Hepatic interferon-stimulated genes are differentially regulated in the liver of chronic hepatitis C patients with different interleukin-28B genotypes

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Pretreatment up-regulation of hepatic interferon (IFN)-stimulated genes (ISGs) has a stronger association with the treatment-resistant interleukin (IL)28B minor genotype (MI; TG/GG at rs8099917) than with the treatment-sensitive IL28B major genotype (MA; TT at rs8099917). We compared the expression of ISGs in the liver and blood of 146 patients with chronic hepatitis C who received pegylated IFN and ribavirin combination therapy. Gene expression profiles in the liver and blood of 85 patients were analyzed using an Affymetrix GeneChip (Affymetrix, Santa Clara, CA). ISG expression was correlated between the liver and blood of the MA patients, whereas no correlation was observed in the MI patients. This loss of correlation was the result of the impaired infiltration of immune cells into the liver lobules of MI patients, as demonstrated by regional gene expression analysis in liver lobules and portal areas using laser capture microdissection and immunohistochemical staining. Despite having lower levels of immune cells, hepatic ISGs were up-regulated in the liver of MI patients and they were found to be regulated by multiple factors, namely, IL28A/B, IFN-λ4, and wingless-related MMTV integration site 5A (WNT5A). Interestingly, WNT5A induced the expression of ISGs, but also increased hepatitis C virus replication by inducing the expression of the stress granule protein, GTPase-activating protein (SH3 domain)-binding protein 1 (G3BP1), in the Huh-7 cell line. In the liver, the expression of WNT5A and its receptor, frizzled family receptor 5, was significantly correlated with G3BP1. Conclusions: Immune cells were lost and induced the expression of other inflammatory mediators, such as WNT5A, in the liver of IL28B minor genotype patients. This might be related to the high level of hepatic ISG expression in these patients and the treatment-resistant phenotype of the IL28B minor genotype. (Hepatology 2014;59:828–838)

Abbreviations
ALT

alanine aminotransferase

AST

aspartate aminotransferase

CCL

CC chemokine ligand

CHC

chronic hepatitis C

CLLs

cells in liver lobules

CPAs

cells in portal areas

CXCL10/IP-10

chemokine (C-X-C motif) ligand 10/interferon-gamma-induced protein 10

CXCR3

chemokine (C-X-C motif) receptor 3

DCs

dendritic cells

DVL

disheveled

FZD5

frizzled family receptor 5

G3BP1

GTPase-activating protein (SH3 domain)-binding protein 1

GGT

gamma-glutamyl transpeptidase

HCV

hepatitis C virus

IFI44

interferon-induced protein 44

IFIT1

interferon-induced protein with tetratricopeptide repeats 1

IFN

interferon

IHC

immunohistochemical

IL

interleukin

ISGs

interferon-stimulated genes

JFH-1

Japanese fulminant hepatitis type 1

LCM

laser capture microdissection

MA

major genotype

MAd

major genotype, down-regulated

MAu

major genotype, up-regulated

MI

minor genotype

Mx

myxovirus (influenza virus) resistance

NK

natural killer

OAS2

2′-5′-oligoadenylate synthetase 2

PALT

portal-tract-associated lymphoid tissue

Peg_IFN

pegylated IFN

RBV

ribavirin

RTD-PCR

real-time detection polymerase chain reaction

SG

stress granule

siRNA

small interfering RNA

SVR

sustained virologic response

WNT5A

wingless-related MMTV integration site 5A

Interferon (IFN) and ribavirin (RBV) combination therapy has been a popular modality for treating patients with chronic hepatitis C (CHC); however, ∼50% of patients usually relapse, particularly those with hepatitis C virus (HCV) genotype 1b and a high viral load.[1] The recently developed direct-acting antiviral drug, telaprevir, combined with pegylated (Peg)-IFN plus RBV, significantly improved sustained virologic response (SVR) rates; however, the SVR rate was not satisfactory (29%-33%) in patients who had no response to previous therapy.[2] Therefore, IFN responsiveness is still an essential clinical determinant for treatment response to triple (Peg-IFN+RBV+DAA) therapy.

