Differential Expression of the CXCR3 Ligands in Chronic Hepatitis C Virus (HCV) Infection and Their Modulation by HCV In Vitro ^▿

Tissue, RNA isolation, and microarray analysis.

Liver biopsy samples were collected from nondiseased and HCV-infected patients and ethics approval was obtained as previously described (13). Nondiseased liver samples were obtained from sites distant from hepatic metastasis and deemed free of infection via histological analysis. All patients included in the study were negative for hepatitis B surface-antigen, and had no other causes of liver disease (Table 1). Total cellular RNA was isolated from liver biopsy samples and nondiseased liver samples by using an RNAqueous RNA extraction kit (Ambion, CA). Microarray analysis was performed as previously described (12), within the UTMB Molecular Genomics Core Facility, utilizing Affymetrix U133 plus 2.0 array gene chips. The gene expression profiles for HCV-infected liver biopsy samples were compared to those for pooled nondiseased liver samples as a baseline (n = 4), and transcripts demonstrating at least a 2.5-fold increase over those for pooled nondiseased liver samples were considered significantly elevated. Microarray analysis of liver samples from experimentally infected chimpanzees was performed as previously described (5).

Liver histology.

Histological assessment of liver biopsy samples was performed by an experienced liver pathologist (A. Ruszkiewicz). Grading of each section was performed in a blind manner and in accordance with the Scheuer index (37).

RNA extraction, cDNA synthesis, and PCR analysis.

Real-time PCR analysis was utilized to quantitate mRNA corresponding to CXCL11, CXCL10, and CXCL9. Total cellular RNA was isolated from liver biopsy material or in vitro cell monolayers by using Trizol (Invitrogen, CA), and first-strand cDNA was synthesized and real-time PCR performed as described previously (13). The primers for cDNA detection were as follows: for CXCL11, 5′-CAAGGCTTCCCCATGTTCA-3′ and 5′-CCCAGGGCGTATGCAAAGA-3′; for CXCL10, 5′-TCCACGTGTTGAGATCATTGC-3′ and 5′-TCTTGATGGCCTTCGATTCTG-3′; for CXCL9, 5′-GAGTGCAAGGAACCCCAGTAGT-3′ and 5′-GGTGGATAGTCCCTTGGTTGGT-3′; and for RPLPO, control gene primers 5′-AGATGCAGCAGATCCGCAT-3′ and 5′-GGATGGCCTTGCGCA-3′.

Analysis of serum chemokine levels.

Sera were collected from 16 treatment-naïve CHC patients (average age = 45.4 ± 8.2) attending the Royal Adelaide Hospital liver clinic and from 16 healthy controls (average age = 39.4 ± 11.3) sourced from healthy blood donors attending the Australian Red Cross donation facility in Adelaide. Serum samples were processed within 2 h of collection and stored at −80°C until required. Samples were diluted appropriately in 0.1% bovine serum albumin and analyzed for CXCR3 chemokine expression. Enzyme-linked immunosorbent assay (ELISA) analysis was performed using R&D Systems antibody pairs, following the manufacturer's instructions, as previously described (13).

Cell lines.

Human hepatoma Huh-7 cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco-BRL, Grand Island, NY) supplemented with 10% fetal bovine serum, l-glutamine, penicillin, and streptomycin in a humidified 37°C/5% CO₂ incubator.

JFH-1 infection of Huh-7 cells.

