Truncated CXCL10 is associated with failure to achieve spontaneous clearance of acute hepatitis C infection


  • Potential conflict of interest: Nikolai V. Naoumov is currently an employee of Novartis Pharma AG, Basel, Switzerland.

  • Supported by the Foundation for Liver Research and by a EASL Sheila Sherlock Fellowship.


The pathogenesis of hepatitis C virus (HCV) infection is strongly influenced by the nature of the host's antiviral immunity. Counterintuitively, elevated serum concentrations of C-X-C chemokine 10 (CXCL10), a potent chemoattractant for antiviral T-cells and NK-cells, are associated with poor treatment outcomes in patients with chronic HCV. It has been reported that an N-terminal truncated form of CXCL10, generated by the protease dipeptidylpeptidase 4 (DPP4), can act as chemokine antagonist. We sought to investigate CXCL10 antagonism in the clinical outcome and evolution of acute HCV infection. We collected serial blood samples from 16 patients, at the clinical onset of acute HCV infection and at 12 standardized follow-up timepoints over the first year. Intact and truncated CXCL10 and DPP4 activity were quantified in all longitudinal samples. In addition, NK-cell frequency/phenotype, and HCV-specific T-cell responses were assessed. Subjects developing chronicity (n = 11) had higher concentrations of CXCL10 (P < 0.001), which was predominantly in a truncated form (P = 0.036) compared to patients who spontaneously resolved infection (n = 5). Truncated CXCL10 correlated with HCV-RNA (r = 0.40, P < 0.001) and DPP4 activity (r = 0.53, P < 0.001). Subjects who resolved infection had a higher frequency of HCV-specific interferon-gamma (IFNγ)-producing T-cells (P = 0.017) and predominance of cytotoxic NK-cells (P = 0.005) compared to patients who became chronic. Patients who became persistently infected had higher proportions of cytokine-producing NK-cells, which were correlated with concentrations of truncated CXCL10 (r = 0.92, P < 0.001). Conclusion: This study provides the first evidence of chemokine antagonism during acute HCV infection. We suggest that the DPP4-CXCL10 axis inhibits antiviral innate and adaptive host immunity and favors establishment of viral persistence. (Hepatology 2014;60:487–496)


alanine amino transferase


chronic hepatitis C, chronic HCV infection


C-X-C chemokine 10


C-X-C chemokine receptor 3, CXCL10-receptor


dipeptidylpeptidase 4


enzyme-linked immunosorbent assay


enzyme-linked immunospot assay


hepatitis C virus


interferon alpha


interferon gamma




HCV non-structural protein 3


peripheral blood mononuclear cell


polymerase chain reaction




World Health Organization

Approximately 3-4 million people are newly infected with hepatitis C virus (HCV) annually, of whom more than 80% will develop chronic infection.[1] It is well established that the host immune response plays a key role in defining the clinical outcome of HCV infection.[2-5] However, the precise mechanisms responsible for the high rates of viral persistence are not fully understood. This paucity of knowledge stems from the shortage of appropriate in vivo / in vitro models and the difficulty of collecting samples during the acute phase, as these subjects are usually asymptomatic or symptomatic for a short period only.[1] This has impeded temporal investigations analyzing the relationship between host immunity and the clinical course of HCV infection.

C-X-C chemokine 10 (CXCL10; also known as interferon-gamma [IFNγ]-inducible-protein-10 or IP-10) is a potent chemokine produced in the liver by hepatocytes and liver-infiltrating lymphocytes during HCV infection, inducing the migration of T-cells and NK-cells to the inflamed organ in order to perform their antiviral effector functions. Counterintuitively, this potent chemoattractant has been implicated in the pathogenesis of chronic HCV infection (cHCV). We and others have shown elevated concentrations of total CXCL10 associated with cHCV in humans,[6-8] chimps,[9] and viral replication.[10] A recent study in a cohort of acute HCV patients also reported higher serum levels of total CXCL10 in patients that established viral persistence.[11] Notably, higher serum concentrations of CXCL10 are a strong negative predictor of response to treatment with pegylated (Peg)-IFNα plus ribavirin in cHCV patients12; yet the underlying mechanisms for these unexpected associations are not fully elucidated.