A recent landmark genome-wide association study identified a polymorphism in the interleukin (IL)28B, IFN-λ3) gene that was associated with either a sensitive (major genotype; MA) or resistant (minor genotype; MI) treatment response to Peg-IFN and RBV combination therapy and was characterized by either up- (-u) or down-regulation (-d) of interferon-stimulated genes (ISGs).[3-5] However, the underlying mechanism for the association of this polymorphism and treatment response has not been clarified. Previously, we showed that up-regulation of the pretreatment expression of hepatic ISGs was associated with an unfavorable treatment outcome and was closely related to the treatment-resistant IL28B genotype (TG or GG at rs8099917).[6] It could be speculated that the pretreatment activation of ISGs would repress additional induction of ISGs after treatment with exogenous IFN. However, it is unknown how hepatic ISGs are up-regulated in treatment-resistant CHC patients and why patients with high levels of ISG expression cannot eliminate HCV. Therefore, other mechanisms should be involved in the unfavorable treatment outcome of patients with the treatment-resistant IL28B genotype.

In the present study, we performed gene expression profiling in the liver and blood and compared the expression of ISGs between them. Furthermore, ISG expression in liver lobules and portal areas was analyzed separately using a laser capture microdissection (LCM) method. Finally, we identified an immune factor that is up-regulated in patients with the treatment-resistant IL28B genotype and mediates favorable signaling for HCV replication.

Materials and Methods

Patients

We analyzed 168 patients with CHC who had received Peg-IFN-α2b (Schering-Plough K.K., Tokyo, Japan) and RBV combination therapy for 48 weeks at the Graduate School of Medicine, Kanazawa University Hospital, Japan and its related hospitals, as reported previously (Table 1 and Supporting Table 1).[6]

Table 1. Clinical Characteristics of 146 Patients Whose Liver and Blood Samples Were Analyzed by RT-PCR
 Major (MA)Minor (MI) 
Clinical CategoryMajor ISG Up (MAu) Major ISG Down (MAd)   P Value
  1. a

    P < 0.05.

  2. Abbreviations: BMI, body mass index; ALT, alanine aminotransferase; WBC, leukocytes; Hb, hemoglobin; PLT, platelets; TG, triglycerides; T-chol, total cholesterol; LDL-chol, low density lipoprotein cholesterol; HDL-chol, high density lipoprotein cholesterol; NA, not applicable; NS, not significant.

No. of patientsn = 42 n = 68 n = 36 NA
Age and sex       
Age (years)55(30-72)56(31-72)55(30-73)NS
Sex (M vs. F)27 vs. 15 34 vs. 34 19 vs. 17 NS
Treatment responses       
SVR/TR/NR24/12/6 30/33/6 6/7/23a MAu vs. MI < 0.0001, MAd vs. MI < 0.0001
IL28B genotype (TT vs. TG+GG)TT TT TG/GG (31/5) NA
Liver factors       
F stage (1/2/3/4)14/13/11/4 30/20/11/7 14/8/10/4 NS
A grade (A0-1 vs. A2-3)16 vs. 26 37 vs. 31 20 vs. 16 NS
ISGs (Mx1, IFI44, IFIT1)3.83a(2.14-9.48)1.30a(0.36-2.08)5.52a(0.86-17.3)MAu vs. MAd < 0.0001, MAu vs. MI < 0.0001, MAd vs. MI < 0.0001
IL28A/B41.3a(4-151)11.7a(1-53)22.7a(3-93)MAu vs. MAd < 0.0001, MAu vs. MI = 0.0004, MAd vs. MI = 0.031
Blood factors       
ISGs (Mx1, IFI44, IFIT1)11.1a(2.78-24.9)4.76(0.41-20.6)5.64(0.71-2.8)MAu vs. MAd < 0.0001, MAu vs. MI < 0.0001
IL28A/B1.6(0.1-7.7)1.3(0.2-6.4)1.3(0.3-3.6)NS
Laboratory parameters       
HCV-RNA (KIU/mL)2,430(160-5,000)2,692(140-5,000)1,854a(126-5,000)MAd vs. MI = 0.017
BMI (kg/m2)24(18.7-31.9)24(16.3-34.7)22.8(19.1-30.5)NS
AST (IU/L)86a(22-258)54(18-192)64(21-178)MAu vs. MAd = 0.0008
ALT (IU/L)112a(17-376)75(16-345)79(18-236)MAu vs. MAd = 0.023
γ-GTP (IU/L)99a(21-392)47(4-367)74(20-298)MAu vs. MAd = 0.0003
WBC (/mm3)4,761(2,100-8,100)4,982(2,800-9,100)4,823(2,500-8,200)NS
Hb (g/dL)14.1(11.4-16.7)14.1(9.3-16.9)13.9(11.2-16.4)NS
PLT (× 104/mm3)15.2(9.2-27.8)16.8(7-39.4)16.3(9-27.8)NS
TG (mg/dL)112(42-248)102(42-260)136a(30-323)MAd vs. MI = 0.02
T-Chol (mg/dL)162(90-221)169(107-229)167(81-237)NS
LDL-Chol (mg/dL)77(36-123)83a(42-134)72(29-107)MAd vs. MI = 0.04
HDL-Chol (mg/dL)40(18-67)43(27-71)47a(27-82)NS
Viral factors       
ISDR mutations ≦ 1 vs. ≧ 223 vs. 19a 51 vs. 17 26 vs. 10 MAu vs. MAd = 0.02
Core aa 70 (wild-type vs. mutant)24 vs. 18 42 vs. 22 16 vs. 20a MAd vs. MI = 0.02

