2A, CD158a and CD94) in CD3 hi CD4 -T-cells were measured in 22 CHB NF and 14 CHB F subjects. CD161 expression was measured in Vδ2 + γδT-cells from 22 CHB NF and 9 CHB F subjects. Error bars indicate 25% and 75% interquartile ranges. CHB, NC and AHB groups were compared by non-parametric Kruskal Wallis test (k = 3) with further comparisons between 2 groups by Mann Whitney U if the initial Kruskal Wallis test yielded p-values < 0.05. https://doi.org/10.1371/journal.ppat.1007715.g008 Circulating gammadelta T cells in acute and chronic hepatitis B frequencies by multiple approaches (e.g. staining for Vδ2 TCR and Vγ9 TCR as well as CD3 hi CD4phenotype). Furthermore, our cohort included diverse clinical CHB phenotypes including immune tolerant, immune active and inactive carriers as well as those with ALT flares, supporting the generalizability of our findings.
There was a notable impact of host factors for Vδ2 + γδT-cell frequency. For example, Asian Americans displayed 2-3 fold higher circulating Vδ2 + γδT-cell frequency compared to Non-Asian Americans. These differences were not due clinical or virological status of CHB as they were also detected in uninfected controls. They were not associated with HIV or CMV coinfection, since HIV-infected or immunosuppressed subjects were excluded in our study and CMV infection generally impacts Vδ1 + (not Vδ2 + ) γδT-cells [40][41][42]. To our knowledge, differential γδT-cell frequency between Asian and Non-Asian Americans has not been reported, although greater Vδ2 + γδT-cell frequency was described in Caucasian compared to African Americans [74]. The inverse association between age and Vδ2 + (but not Vδ1 + ) γδT-cell frequency was consistent with previous reports [59,60]. The underlying mechanisms that govern γδT-cell homeostasis and their clinical implications are not well-defined, although both environmental factors (e.g. diet, infection, microbes) as well as genetics are shown to impact human immune system [75][76][77]. In fact, Vδ2 + Vδγ9 + γδT-cells showed greater heritability than other innate-line T-cells such as Vδ1 + γδT-cells and NKT-cells, although innate immune traits were influenced more by environment whereas adaptive immune traits were impacted more by genetic [78]. In any case, the marked differences in Vδ2 + γδT-cell frequency between Asians and Non-Asians suggest that race/ethnicity must be considered in clinical studies involving Vδ2 + γδT-cells, including immunotherapy [31][32][33][34][35].

Fig 9. Early IFNγ/TNF responses in Vδ2 + γδT-cells to brief PMA/Ionomycin stimulation improve with the resolution of AHB but not CHB Flare. A/B. Clinical, virological and immunological measures during and after CHB Flare or AHB.
Graphs compare serum ALT (log IU/L), HBV DNA (log IU/L), %Vδ2 + γδT-cells/CD3 + T-cells, %Tbet/Vδ2 + γδT-cells, %IFNγ + /Vδ2 + γδT-cells, %IFNγ -TNF -/Vδ2 + γδT-cells for 7 CHB-Flare subjects (A) and 7 AHB subjects (B) at the earliest time point (T1) within 1 weeks from initial clinical presentation for CHB flare or AHB and a later time point T2 with resolution of ALT flare or AHB. Subject ID and the timing of T1 and T2 blood draws for immune analyses (relative to clinical presentation) are shown on far right. Dotted lines with unfilled markers indicate that ALT or HBV DNA values were missing from those time points and substituted from the closest available time points. Cryopreserved PBMC from T1 and T2 time points for each subject were assayed concurrently for better comparability. As shown, ALT and HBV DNA levels declined by T2 in most subjects. No significant changes were detected for %Vδ2 + γδT-cells/CD3 + T-cells or %Tbet/Vδ2 + γδT-cells. AHB (but not CHB) subjects showed increased %IFNγ + /Vδ2 + γδT-cells and decreased %IFNγ -TNF -/Vδ2 + γδT-cells between T1 and T2. C. Evolution of serum ALT and HBV DNA levels relative to IFNγ expression in Vδ2 + γδT-cells in CHB Flare and AHB subjects. Serum HBV DNA (green diamond) and ALT (red diamond) levels are shown over time (in weeks from presentation of CHB flare or AHB), and