Huh-7 cells were seeded at 1.6 × 10⁶ cells in a T75-cm³ flask for 24 h before being infected with 2 × 10⁴ focus-forming units/ml of JFH-1 virus (provided by T. Wakita, National Institute of Infectious Diseases, Tokyo, Japan) (48) in a 2-ml volume of supplemented DMEM for 6 h. Following initial infection, the medium volume was increased to 10 ml and the cells were left for 48 or 72 h before being seeded for the luciferase assay or mRNA quantification experiments, respectively. Upon completion of the experiment, JFH-1-infected Huh-7 cells were analyzed to ensure HCV protein expression by using indirect immunofluorescence microscopy essentially as described previously (21), with the exception that cells were incubated at a 1/300 dilution of mixed sera taken from patients chronically infected with HCV and a fluorescein isothiocyanate-conjugated rabbit anti-human immunoglobulin G secondary antibody (Chemicon, Temecula, CA). Cells were visualized using an Olympus Provis AX70 microscope, and all experiments were performed at least in triplicate. All cells were at least 80% JFH-1 infected at the completion of the experiments, as determined by cell enumeration (data not shown).

CXCR3 chemokine promoter analysis.

DNA fragments comprising 934 and 1,013 bp and located upstream of the CXCL11 and CXCL9 ATG start sites, respectively, were cloned into the luciferase reporter vector pGL3 (Promega, WI). This was performed using directional cloning and the following primer pairs: for CXCL11, 5′-GGCTTCACCCCTCAACGAGTGG-3′ (forward) and 5′-GTTTGTTTTTTGCTGTTGCTGCTG-3′ (reverse), tagged with the NheI and XhoI restriction sites; and for CXCL9, 5′-TGATTATGGCCATTCTTGCAGG-3′ (forward) and 5′-AGTGATAGAATGGAGTTCCAAG-3′ (reverse), tagged with the XhoI and HindIII restriction sites. The CXCL10 promoter construct was a kind gift from R. M. Ransohoff (Lerner Research Institute, Cleveland, OH) (25).

Promoter studies were performed via transfection of constructs into naïve and JFH-1-infected Huh-7 cells. Cells (7 × 10⁴) were transfected with 400 ng of a reporter construct either alone or in combination with (i) 400 ng of HCV protein-expressing plasmid (pCDNA3.1 CP7, NS3, and NS4a; pCDNA6 NS3/4a and 3-5B; and pCMV NS5a-Flag and NS5b-Flag, all kind donations from Kui Li, University of Texas Medical Branch) (9, 17, 20); (ii) RIG-I-wt, RIG-N, IRF3-wt, or IRF3-5D (all kind donations from Stephen Polyak, University of Washington, Seattle, WA) (22, 45), using Fugene 6 reagent (Roche, Indianapolis, IN); or (iii) in vitro-transcribed JFH-1 or 5′ untranlated region (5′UTR) RNA (JFH-1) transfection 24 h after reporter construct transfection as described below. Ten nanograms of PhRL-TK, a Renilla-expressing vector, was also cotransfected to normalize transfection efficiency. Cultures were incubated for 24 h before treatment with 1,000 U/ml IFN-γ either alone or in combination with 40 ng/ml TNF-α (Peprotech, Rehovot, Israel) for a further 24 h. Luciferase activity was measured using the luciferase assay system (Promega) and a Turner TD 20/20 luminometer. All experiments were performed at least in triplicate.

Treatment of cells with dsRNA, interferon, and TNF-α.

Huh-7 cells were seeded at 2 ×10⁵ in a six-well plate without antibiotics and allowed to settle for 24 h before RNA transfection with either 2 μg of T7 RNA polymerase-derived JFH-1 RNA, HCV 5′UTR (ApaI-digested JFH-1), or poly(I)-poly(C) RNA (Sigma, MO), using DMRIE C (Invitrogen). Transfection reagent was removed at 4 h and replaced with either complete DMEM or IFN-γ (1,000 units/ml) in combination with TNF-α (40 ng/ml) for 4 h. RNA was then harvested and real-time PCR performed as described previously (12). Experiments involving poly(I)-poly(C) RNA stimulation of the CXCR3 promoters were performed as described above 24 h after transfection of the CXCR3 luciferase promoter constructs into Huh-7 cells. All experiments were performed at least in triplicate.

Statistical analysis.