Preparation of Liver Tissue and Blood Samples

A liver biopsy was performed on samples from 168 patients, and blood samples were obtained from 146 of these patients before starting therapy (Table 1 and Supporting Table 1). Detailed procedures are described in the Supporting Materials and Methods.

Affymetrix GeneChip Analysis

Liver tissue samples from 91 patients and blood samples from 85 patients were analyzed using an Affymetrix GeneChip (Affymetrix, Santa Clara, CA). LCM analysis was performed in 5 MAu, MAd, and MI patients. Affymetrix GeneChip analysis and LCM were performed, as described previously.[6, 7] Detailed procedures are described in the Supporting Materials and Methods.

Hierarchical Clustering and Pathway Analysis of GeneChip Data

GeneChip data analysis was performed using BRB-Array Tools (http://linus.nci.nih.gov/BRB-ArrayTools.htm), as described previously.[7] Pathway analysis was performed using MetaCore (Thomson Reuters, New York, NY). Detailed procedures are described in the Supporting Materials and Methods.

Quantitative Real-Time Detection Polymerase Chain Reaction, Cell Lines, Cell Migration Assay, Vector Preparation, HCV Replication Analysis, and Statistical Analysis

These procedures are described in detail in the Supplemental Material and Methods.

Results

Differential ISG Expression in Liver and Blood of Patients With Different IL28B Genotypes

Previously, we showed that pretreatment up-regulation of hepatic ISGs was associated with an unfavorable treatment outcome and was closely related to the treatment-resistant IL28B MI (TG or GG at rs8099917).[6] To examine whether expression of hepatic ISGs would reflect the expression of blood ISGs, we compared ISG expression between the liver and blood. We utilized three ISGs (interferon-induced protein 44 [IFI44], interferon-induced protein with tetratricopeptide repeats 1 [IFIT1], and myxovirus (influenza virus) resistance [Mx1]) with a high dynamic range, comparable relative expression, and good predictive performance.[6] Mean values of the three ISGs detected by real-time detection polymerase chain reaction (RTD-PCR) in 168 liver tissue samples (Supporting Table 1) showed a significant up-regulation of their expression in nonresponder or treatment-resistant IL28B MI (TG/GG; rs8099917) patients, compared to responder (SVR+TR) or treatment-sensitive IL28B MA (TT; rs8099917) patients, as reported previously (Fig. 1A and Supporting Fig. 1A).[6] However, ISG expression in 146 blood samples (Table 1) showed no difference between responders and nonresponders or the IL28B major and minor genotypes (Fig. 1B and Supporting Fig. 1B). To explore these findings further, gene expression profiling using Affymetrix GeneChips was performed on liver and blood samples from 85 patients (Supporting Tables 2 and 3), and the expression of 37 representative ISGs[6] was compared (Fig. 1C-E). MA patients were divided into two groups according to their ISG expression pattern in the liver: MAu and MAd. MI patients expressed ISGs at a higher level than MAu patients. Interestingly, ISG expression in MA patients showed a similar expression pattern in liver and blood, and ISGs were up-regulated in MAu patients and down-regulated in the MAd patients. However, MI patients showed a different ISG expression pattern in liver and blood, where ISGs were up-regulated in the liver, but down-regulated in the blood (Fig. 1C). The correlation of the mean values of the three ISGs (IFI44, IFIT1, and Mx1) between liver and blood from 146 patients demonstrated a significant correlation between values in MA patients (Fig. 1D), whereas no correlation was observed in MI patients (Fig. 1E). Interestingly, ISG expression correlated significantly between liver and blood of responders, but not of nonresponders, in MA and MI patients (Supporting Fig. 1C-F). These results indicate that the correlation of ISG expression in the liver and blood is an important predictor of treatment response.

