Intrahepatic Transcriptional Signature Associated with Response to Interferon-α Treatment in the Woodchuck Model of Chronic Hepatitis B

Abstract

Recombinant interferon-alpha (IFN-α) is an approved therapy for chronic hepatitis B (CHB), but the molecular basis of treatment response remains to be determined. The woodchuck model of chronic hepatitis B virus (HBV) infection displays many characteristics of human disease and has been extensively used to evaluate antiviral therapeutics. In this study, woodchucks with chronic woodchuck hepatitis virus (WHV) infection were treated with recombinant woodchuck IFN-α (wIFN-α) or placebo (n = 12/group) for 15 weeks. Treatment with wIFN-α strongly reduced viral markers in the serum and liver in a subset of animals, with viral rebound typically being observed following cessation of treatment. To define the intrahepatic cellular and molecular characteristics of the antiviral response to wIFN-α, we characterized the transcriptional profiles of liver biopsies taken from animals (n = 8-12/group) at various times during the study. Unexpectedly, this revealed that the antiviral response to treatment did not correlate with intrahepatic induction of the majority of IFN-stimulated genes (ISGs) by wIFN-α. Instead, treatment response was associated with the induction of an NK/T cell signature in the liver, as well as an intrahepatic IFN-γ transcriptional response and elevation of liver injury biomarkers. Collectively, these data suggest that NK/T cell cytolytic and non-cytolytic mechanisms mediate the antiviral response to wIFN-α treatment. In summary, by studying recombinant IFN-α in a fully immunocompetent animal model of CHB, we determined that the immunomodulatory effects, but not the direct antiviral activity, of this pleiotropic cytokine are most closely correlated with treatment response. This has important implications for the rational design of new therapeutics for the treatment of CHB.

Fig 1. Design of the wIFN-α treatment study in woodchucks chronically infected with WHV.

Chronic WHV carrier woodchucks were dosed three times a week (TIW) for 15 weeks with placebo, or for 7 weeks with 20 μg wIFN-α followed by another 8 weeks with 100 μg wIFN-α (15 weeks total). Animals were typically followed for additional 8 weeks after the treatment period (follow-up), although this was extended by two weeks for two wIFN-treated woodchucks (animals M1002 and M1004) that had no evidence of viral recrudescence at the end-of-study (week 23). *PK analysis of serum wIFN-α levels was not performed due to the lack of a sufficiently sensitive quantitative method (see Methods).

Fig 2. wIFN-α treatment of chronic WHV carriers induces suppression of serum antigenemia and viremia.

Change in serum (a) WHsAg and (b) WHV DNA relative to week -3 (pre-treatment baseline). Circles indicate the mean of each group (open: placebo, closed: wIFN-α), and the error bars represent the standard error of the mean. The WHsAg level for two wIFN-treated animals was ≤ lower limit of detection (LLOD; 20 ng/mL) at various times during the study (S2 Fig); the LLOD was used to estimate the WHsAg decline at these timepoints. Note that seven animals (three in the placebo group, four in the wIFN-α group) died during the study, and one animal in the wIFN-α group (M1004) was excluded from the analysis since it was likely naturally clearing WHV as the study initiated (see Table 1).

Fig 3. High dose wIFN-α significantly inhibits WHV.

Maximum reductions in (a) serum and (b) intrahepatic viral parameters, and (c) maximum serum ALT and AST levels in response to placebo, low dose (20 μg) wIFN-α and high dose (100 μg) wIFN-α treatment. Changes in viral parameters were calculated relative to week -3 (pre-treatment baseline), with the exception of F1018 (wIFN-α group), for which week 0 was used as the baseline for the intrahepatic cccDNA and RNA analyses. The bar height indicates the mean of each group, and the errors bars represent the standard error of the mean. The lowest WHsAg level for two wIFN-treated animals (M1002 and F1022) was ≤ LLOD (20 ng/mL), and so the LLOD was used to estimate the maximum WHsAg decline. Per the sampling scheme outlined in Fig 1, the following data was included in the analyses: maximum reduction in serum WHV DNA and WHsAg at weeks 0–16 (placebo), 0–7 (20 μg dose) and weeks 8–16 (100 μg dose); maximum reduction in intrahepatic cccDNA and RNA at weeks 0–15 (placebo), weeks 0 and 3 (20 μg dose) and weeks 7, 11 and 15 (100 μg dose); maximum serum ALT and AST levels at weeks 0–16 (placebo), 0–7 (20 μg dose) and weeks 11–16 (100 μg dose). Animals in the wIFN-α group were only included if data from all relevant time-points was available, with the exception of F1020, for which no week 16 sample was available for serum WHV DNA, WHsAg, ALT and AST analysis. Statistical significance was calculated by one-way ANOVA with Tukey's multiple comparison correction.