CXCL10 binds to the receptor CXCR3 on T-cells and NK-cells and the engagement of this synapse should result in the activation/migration of effector lymphocytes to the liver. It has been suggested that CXCL10 in the serum of cHCV patients may not represent the biologically active form. Indeed, there is evidence that a truncated form of CXCL10 exists, resulting from the N-terminal cleavage of two amino acids by the protease dipeptidylpeptidase 4 (DPP4, or CD26). Importantly, the truncated form of CXCL10 retains CXCR3 binding, but does not induce signaling. As such, it acts as a dominant negative antagonizing the effects of the biologically active long CXCL10.13,14

The aims of this study were to longitudinally investigate and directly compare the truncated/long forms of CXCL10, DDP4 activity, and the functionality of HCV-specific T-cells and NK-cells during the clinical evolution of acute HCV at standardized and structured timepoints from the clinical onset of infection.

Patients and Methods

Patient Recruitment and Study Design

This investigation was a prospectively designed clinical protocol developed jointly between the Institute of Hepatology, London (IoH), and the National University Hospital of Infectious Diseases, Lithuania (LID), to capture and serially follow new cases of symptomatic acute HCV infection for 1 year (Supporting Methods). The study was coordinated by IoH, where the functional analysis of the immune response was performed blindly. CXCL10 and DPP4 measurements were conducted by INSERM, Paris. Sample collection at LID was conducted using standardized procedures provided by IoH using previously established criteria15; training and reagents were also supplied. After serial collection, samples were shipped to IoH in accordance with the UK Health and Safety Executive regulations (Biological Substances, IATA Packing Instructions) and WHO guidelines for transport of infectious substances (WHO/CDS/CSR/LYO/2005 22).

Sixteen patients with acute hepatitis C were enrolled in the study. Diagnosis and recruitment of acute hepatitis C were based on established criteria[16] and included the acute symptomatic onset of hepatitis in previously healthy individuals. Inclusion criteria were HCV-RNA positivity, alanine aminotransferase (ALT) >10× upper limit of normal (ULN) and known/suspected exposure to HCV within the previous 4 months. Patients with other causes of acute or chronic liver disease including hepatitis B virus, human immunodeficiency virus, autoimmune hepatitis, alcohol abuse, toxic hepatitis, metabolic conditions, and hematological/coagulation problems were excluded. None of the patients had a previous history of HCV infection and none were treated during this study. Longitudinal immune-surveillance was performed in serial blood samples collected at 12 predefined structured and standardized timepoints over 1 year, starting from clinical presentation of acute HCV infection and at the following subsequent visits: weeks 1-3 and months 1-4, 6, 8, 10, and 12. Forty mL of peripheral blood was taken at each timepoint for clinical, virological, and immunological assessment. This study protocol was approved by the Lithuanian Bioethics Committee. All patients gave written informed consent. Table 1 summarizes baseline characteristics of the patients. Definition of clinical outcomes was based on HCV-RNA and ALT levels according to previously established criteria16; subjects achieving sustained undetectable HCV-RNA and normalization of ALT by month 6 postacute infection were considered spontaneous resolvers.

Table 1. Baseline Characteristics
GroupAgeGenderReported Risk FactorHCV GenotypeIL28B AllelesHCV RNA (Log10 copies/mL)ALT (IU/L)ALT (xULN, F=35/M=40 IU/L)ALT (IU/L)Pre-exposure HCV Results
Chr21MSexual1a/1bCC3.99140035.0372Yes, antiHCV(-) 9 months before
Chr42FSexual1bCC5.53163046.6758Yes, HCV-RNA(-) 6 months before
Chr23MDentist1aCT6.7171317.8713Yes, antiHCV(-) 3 & 12 months before
Chr35FSurgery; Sexual1bCT8.0993026.61104No
 median [range]M / FMedical / Sex / IVDU / Sharing / Unknown1 (1a / 1b) / 2 / 3CC / CTmean ± SEmean ± SEmean ± SEmean ± SE 
 35 [21; 57]5 / 66 / 3 / 2 / 0 / 19 (4 / 6) / 1 / 18 / 36.28 ± 0.381351.55 ± 266.4136.5 ± 7.4902.27 ± 152.30 
Res21FRazor; Sexual1bCC4.5982723.6166No
Res19FSexual (domestic); IVDU3aCC3.88163046.61109Yes, antiHCV(-) 6 months before
 median [range]M / FMedical / Sex / IVDU / Sharing / Unknown1 (1a / 1b) / 2 / 3CC / CTmean ± SEmean ± SEmean ± SEmean ± SE 
 21 [19; 47]1 / 40 / 3 / 2 / 2 / 02 (0 / 2) / 0 / 34 / 15.72 ± 0.621307.40 ± 137.4036.3 ± 3.71127.40 ± 307.92 