Figure 1.

Comparison of ISG expression in liver and blood of patients with different IL28B genotypes. (A and B) RTD-PCR results of mean ISG expression (IFI44+IFIT1+Mx1) in liver (A) and blood (B) of IL28B major (MAu/Mad) and minor (MI) genotype patients. (C) One-way hierarchical clustering analysis of 85 patients using 37 representative ISGs derived from liver (upper) and blood (lower). (D and E) Correlation of mean ISG expression (IFI44+IFIT1+Mx1) in liver and blood of IL28B major (MA; D) and minor (MI; E) genotype patients.

Clinical Characteristics of IL28B MA Patients With Up- and Down-Regulated ISGs and IL28B MI Patients

From the expression pattern of ISGs and mean values of the three ISGs (IFI44, IFIT1, and Mx1), we could use receiver operating characteristic curve analysis to set a threshold of 2.1-fold to differentiate MAu and MAd patients. Following this criterion, 42 MAu, 68 MAd, and 36 MI patients (total, 146) were grouped (Table 1). Hepatic ISG expression was highest in MI patients, whereas blood ISG expression was highest in MAu patients. Conversely, hepatic IL28A/B (IFN-λ2/3) expression was highest in MAu patients, whereas blood IL28A/B expression showed no difference among the three groups. Serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), and gamma-glutamyl transpeptidase (GGT) levels were significantly higher in MAu patients than in MAd patients. Interestingly, serum ALT levels were significantly correlated with ISG expression in MA patients, but not in MI patients (Supporting Fig. 2E,F).

Gene expression profiling in peripheral immune cells showed the presence of active inflammation in MAu patients, whereas the inactive or remissive phase of inflammation was observed in MAd patients. In contrast, monophasic and intermediate inflammation existed in MI patients (Supporting Fig. 3).

Reduced Number of Immune Cells in the Liver Lobules of IL28B MI Patients

To examine the discordant expression of ISGs in liver and blood of MI patients, we performed LCM to collect cells in liver lobules (CLLs) and cells in portal areas (CPAs) separately from each of five liver biopsied samples from MAu, MAd, and MI patients (Fig. 2A). Interestingly, the ISG expression pattern in CLLs from MA patients was similar to that of CPAs, and ISGs were up-regulated in MAu patients and down-regulated in MAd patients. ISG expression in CLLs from the MI patients was different to that in CPAs, and ISGs were up-regulated in CLLs, but down-regulated in CPAs (Fig. 2A). We hypothesized that the discordance of ISG expression between CLLs and CPAs in MI patients might be the result of the lower number of immune cells that infiltrated the liver lobules of these patients. To prove this hypothesis, immunohistochemical (IHC) staining was performed (Fig. 2B). IHC staining showed that IFI44 was strongly expressed in the cytoplasm and nucleus of CLLs from MI patients, whereas it was intermediately expressed in MAu patients and weakly expressed in MAd patients. Interestingly, IFI44 was strongly expressed in CPAs of MAu patients and weakly expressed in CPAs of MAd patients, showing a correlation between expression in CLLs and CPAs of MA patients, whereas ILI44 expression was relatively weak in CPAs, compared with CLLs, in MI patients (Fig. 2B). In the same section of the specimens, there were less CD163-positive monocytes and macrophages in MI patients than in MAu and MAd patients. Similarly, there were fewer CD8-positive T cells in MI patients than in MAu and MAd patients (Fig. 2B). Semiquantitative evaluation of CD163- and CD8-positive lymphocytes in liver lobules showed a significantly lower number of cells in MI patients than in MAu and MAd patients (Supporting Fig. 4A,B). To support these findings, we examined the expression of 24 surface markers of immune cells in CLL, including dendritic cells (DCs), natural killer (NK) cells, macrophages, T cells, B cells, and granulocytes (Supporting Fig. 5A). The expression of immune cell-surface markers was repressed in MI patients, compared to MAu and MAd patients. Furthermore, whole-liver expression profiling in 85 patients showed the reduced expression of these surface markers in MI patients, compared to MAu and MAd patients (Supporting Fig. 5B). These results indicated that fewer immune cells had infiltrated the liver lobules of MI patients.