juxtaposed to %IFNγ + /Vδ2 + γδT-cells (blue filled circle) for 7 CHB Flare and 7 AHB subjects. Notably, %IFNγ + /Vδ2 + γδT-cells increased between T1 and T2 in AHB but not CHB subjects. https://doi.org/10.1371/journal.ppat.1007715.g009 Circulating gammadelta T cells in acute and chronic hepatitis B Tbet/Eomes expression patterns in γδT-cells correlated significantly with their expression of NK and T-cell activation or co-stimulatory markers, but not serum ALT or HBV DNA levels. For example, percentages of Tbet + , Eomes + or Tbet hi Eomes dim cells in Vδ1 + γδT-cells showed significant positive correlations with their CD56, CD16 and CD161 expression (S3 Fig). Percentages of Tbet + or Tbet hi Eomes dim in Vδ2 + γδT-cells also correlated positively with their expression of NK markers such as NKG2A, CD94, CD56 and CD16, but inversely with T-cell markers (e.g. PD-1, CD28 and CD127) as well as a killer immunoglobulin-like receptor CD158a (Fig 3D). Inverse associations between Tbet and PD1 as well as other co-inhibitory receptors have been reported in CD8 T-cells whereby Tbet mediates direct transcriptional repression of PD1 [79]. To our knowledge, similar associations between Tbet and PD-1 as well as other T/NK markers have not been reported in γδT-cells, but suggest a critical regulatory role for Tbet in γδT-cells. These broad and inter-related phenotypic alterations in circulating γδT-cells from AHB and CHB patients also suggest that γδT-cells participate in lymphocyte stress surveillance during HBV infection.
Several new findings emerged in our study regarding γδT-cell function in HBV infection, despite different kinetics and strengths whereby Vδ2 + γδT-cells were stimulated in-vitro by pAg (through the TCR) or non-specifically by PMA/Ionomycin (through protein kinase C and calcium signaling) [16,26,69,80]. First, compared to NC subjects, CHB subjects showed wellpreserved IFNγ/TNF responses in circulating Vδ1 + and Vδ2 + γδT-cells (even greater for early PMA/Ionomycin responses in Vδ2 + γδT-cells). This contrasted from functional impairment reported for conventional T-cells and B-cells in CHB [10,11,81,82], but resemble recent findings in HCV-infected patients [64]. Second, AHB subjects displayed significantly weaker IFNγ/TNF responses in Vδ2 + γδT-cells compared to CHB subjects for early PMA/Ionomycin response, and compared to both CHB and NC subjects for late responses to pAg and PMA/ Ionomycin. By contrast, Vδ1 + γδT-cells from AHB subjects showed preserved early IFNγ/TNF responses to PMA/ionomycin and increased Tbet hi Eomes dim phenotype, compared to Vδ1 + γδT-cells from CHB and NC subjects. In this regard, the poor effector function of Vδ2 + γδTcells from AHB subjects may reflect their enrichment for Tbet dim Eomes hi phenotype associated with exhausted HIV-specific CD8 T-cells with increased PD1 expression [65]. However, alternative mechanisms are likely, as PD1 expression was not increased in Vδ2 + γδT-cells from AHB subjects and did not correlate with IFNγ/TNF expression, despite positive correlations between PD1 expression and Tbet dim Eomes hi phenotype in Vδ2 + γδT-cells. Furthermore, we cannot rule out hepatic compartmentalization of more functional γδT-cells in AHB. Third, late IFNγ/TNF responses to pAg in Vδ2 + γδT-cells correlated with their expression of NKG2A/CD94 and CD158a as well as Tbet hi Eomes dim phenotype. For PMA/Ionomycin, late responses in Vδ2 + γδT-cells correlated significantly with CD94 and Tbet/Eomes expression, whereas early responses correlated with NKG2A and CD94 but not Tbet/Eomes expression. In this regard, increased NKG2A and CD94 expression has been associated with more cytolytic Vδ2 + γδT-cells [83] although direct stimulation through CD94/NKG2A is inhibitory to Vδ2 + γδT-cells [23,84]. Collectively, these findings show that circulating Vδ2 + γδT-cells are functionally preserved in CHB but suppressed in AHB, with regulatory roles for NK inhibitory receptors and Tbet/Eomes in their effector function, without increased T-cell activation or regulatory markers.