Fig 4. Serum WHsAg and liver enzymes for individual wIFN-treated animals.

Serum WHsAg (black open circles) is plotted on the left y-axis. Serum ALT (red circles), AST (blue circles) and SDH (green circles) are all plotted on the right y-axis. The treatment response group classifications (a-d) are described in Table 1. *Animals died prior to end-of-study. Note all data was from pre-dose (or equivalent).

Fig 5. Differential induction of whole blood gene expression by low dose and high dose wIFN-α.

qRT-PCR data for (a) ISGs and (b) T_H1-type cytokine genes expressed as fold-change relative to week 0 pre-dose (pre-treatment baseline). The bar height indicates the mean maximal fold-change for each group, and the errors bars represent the standard error of the mean. Placebo: maximal induction at 6 hours post-dose at weeks 0, 1, 3, 7, 11 and 15. 20 μg: maximal induction at 6 hours post-dose at weeks 0, 1 and 3. 100 μg: maximal induction at 6 hours post-dose at weeks 7, 11 and 15. Animals from the wIFN-α treatment group were only included if data from all relevant time-points was available. Statistical significance was calculated with log-transformed values by one-way ANOVA with Tukey's multiple comparison correction.

Fig 6. Modular analysis of intrahepatic transcriptional signatures in the wIFN-α treatment group.

Data from all available wIFN-treated animals (n = 5–11) were included at each time-point. Spot intensity (red: over-expressed; blue: under-expressed) denotes the percentage of transcripts significantly changed in each module (M) and is defined by the scale bar. The functional interpretation of each module [29] is displayed on the right. Only modules with enrichment greater than 10% at one or more time-point are displayed. At each time-point, all genes selected for modular analysis had an absolute fold-change > 1.5 with a Benjamini-Hochberg corrected FDR<0.05 relative to the time-matched placebo group. The horizontal bars together with the week numerators indicate the study stage, as described in Fig 1.

Fig 7. Comparable induction of intrahepatic expression of most ISGs with low dose and high dose wIFN-α.

(a) Unsupervised hierarchical clustering of differentially expressed intrahepatic ISGs of animals that were responders (R, n = 3), partial responders (PR, n = 2) or non-responders (NR, n = 2) to wIFN-α treatment. Note that there was no week 19 sample for the responder group animal M1002 and the week 25 sample (end-of-study for this responder animal) was included at week 23 (end-of-study for most animals) for ease of data comparison. The sources of the ISGs are described in S3 Table. Heatmap columns represent samples from individual animals collected at the indicated times, and rows represent different genes (n = 209). Red and blue coloring of cells represents high and low expression levels (normalized count data), respectively, as indicated by the scale bar for log₂ normalized values. (b) qRT-PCR data expressed as fold-change relative to week -3 (pre-treatment baseline). The bar height indicates the mean of each group, and the errors bars represent the standard error of the mean. Placebo: maximum induction at 6 hours post-dose at weeks 0 and 7. 20 μg: sample collected 6 hours post-first dose of 20 μg wIFN-α (week 0). 100 μg: sample collected 6 hours post-first dose of 100 μg wIFN-α (week 7). Animals from the wIFN-α treatment group were only included if both 20 μg wIFN-α (week 0) and 100 μg wIFN-α (week 7) data was available (see S4 Table). Statistical significance was calculated with log-transformed values by one-way ANOVA with Tukey's multiple comparison correction.

Fig 8. Characterization of intrahepatic transcriptional signature associated with response to wIFN-α treatment.

Analysis of genes (n = 468) differentially induced by high dose (100 μg) wIFN-α (see S7 Fig, “High dose”). (a) Modular analysis of intrahepatic gene expression, as described in Fig 6. Only the cytotoxic cell module (M2.1) had enrichment greater than 10% at one or more time-point. (b) Top canonical pathways identified by Ingenuity Pathway Analysis. Pathway enrichment was calculated with the Fisher’s exact test with multiple testing correction by the Benjamini and Hochberg method. The–log(p-value) for p = 0.05 and p = 0.01 significance levels are indicated. (c) qRT-PCR data expressed as fold-change relative to week -3 (pre-treatment baseline). The bar height indicates the mean of each group, and the errors bars represent the standard error of the mean. Placebo: mean induction at 6 hours post-dose at weeks 0, 7 and 15 in placebo-treated animals. wIFN-α: induction at 6 hours post-dose at week 15 in non-responder (NR, n = 2) or responder (R, n = 3) animals. No samples from partial responder (PR) animals were available for qRT-PCR analysis (S6 Table). Statistical significance was calculated with log-transformed values by one-way ANOVA with Tukey's multiple comparison correction.