Figure 2.

LCM and IHC staining of biopsied liver specimens. (A) Comparison of the ISG expression pattern of whole liver (upper), CLLs (upper middle), CPAs (lower middle), and blood (bottom). CLLs and CPAs were obtained from 5 MI, MAu, and MAd patients, who are indicated by small black bars. (B) IHC staining of IFI44, CD163, and CD8 in MI, MAu, and MAd patients.

In addition to these findings, various chemokines, such as CC chemokine ligand (CCL)19, CCL21, CCL5, and chemokine (C-X-C motif) ligand (CXCL)13, which are important regulators for the recruitment of DCs, NK cells, T cells, and B cells in the liver, were significantly down-regulated in MI patients, compared to MAd and MAu patients (Supporting Fig. 4C-F).

Hepatic ISG Expression Is Significantly Correlated With IL28A/B, but not IFN-α or IFN-β

The lower number of immune cells in the liver lobules of MI patients implies that reduced levels of IFN are produced from DCs, macrophages, and so on. These findings prompted us to examine the relationship between hepatic ISGs and IFN-α, IFN-β, IL29/IFN-λ1, and IL28A/B in CHC patients. Hepatic ISG expression was significantly correlated with IL28A/B, but not IFN-β (Fig. 3A-C) or IFN-α (data not shown) in MAu, MAd, and MI patients. Expression of IL29 was correlated with hepatic ISG expression only in MAu patients. These results indicate that hepatic ISGs would be mainly induced by IL28A/B in CHC patients. Interestingly, the correlation between hepatic ISGs and IL28A/B was strongest in MA patients (P < 0.0001 in MAu; P = 0.0006 in MAd), whereas rather a weak correlation was observed in MI patients (P = 0.015). Moreover, the ratio of hepatic ISGs to IL28A/B was larger in MI patients than in MA patients (S = 0.061 in MI; S = 0.028 in MAu; S = 0.020 in MAd), suggesting the presence of additional factors that can induce expression of ISGs in MI patients. Therefore, we evaluated the expression of the recently discovered IFN-λ4 in MI patients. Interestingly, there was a significant correlation between hepatic ISG and IFN-λ4 expression (P = 0.0003; Fig. 3C).

Figure 3.

Correlation analysis of hepatic ISGs and IL28A/B, IL29, IFN-β, and IFN-λ4. Correlation of mean ISG (IFI44+IFIT1+Mx1) and IL28A/B, IL29, IFN-β, and IFN-λ4 expression was evaluated in MAu (A), MAd (B), and MI (C) patients. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Wingless-Related MMTV Integration Site 5A and Its Receptor, Frizzled Receptor 5, Are Significantly Up-Regulated in the Liver of Patients With the IL28B MI

IFN-λ4 is a promising factor to induce ISG expression in MI patients,[8] and the functional relevance of IFN-λ4 for the pathogenesis of CHC is under investigation. We searched for other factors that could induce ISG expression in MI patients. A closer observation of gene expression profiling in CLLs obtained by LCM demonstrated that WNT signaling was specifically up-regulated in MI patients (Supporting Fig. 6). Further observation enabled us to identify that the WNT ligand, wingless-related MMTV integration site 5A (WNT5A), and its receptor, frizzled receptor 5 (FZD5), were up-regulated in MI patients. RTD-PCR results on 168 liver-biopsied samples confirmed the significant up-regulation of WNT5A and FZD5 in MI patients, compared to MAu and MAd patients (Fig. 4A,B). Interestingly, WNT5A expression was negatively correlated with chemokine expression (Supporting Fig. 7). IHC staining showed up-regulation of FZD5 in liver lobules of MI patients, but not in MAu or MAd patients (Fig. 4C). WNT5A expression was significantly correlated with hepatic ISG expression in MI and MAd patients (Fig. 4D). Interestingly, we found a weak, but significant, correlation between WNT5A and IFN-λ4 expression in MI patients (Fig. 4E).