HCV-RNA Detection, Quantification, and Genotyping

Qualitative detection of serum HCV-RNA to confirm the diagnosis of HCV infection was performed with the polymerase chain reaction (PCR)-based AMPLICOR Hepatitis C Virus Test v. 2.0 assay from Roche (sensitivity: 50 IU/mL) (Roche, Nutley, NJ) according to the manufacturer's instructions. Quantitative serum levels of HCV-RNA were also analyzed at all timepoints during the study protocol. Total RNA was extracted from 140 μL of thawed plasma using the QIAamp Viral RNA Mini Kit (Qiagen, Crawley, UK). HCV-RNA was then quantitated in triplicate by real-time PCR according to established protocols[2, 4] using an internal WHO International Standard for HCV [96/790] (National Institute for Biological Standards and Controls, Potters Bar, UK). HCV genotypes were determined using the VERSANT HCV Genotype Line Probe Assay (LIPA) (Bayer Healthcare, Belgium) according to the manufacturer's instructions. HCV genotyping was performed by Lab21 (Cambridge, UK).

Interleukin (IL)28B Genotyping

Genomic DNA was extracted using QIAamp DNA Blood Midi Kit (Qiagen) from 1 × 106 peripheral blood mononuclear cells (PBMCs) isolated from heparinized blood by Lymphoprep gradient centrifugation (Nygaard, Oslo, Norway). Alleles C and T from the single nucleotide polymorphic locus (SNP) rs12979860 near the IL28B gene were analyzed in triplicate by ABI Prism 7500 TaqMan Allelic Discrimination assay according to the manufacturer's instructions (Applied Biosystems, Warrington, UK).

Quantitation of Long/Truncated CXCL10, DPP4 Activity, and IL15

Plasma concentrations of total, long/nontruncated (amino acids 1-77) and short/truncated (amino acids 3-77) forms of CXCL10 were measured with the Luminex-based 3-plex assay previously described[13] (Rules Based Medicine, TX). The enzymatic activity of DPP4 was quantified using the luciferase-based DPP4-Glo protease assay (Promega, Southampton, UK) according to the manufacturer's instructions. Plasma samples were collected appropriately for the quantification of truncated CXCL10 and DPP4 activity. Serum levels of IL15 were quantitated in triplicate by standard sandwich enzyme-linked immunosorbent assay (ELISA) (Quantikine ELISA, R&D Systems, UK).

Enumeration of HCV-Specific IFNγ- and IL10-Producing T-Cells by Enzyme-Linked Immunospot Assay (ELISpot)

PBMCs were isolated in Lithuania from heparinized blood within 4 hours of venesection by Lymphoprep gradient centrifugation, cryopreserved in 10% dimethyl sulfoxide (DMSO) as previously described,[2, 4, 15, 17] and stored in liquid nitrogen for quarterly batch-shipments to the IoH, London, for immunological analysis. After thawing, cell recovery was >90% and viability by Trypan Blue exclusion was >95% in all samples. The ELISpot assays were performed as previously described,[15] on longitudinal samples collected between baseline and month 6. Recombinant HCV-Core and HCV-NS3 antigens (Mikrogen, Neuried, Germany) were used at 1 μg/mL. Phytohemagglutinin (PHA) (Sigma, Dorset, UK) (1 μg/mL) and Tetanus Toxoid (TT) (Merck, Nottingham, UK) (2 μg/mL) were used as positive controls. Spots were counted with an AID ELISpot plate reader (AID Diagnostika, Strassberg, Germany) (Supporting Methods).

NK-Cell Subsets Analysis

Flow cytometric examination of NK-cells was performed by staining PBMCs with fluorochrome-labeled antibodies to CD3, CD56, and CD16 (BD Biosciences, Oxford, UK). Dead cells and debris were excluded by dimensional gating (FSC/SSC). Total NK-cells were selected as CD3−/CD56+ lymphocytes. NK-cell subsets were identified by expression of CD56 and CD16 (cytotoxic NK-cells: CD56dim/CD16hi; cytokine-producing NK-cells: CD56bright/CD16dim). Data were acquired and analyzed on a FACS Canto II flow cytometer using FACS Diva 6 software (Becton Dickinson, Oxford, UK). In addition, NK analysis was performed in healthy controls. CXCL10-receptor analysis was not performed due to lack of sample availability.