The nonparametric Spearman correlation was used for all correlations. Student t tests and the Wilcoxon rank sum method were utilized to analyze the distributions of two normally distributed and nonnormally distributed independent data sets. All statistical analysis was performed using SPSS 10.

The CXCR3 ligands are highly expressed in CHC.

We have previously reported elevated expression of the CXCR3 ligand CXCL11 in CHC (12). To gain further insight into possible global chemokine networks at play in CHC, we further interrogated our array data to identify differential expression levels of chemokine transcripts in livers of both human CHC patients and chimpanzees experimentally infected with HCV. Analysis of liver samples from eight humans chronically infected with HCV revealed seven chemokine transcripts elevated in at least three out of eight patients (Table 2). Most striking were the CXCR3 ligands CXCL10, CXCL11, and CXCL9, which were upregulated in all eight patients, with average mRNA increases of 62-, 21-, and 19-fold, respectively, compared with levels for nondiseased liver samples. The CCR5 ligand RANTES was also upregulated in six out of eight patients (average increase of 7.6-fold) and, along with the three CXCR3 ligands, has previously been reported to be elevated in CHC (10, 12, 13, 18, 39). Other chemokines found to be upregulated in fewer patients were GCP-2, Mip-3β, and SCM-1β.

We next investigated whether a similar profile of chemokine expression occurs in chimpanzees chronically infected with HCV. Analysis of Affymetrix GeneChip data from 18 chimpanzees chronically infected with HCV revealed that CXCL10 and CXCL11 were the most significantly elevated chemokines (24.3- and 28-fold, respectively); however, in contrast to what was found for humans, CXCL9 and RANTES were not elevated. Together with the human data, this suggests that the CXCR3 chemokines play a significant role in recruitment of T lymphocytes into the liver and that, while similarities exist, there are also differences between the livers of chimpanzees and humans chronically infected with HCV.

Real-time PCR was performed on 19 separate liver biopsy samples from CHC patients to analyze the relative distributions of the CXCR3 ligands in individual patients (Fig. 1). All patients had increased mRNA levels of all three chemokines, with the average increases above the levels for pooled nondiseased human liver samples being 17.54-, 31.95-, and 15.23-fold for CXCL11, CXCL10, and CXCL9, respectively (Fig. 1B). However, the upregulation of these three chemokines is not uniform in all patients; this is further supported by analysis of the correlation of these individual data sets, where upregulation of CXCL11 correlates with that of CXCL10 and CXCL9 (P < 0.001 and P = 0.01, respectively) (Fig. 1B) whereas no correlation can be seen when increases of CXCL9 and CXCL10 for different patients are compared. These data are suggestive of some form of variable stimulation of these transcripts within the livers of CHC patients, perhaps in addition to that by endogenous IFN-γ and TNF-α.

CXCR3 ligand expression correlates with liver inflammation.

In the present study, CXCL11, CXCL10, and CXCL9 mRNA levels were found to correlate significantly with the Scheuer classification lobular inflammation scores for each biopsy sample (P = 0.001, P = 0.014, and P = 0.002, respectively) (Fig. 2). However, a significant correlation between chemokine mRNA increase and portal inflammation was demonstrated only for CXCL11 (P = 0.033). No correlations were observed between chemokine mRNA levels and total Scheuer scores or fibrosis scores (data not shown). Collectively, these results suggest that the CXCR3 ligands may play a role in immune-mediated pathogenesis of hepatitis C, with CXCL11 potentially having the more dominant role.

CXCR3 chemokine expression in sera from CHC patients versus healthy controls.

Sera taken from CHC patients (n = 16) or healthy controls (n = 16) were analyzed using ELISA for CXCL11, CXCL10, and CXCL9. Figure 3 demonstrates that only CXCL10 levels were significantly increased in the sera of HCV patients compared with levels for healthy controls (P = 0.0037). The mean CXCL11, CXCL10, and CXCL9 protein concentrations in sera from CHC patients were 368, 2,218, and 633 pg/ml, respectively, and those in sera from healthy controls were 261, 1,405, and 693 pg/ml, respectively. These results suggest that, unlike for CXCL10, a significant chemokine gradient exists for CXCL11 and CXCL9 between the liver (increased levels) (Fig. 1) and the peripheral circulation (no increase) (Fig. 3).