Figure 4.

WNT5A and FZD5 are up-regulated in IL28B MI patients. (A) RTD-PCR results of WNT5A expression in liver of MAu, MAd, and MI patients. (B) RTD-PCR results of FZD5 expression in liver of MAu, MAd, and MI patients. (C) IHC staining of IFI44 and FZD5 expression in liver of MAu, MAd, and MI patients. (D) Correlation of mean ISG (IFI44+IFIT1+Mx1) and WNT5A expression in liver of MAu, MAd, and MI patients. (E) Correlation of WNT5A and IFN-λ4 expression in liver of MI patients.

WNT5A Induces ISG Expression, but Stimulates HCV Replication in Huh-7 Cells

To examine the functional relevance of up-regulated expression of WNT5A in MI patients, we first evaluated expression levels of WNT5A and ISGs (2′-5′-oligoadenylate synthetase 2 [OAS2], Mx1, IFI44, and IFIT1) in two immortalized human hepatocyte cell lines, THLE-5b and TTNT cells (Supporting Materials and Methods), and one human hepatoma cell line, Huh-7 cells (Supporting Fig. 8A,B). WNT5A was moderately expressed in THLE-5b and TTNT cells, whereas its expression in Huh-7 cells was minimal. Interestingly, ISG expression in these cells correlated well with expression of WNT5A (Supporting Fig. 8B). Small interfering RNA (siRNA) to WNT5A efficiently repressed WNT5A expression to ∼20% of the control in THLE-5b cells, and in this condition, ISG expression was significantly decreased to 30%-50% of the control (Supporting Fig. 8C). Conversely, transduction of WNT5A using a lentivirus expression system in Huh-7 cells significantly increased OAS2 expression (Supporting Fig. 8D), as well as Mx1 and IFIT1 expression (data not shown), in the presence and absence of HCV infection. Surprisingly, HCV replication, as determined using Gaussia luciferase activity, increased in WNT5A-transduced cells (Supporting Fig. 8E). Furthermore, WNT5A-transduced cells supported more HCV replication than nontransduced cells under IFN treatment (Supporting Fig. 8F).

WNT5A-FZD5 Signaling Induces the Expression of the Stress Granule Protein, GTPase-Activating Protein (SH3 Domain)-Binding Protein 1, Which Supports HCV Replication

These findings were further confirmed by using Huh-7 cells that were continuously infected with Japanese fulminant hepatitis type 1 (JFH-1; Huh7-JFH1), which is a genotype 2a HCV isolate.[9] Interestingly, expression of WNT5A in Huh7-JFH1 cells was significantly up-regulated, compared with uninfected Huh-7 cells, and showed an equivalent expression level with THLE-5b cells (Fig. 5A). siRNA to WNT5A efficiently repressed WNT5A expression to ∼20% of the control, and in this condition, ISG expression (IFI44 was not expressed in Huh-7 cells), HCV RNA, and infectivity were repressed to 25%-65%, 60%, and 40% of the control, respectively (Fig. 5B and Supporting Fig. 9A). Interestingly, CXCL13 expression was significantly increased in this condition. We evaluated the expression of GTPase-activating protein (SH3 domain)-binding protein 1 (G3BP1), a recently recognized stress granule (SG) protein that supports HCV infection and replication.[10] Expression of G3BP1 was repressed to 60% of the control by knocking down WNT5A. Conversely, overexpression of WNT5A in Huh7-JFH1 cells significantly decreased CXCL13 expression and increased HCV RNA, infectivity, and G3BP1 expression (Fig. 5C and Supporting Fig. 9B). A recent report demonstrated that G3BP1 is a disheveled (DVL)-associated protein that regulates WNT signaling downstream of the FZD receptor.[11] Knocking down FZD5 in Huh7-JFH1 cells significantly reduced the expression of DVL1-3, G3BP1, Mx1, and IFIT1 as well as HCV infectivity (Supporting Fig. 9C,D). Interestingly, G3BP1 expression was significantly up-regulated in liver of MI patients (Fig. 5D). Furthermore, G3BP1 expression was significantly correlated with WNT5A expression in liver of the CHC patients (Fig. 5E). More dramatically, a strong correlation was observed between expression of FZD5 and G3BP1 in liver of CHC patients (Fig. 5F).