Statistical Analysis

Data were analyzed using MS Excel 2007 and SPSS 18. Group comparisons based on clinical outcome of acute infection were performed by Mixed Model analysis for repeated measures incorporating all the timepoints of the longitudinal dataset. Individual timepoints were compared using a 2-tailed Mann-Whitney U-Test. Nominal factors were analyzed by the Chi-square test. Correlations were estimated by way of linear regression, unless otherwise specified. The statistical significance was set at P = 0.05.


Clinical Outcomes

Serial sample collection was performed in 16 patients presenting with acute HCV infection at 12 interval timepoints over 1 year. All patients were HCV-RNA-positive at recruitment. After 6 months we found that five patients underwent spontaneous viral clearance and 11 progressed to chronicity. In those who resolved infection, serum HCV-RNA levels temporally decreased and became undetectable after month 3; in subjects who progressed to chronicity viral load remained consistently high throughout the 1-year study period. Serum ALT levels followed similar kinetics and normalized after month 3 only in subjects who resolved the infection (Fig. 1). It is widely accepted that acute HCV patients who manifest symptoms will do so within 6-12 weeks of exposure.[16] The kinetics of decreasing viremia and ALT in spontaneous resolvers suggests a close correlation in the time course of infection postexposure within this group.

Figure 1.

Kinetics of serum HCV-RNA and ALT concentrations in subjects who became chronically infected (dark bars) or subjects who spontaneously resolved the infection (light bars). Bars represent levels of serum HCV-RNA; lines represent levels of serum ALT. The lower limit of detection (LLD) of HCV-RNA was 20 copies/mL (shaded areas). The upper limit of normal (ULN) for ALT values was 35 IU/L for women (dotted line) and 40 IU/L for men (dashed line). Values are mean ± standard error.

Age, gender, HCV genotype, and baseline serum HCV-RNA (P = 0.43) and ALT (P = 0.47) were comparable in the two groups. The IL28B rs12979860 CC/CT genotypes were equally distributed between groups and were not used as discriminant factors for subsequent analyses. The IL28B TT genotype was not represented in this cohort. Table 1 summarizes the baseline characteristics of the patients.

Lower Plasma Concentrations of Truncated CXCL10 and Decreasing DPP4 Activity Are a Feature of Spontaneously Resolving HCV Infection

In this longitudinal cohort of acute HCV subjects, the plasma concentration of total CXCL10 (long and truncated forms) was significantly lower in patients who spontaneously resolved the infection compared to patients who became persistently infected (P < 0.001) (Fig. 2A). While the levels of biologically active long CXCL10 did not differ between the two cohorts (P = 0.46) (Fig. 2B), we did observe significantly lower levels of the circulating truncated CXCL10 antagonist in patients who cleared the infection (P = 0.036) (Fig. 2C). Furthermore, the concentration of truncated CXCL10 was significantly correlated with viral replication and liver inflammation (HCV-RNA: r = 0.40, P < 0.001; ALT: r = 0.52, P < 0.001).

Figure 2.

Longitudinal evaluation of concentrations of total (A), long (B) or truncated (C) CXCL10, and measurement of DPP4 activity (D) in plasma samples of subjects who became chronic (dark bars) or subjects who spontaneously resolved the infection (light bars). Overall comparisons between groups were obtained by Mixed Model analysis for repeated measures incorporating all the timepoints of the longitudinal dataset. Comparisons at the same timepoint were evaluated by 2-tailed Mann-Whitney U-test. Values are mean ± standard error.

The antagonistic form of CXCL10 is generated by proteolytic cleavage performed by the protease DPP4. We therefore hypothesized that differences in DPP4 activity could be associated with the different concentrations of truncated CXCL10 and with the clinical outcome of infection. Indeed, DPP4 activity progressively decreased over time in patients who spontaneously cleared the infection, but remained consistently high in those who developed chronicity (P = 0.06) (Fig. 2D). After week 3, this difference became significant (P = 0.021).

Furthermore, DPP4 activity was correlated with concentrations of CXCL10 antagonist (r = 0.53, P < 0.001), while no correlation was observed with the biologically active long CXCL10 (r = 0.14, P = 0.13). DPP4 activity was also correlated with HCV-RNA (r = 0.50, P < 0.001) and ALT (r = 0.68, P < 0.001).