Surprisingly, early IFNγ/TNF responses in Vδ2 + γδT-cells to PMA/Ionomycin stimulation showed significant inverse correlations with serum ALT (but not HBV DNA) levels, despite non-specific nature of PMA/Ionomycin stimulation that bypasses the early steps of TCR signaling [69,80]. Moreover, Vδ2 + γδT-cells showed less IFNγ/TNF expression and more "anergic" IFNγ -TNFdouble-negative phenotype in CHB Flare compared to CHB Non-flare subjects. Conversely, CHB Non-flare subjects showed greater IFNγ/TNF expression in Vδ2 + γδT-cells compared to uninfected controls and AHB subjects. Significant inverse association was also detected for early IFNγ responses to PMA/Ionomycin in Vδ1 + γδT-cells, although limited to IFNγ and without significant differences between CHB with and without ALT flare. However, serum ALT did not correlate with pAg-specific IFNγ/TNF responses in Vδ2 + γδTcells, although pAgs provide more specific and physiological stimulation to Vδ2 + γδT-cells [16,[25][26][27][28]. Collectively, these findings suggest that circulating γδT-cells are functionally altered in CHB and that their early IFNγ/TNF responses to PMA/Ionomycin may provide a biomarker for immune active phenotype beyond ALT.
Given the known antiviral effects of IFNγ and TNF against HBV [4][5][6][7][8][9], it was surprising that HBV DNA levels in CHB did not correlate with the circulating frequency or IFNγ/TNF expression by γδT-cells. This lack of correlation could reflect insufficient in-vivo activation of γδT-cells to exert antiviral activity (especially given increased expression of inhibitory NKG2A/CD94 receptors) although Vδ2 + γδT-cells from our CHB subjects were responsive to phosphoantigens in-vitro. In fact, circulating γδT-cells from CHB, NC and AHB subjects showed little to no cytokine expression when stained without further stimulation in-vitro (Figs 4B, 6A and 6B). As a caveat, it should be acknowledged that we did not study the intrahepatic compartment which is enriched in γδT-cells, although both Vδ1 + and Vδ2 + γδT-cells have been detected in the liver of HBV-infected and uninfected subjects [45,85].
The inverse association between Vδ2 + γδT-cell function and ALT activity might also suggest a protective role for Vδ2 + γδT-cells against HBV-associated liver inflammation and hepatocellular injury. In fact, a late tissue protective and immune regulatory role have been suggested for γδT-cells, by directly killing activated macrophages, producing regulatory cytokines and secreting factors that promote tissue repair, beyond early pro-inflammatory effects [15,[86][87][88][89][90]. Consistent with this possibility, Vδ2 + γδT-cell depletion and anergy have been associated with adverse outcomes in HIV-infection [37][38][39]. Although beyond the scope of this study, potential regulatory roles in HBV pathogenesis have been suggested for γδT-cells with IFNγ-dependent suppression of Th17 + CD4 T-cells [45] and the induction of myeloidderived suppressor cells [49]. In this context, it is tempting to speculate if hepatitis flare in CHB might represent a failure of IFNγ−dependent regulatory function by Vδ2 + γδT-cells.
Alternatively, IFNγ/TNF deficit in Vδ2 + γδT-cells from CHB flare could be a consequence of inflammatory and regulatory mediators induced during active hepatocellular injury in-vivo [12,91,92], as CHB flares are associated with necroinflammatory changes in the liver [71][72][73]. However, IFNγ/TNF expression by Vδ2 + γδT-cells did not improve even months after the resolution of CHB flare (including in one subject with therapeutic control of HBV DNA). In this regard, persistent IFNγ/TNF deficit in Vδ2 + γδT-cells from CHB Flare patients may represent continued cellular stress with fibrogenesis and cell turnover in the chronicallly HBV-infected liver despite apparent improvement in ALT, although we could not address this question due to limited number of liver biopsy in our study. By contrast, early IFNγ response to brief PMA/ Ionomycin stimulation improved in Vδ2 + γδT-cells from AHB subjects after 2-7 months from initial presentation, suggesting that regulatory effect on Vδ2 + γδT-cell function is reversible in AHB unlike CHB.
In conclusion, circulating γδT-cells are preserved in frequency and function, but with distinct and innate phenotype in acute and chronic HBV infection, with a significant inverse associations between early IFNγ/TNF responses in Vδ2 + γδT-cells and serum ALT in CHB. Our findings suggest that circulating γδT-cells participate in lymphoid stress surveillance in HBV infection, with differential activation and differentiation with potential relevance to HBV pathogenesis.