Figure 5.

Relationship between WNT5A and FZD5 signaling and the SG protein, G3BP1. (A) WNT5A expression in THLE-5b, Huh-7, and Huh7-JFH1 cells. (B) Knocking down WNT5A and changes of OAS2, Mx1, IFIT1, CXCL13, and G3BP1 expression, HCV RNA, and infectivity in Huh7-JFH1 cells. (C) Overexpression of WNT5A after transfection with pCMV-WNT5A and decrease in CXCL13 expression and increase in HCV RNA, infectivity, and G3BP1 expression. (A-C) Experiments were performed in duplicate and repeated three times (n = 6). Values are the means ± standard error. *P < 0.05; **P < 0.01; ***P < 0.005. (D) RTD-PCR results for G3BP1 expression in liver of MAu, MAd, and MI patients. (E) Correlation of WNT5A and G3BP1 expression in the liver. (F) Correlation of FZD5 and G3BP1 expression in the liver. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Discussion

The underlying mechanism for the association of the IL28B genotype with treatment responses to IFN-based therapy for HCV has not yet been clarified. We and others have shown that pretreatment up-regulation of hepatic ISGs was associated with an unfavorable treatment outcome[7, 12, 13] and was closely related to treatment-resistant MI IL28B, compared with treatment-sensitive MA IL28B.[6]

By comparing ISG expression in liver and blood, we found that their expression was correlated in MA patients, but not in MI patients. LCM analysis of ISG expression in CLLs and CPAs showed the loss of the correlation between CLLs and CPAs in MI patients (Fig. 2A). This might be the result of the impaired migration of immune cells into liver lobules that was demonstrated by decreased expression of immune cell-surface markers in CLLs by LCM (Supporting Fig. 5A) and IHC staining (Fig. 2B). Lymphocyte accumulation in the portal area (portal-tract-associated lymphoid tissue; PALT) might be involved in extravasation of lymphocytes from vessels in the portal area, but others demonstrated that DCs appeared in the sinusoidal wall and passed through the space of Disse to PALT, where the draining lymphatic duct is located.[14] There should be an active movement of immune cells between liver lobules and PALT, as reflected by the correlation of ISG expression in CLLs and CPAs in the MA patients of this study.

ISGs were reportedly up-regulated in hepatocytes of treatment-resistant IL28B genotype patients, but were up-regulated in Kupffer cells of treatment-sensitive genotype patients.[15] Our results confirmed these findings; however, we also showed that expression of various immune cell-surface markers, such as those on DCs, NK cells, macrophages, T cells, B cells, and granulocytes, was lower in MI than in MA patients (Supporting Fig. 5). In addition, we showed that expression of various chemokines was also repressed in MI patients, compared to MA patients (Supporting Fig. 4C-F).

Up-regulation of pretreatment chemokine (C-X-C motif) ligand 10/interferon-gamma-induced protein 10 (CXCL10/IP-10) serum levels is also associated with an unfavorable treatment outcome.[16] CXCL10 expression in the liver was significantly correlated with hepatic ISG expression and was higher in nonresponders than in responders (Supporting Fig. 10). Our results support the usefulness of serum CXCL10 for prediction of treatment outcome. Chemokine (C-X-C motif) receptor 3 (CXCR3) expression, a receptor for CXCL10, was inversely correlated with hepatic ISG expression and was significantly lower in MI than in MA patients (Supporting Fig. 10).