HCV replication regulates CXCR3 expression in response to IFN-γ and TNF-α.

We have previously shown that HCV replication can increase CXCL11 expression following stimulation of HCV genomic replicon cells with IFN-γ and TNF-α (13). To extend this observation, we next investigated the role of HCV replication on CXCL9 and CXCL10 expression following IFN-γ and TNF-α stimulation by using the HCV JFH-1 cell culture system. Preliminary experiments showed that the sensitivity of our ELISA was not sufficient to detect CXCL11 in the Huh-7 line used for JFH-1 experiments, precluding ELISA analysis. Hence, mRNA expression of CXCR3 ligands was used for JFH-1 experiments. JFH-1 infection of Huh-7 cells demonstrated a significant (P < 0.01) increase in CXCL11 mRNA under IFN-γ and IFN-γ/TNF-α stimulation; however, no significant differences were observed for CXCL9 or CXCL10 mRNA (Fig. 4A). IFN-γ is known to inhibit HCV replication, and these experiments were performed at 16 h poststimulation, at which point HCV replication is still active, as determined via immunofluorescence staining (results not shown). These results suggest that HCV replication can selectively increase expression of CXCL11 but not CXCL9 or CXCL10 in response to IFN-γ and TNF-α.

HCV replication activates CXCR3 ligand transcription via a dsRNA mechanism.

To examine the effect of HCV replication on CXCR3 ligand transcription, the CXCL11, CXCL9, and CXCL10 promoters were cloned upstream of the luciferase reporter gene (Fig. 4B). JFH-1-infected Huh-7 cells were transfected with CXCR3 chemokine promoter-luciferase constructs and then stimulated with either IFN-γ alone or IFN-γ and TNF-α for 24 h. Cell lines replicating HCV JFH-1 (Fig. 4C) showed significant increases (six- to sevenfold) in luciferase output in the presence of IFN-γ and TNF-α for the CXCL11 promoter-luciferase reporter plasmid (P < 0.01), compared with control levels. No selective increase was noted following stimulation with IFN-γ/TNF-α for the CXCL9 or CXCL10 promoter constructs (Fig. 4C).

To determine if HCV protein expression was responsible for the selective transcriptional activation of the CXCL11 promoter, reporter assays were repeated in the presence of individual HCV proteins or the complete complement of nonstructural proteins by using published expression constructs (HCV protein expression was confirmed in an immunofluorescence assay using anti-HCV patient sera or specific anti-HCV antibodies where appropriate; results not shown). As shown in Fig. 5, expression of HCV proteins did not modulate CXCR3 ligand promoter activity in either untreated or stimulated (IFN-γ or TNF-α) Huh-7 cells. Collectively, these results demonstrate that HCV replication is able to selectively increase transcription of CXCL11 but not CXCL9 or CXCL10. However, this activation does not appear to be due to the actions of any single HCV protein.

To investigate if the selective increase in CXCL11 transcription noted above was a result of dsRNA signaling, we examined the ability of dsRNA to act synergistically with IFN-γ and TNF-α to increase CXCR3 ligand promoter activity and de novo mRNA expression. First, we transfected Huh-7 cells with the CXCR3 ligand promoter reporter constructs, followed by transfection with dsRNA in the form of poly(I:C) or in vitro-transcribed HCV RNA representing either the complete HCV genome or the 5′UTR. Transfection of the synthetic dsRNA analogue poly(I:C) resulted in a small increase in CXCR3 ligand promoter activity (Fig. 6A). However, in the presence of IFN-γ/TNF-α and poly(I:C) stimulation, there was a synergistic increase in CXCL11 promoter activity compared to the level for IFN-γ/TNF-α stimulation alone. This was not observed for CXCL9 or CXCL10 promoter activity (Fig. 6A). A selective synergistic increase in promoter activity was also seen for CXCL11 (but not CXCL9 or CXCL10) following transfection of either the HCV 5′UTR or the complete HCV genome, each of which has the capacity to result in significant regions of dsRNA (Fig. 6B) (41).