The lower number of immune cells in the liver lobules of MI patients would imply the reduced production of IFN from DCs, macrophages, and so on. Correlation analysis showed that hepatic ISGs were mainly associated with type III IFNs (IL28A/B and IL29), but not type I IFNs (IFN-α or IFN-β), although a significant association with IL29 was only observed in MA patients with up-regulated ISGs. This might be related to the high serum ALT levels in MAu patients (Fig. 3). Closer examination of hepatic ISGs and IL28A/B suggested that factors other than IL28A/B might regulate ISG expression in MI patients. During the preparation of this study, IFN-λ4 was newly identified to be expressed in hepatocytes from treatment-resistant IL28B genotype patients.[8] Interestingly, we found a significant correlation between hepatic ISGs and IFN-λ4 in MI patients (Fig. 3C). Moreover, a closer examination of gene expression profiling in MI patients enabled us to detect up-regulation of the noncanonical WNT ligand, WNT5A. RTD-PCR analysis of 168 patients confirmed up-regulation of WNT5A and its receptor, FZD5, in MI patients. Importantly, WNT5A expression was significantly correlated with hepatic ISG expression in MI patients. A recent report showed that WNT5A induces expression of ISGs, increases sensitivity of keratinocytes to IFN-α,[17] and might be involved in the immune response to influenza virus infection.[18] Therefore, we examined the role of WNT5A in hepatocytes. Interestingly, expression of WNT5A and ISGs was well correlated, and knocking down WNT5A using siRNA reduced expression of ISGs in THLE-5b cells (Supporting Fig. 8). Conversely, transduction of Huh-7 cells with WNT5A using a lentivirus system increased expression of ISGs. Despite the increase in ISG expression, WNT5A did not suppress HCV replication, but rather increased it in Huh-7 cells (Supporting Fig. 8). These results were also confirmed by using Huh-7 cells continuously infected with JFH-1. By knocking down or overexpressing WNT5A in Huh7-JFH1 cells, we showed that HCV-RNA was positively regulated by WNT5A (Fig. 5B,C).

WNT5A and its receptor, FZD5, mediate noncanonical WNT signaling, such as planar cell polarity and the WNT-Ca2+-signaling pathway through G proteins. WNT5A reportedly inhibits B- and T-cell development by counteracting canonical WNT signaling.[19] We found that G3BP1, an SG assembly factor, was up-regulated by WNT5A (Fig. 5C). SGs were reportedly formed by endoplasmic reticulum stress, followed by HCV infection, and localized around lipid droplets with HCV replication complexes.[10] G3BP1 contributes to SG formation and increases HCV replication and infection in Huh-7 cells.[10] Moreover, a recent report demonstrated that G3BP1 is a DVL-associated protein that regulates WNT signaling downstream of the FZD receptor.[11] In this study, repression of WNT5A or FZD5 significantly reduced expression of DVL1-3, G3BP1, Mx1, and IFIT1 as well as HCV infectivity in Huh7-JFH1 cells (Fig. 5 and Supporting Fig. 9).

Importantly, we found a significant correlation between WNT5A and G3BP1 expression in liver tissue samples (Fig. 5E). We also found a significant correlation between FZD5 and G3BP1 expression in liver tissue samples (Fig. 5F). Thus, up-regulated noncanonical WNT5A-FZD5 signaling participates in the induction of ISG expression, but preserves HCV replication and infection in hepatocytes by increasing levels of the SG protein, G3BP1. These findings may explain the pathophysiological state of the treatment-resistant phenotype in MI patients.

In this study, we demonstrated impaired immune cell infiltration of the liver in treatment-resistant IL28B genotype patients, and we also demonstrated that up-regulation of hepatic ISGs in treatment-resistant IL28B genotype patients was mediated by multiple factors, including IL28A/B, IFN-λ4, and WNT5A. We found a significant negative correlation between WNT5A and various chemokines in liver of CHC patients (Supporting Fig. 7). Interestingly, WNT5A directly repressed one of these chemokines, CXCL13, a B-lymphocyte chemoattractant, in HCV-infected hepatocytes. These results indicate that loss of immune cells from the liver may be associated with the induction of other inflammatory factors, such as WNT5A, in MI patients, although we did not identify which cells express WNT5A. Further studies are needed to explore their functional relevance in the pathogenesis of CHC.

Acknowledgment

The authors thank Mina Nishiyama for her technical assistance.

Ancillary

Advertisement