We next examined the ability of in vitro-transcribed HCV JFH-1 RNA to stimulate de novo mRNA production of CXCR3 ligands and the effect of IFN-γ and TNF-α stimulation. Transfection of Huh-7 cells with JFH-1 RNA followed by IFN-γ/TNF-α stimulation for 8 h resulted in mRNA increases of 525-, 306-, and 63-fold for CXCL11, CXCL9, and CXCL10, respectively, compared to the levels for untreated cells (Fig. 6C). The increase for CXCL11 mRNA was significant (P < 0.01), but those for CXCL9 and CXCL10 mRNA were not, as treatment of Huh-7 cells with IFN-γ/TNF-α alone demonstrated mRNA increases of 302-, 284-, and 61-fold for CXCL11, CXCL9, and CXCL10, respectively, compared to the levels for untreated cells. Furthermore, similar results were seen following stimulation with poly(I:C), IFN-γ, and TNF-α (Fig. 6C). Thus, this selective response of CXCL11 expression seems not to be specific for HCV RNA but may be part of a more generic cellular response to dsRNA.

dsRNA sensor proteins selectively increase CXCL11 expression following IFN and TNF stimulation.

Our previous experiments established that the selective activation of CXCL11 transcription to IFN-γ and TNF-α was mediated by a dsRNA response. To characterize this effect in more detail, we investigated the dsRNA RIG-I signaling pathway that is central for triggering the host response to HCV infection. Activation of this pathway results in IRF3 and NF-κB activation following either viral infection or dsRNA stimulation (9, 27, 41, 45). Using transfection of a constitutively active RIG-I molecule (RIG-N) together with the CXCR3 ligand promoters resulted in a significant increase (threefold) for CXCL11 transcription, but not for CXCL9 or CXCL10 transcription, that was independent of dsRNA stimulation (Fig. 7A). As predicted, transfection of RIG-wt did not stimulate promoter activity. It is surprising that the CXCL10 promoter was not activated, considering that it contains an ISRE and suggests that other factors are required in addition to IRF3 for activation of CXCL10 transcription. As predicted, the CXCL9 promoter was not responsive to RIG-N, presumably due to the absence of an ISRE within the CXCL9 transcriptional promoter (Fig. 4B).

RIG-I signals to IRF3 and is essential for regulation of the transcription of a number of other chemokines, such as RANTES (23) and IL-8 (43). Furthermore, IRF3 activation is fundamental to establishing the host response to HCV and is a target of the HCV NS3/4a protease, which blocks innate sensing of dsRNA (9). Therefore, we next investigated the effects of IRF3 expression on CXCR3 ligand transcription by using IRF3-wt and a constitutively active form of IRF3 (IRF3-5D) that has previously been described (22). Under basal conditions, IRF3-5D resulted in a significant increase (3.5-fold; P < 0.01) in CXCL11 transcription while there was no increase with IRF3-wt, as expected (Fig. 7B). In concordance with the RIG-I data presented above and the lack of an ISRE within the CXCL9 promoter, IRF3-5D did not stimulate CXCL9 or CXCL10 promoter activity (Fig. 7B). Interestingly, in the presence of IRF3-5D and stimulation with IFN-γ and TNF-α, a selective increase in CXCL11 promoter activity compared to the level for transfection with IRF3-wt was seen (17-fold versus 7-fold; P < 0.0001) (Fig. 7C). Collectively, our data suggest that HCV infection can selectively activate CXCL11 transcription synergistically in the presence of IFN-γ and TNF-α via the RIG-I/IRF3 pathway.