Spontaneous Resolution of Acute HCV Infections Correlates With the Early Development of Strong HCV-Specific IFNγ T-Cell Responses

The main biological role of CXCL10 is the recruitment and activation of host antiviral immune effector cells to the site of infection. This chemokine preferentially targets T-cells and NK-cells, both pivotal in the control of HCV replication.[2-5, 18] In patients who spontaneously resolved the infection the baseline frequency of HCV-specific IFNγ-producing T-cells was ∼6-fold higher than in patients who developed chronicity (P = 0.017) and remained markedly higher at week 1 (P = 0.017) and week 2 (P = 0.017) (Fig. 3A). Conversely, the presence of anti-inflammatory antigen-specific IL10-producing T-cells was associated with the development of viral persistence and significantly higher frequencies were observed from week 3 when compared to patients who were able to continue to resolution (P = 0.017) (Fig. 3B).

Figure 3.

Enumeration of HCV-specific IFNγ (A) and IL10 (B) production by ELISpot. All counts of spot-forming-cells (SFCs) were normalized to 106 PBMCs. Dark lines represent subjects who became chronic, light lines represent subjects who spontaneously resolved the infection. Comparisons between groups were evaluated by 2-tailed Mann-Whitney U-test. Comparisons between consecutive timepoints within the same study group were evaluated by Wilcoxon paired rank test. Values are mean ± standard error.

We have previously shown that the frequency of cytokine-producing HCV-specific T-cells in healthy controls is negligible and narrowly focused to HCV antigens and peptides.[4, 19]

Cytotoxic NK-Cells Are Associated With Spontaneous Resolution of the Acute HCV Infection

While T-cells represent the main effector cells of the adaptive immunity, NK-cells are key effectors of the innate immunity and represent the first line of immune defenses against infections. Proportions of total NK-cells (CD3-/CD56+) were comparable in both patient groups (Fig. 4B). Based on well-established criteria, we assessed the proportions of cytotoxic CD56dim and cytokine-producing CD56bright NK-cells[20] (Fig. 4A). Patients who spontaneously resolved the infection had more cytotoxic CD56dim NK-cells (P = 0.005) and less cytokine-producing CD56bright NK-cells (P = 0.020) than patients who became chronically infected (Fig. 4B). In subjects who spontaneously cleared HCV, the progressive reduction of HCV-RNA was correlated with an increasing CD56dim/CD56bright NK-cell ratio (r = −0.997, P = 0.051) (Fig. 4C). Interestingly, this ratio was restored to levels comparable with healthy controls by month 6 (Supporting Fig. 1), when HCV-RNA levels were undetectable. In patients who developed chronicity, serum levels of HCV-RNA at baseline (r = −0.68, P = 0.030) and at month 6 (r = −0.63, P = 0.037) were negatively correlated with the CD56dim/CD56bright ratio. Proportions of unfavorable CD56bright NK-cells in patients who became persistently infected were also directly correlated with liver inflammation (ALT, month 1; r = 0.89, P = 0.001) (Fig. 4D).

Figure 4.

Evaluation of NK-cell subsets and correlation with clinical data. NK-cells were identified by flow cytometry as CD3−/CD56+ lymphomonocytes (A). Proportions of NK-cells with bright or dim expression of CD56 were measured, and represented as line graphs in (B). Dark lines represent subjects who became chronic, light lines represent subjects who spontaneously resolved the infection. Values are mean ± standard error. No differences were observed between timepoints within the same study group, but overall differences in relative proportions of CD56dim and CD56bright NK-cells were found between groups. Among spontaneous resolvers, the temporal reduction of HCV-RNA correlated with increasing predominance of cytotoxic CD56dim NK-cells (C, light squares. Power regression. Each square is the mean ± standard error of all the values for each timepoint). Even in subjects who became chronic lower levels of HCV-RNA and ALT correlated with higher ratios of CD56dim over CD56bright NK-cells (D, dark diamonds).

Furthermore, we found that truncated CXCL10 was directly correlated with frequencies of unfavorable cytokine-producing CD56bright NK-cells (r = 0.92, P < 0.001) and inversely correlated with frequencies of favorable cytotoxic CD56dim NK-cells (r = −0.84, P = 0.004) (Fig. 5).

Figure 5.

CXCL10 antagonism alters relative proportions of NK-cell subsets. Higher concentrations of truncated CXCL10 always correlated with increased proportions of unfavorable cytokine-producing CD56bright NK-cells and decreased proportions of favorable cytotoxic CD56dim NK-cells.

IL15 acts as a differentiation factor for NK-cells[20, 21] and we observed a direct correlation between plasma concentrations of truncated CXCL10 and circulating IL15 (r = 0.48, P = 0.021) (Supporting Fig. 2). The concentration of IL15 in our cohorts was similar to those previously reported[22-24] but, as with these studies, the majority of detected values fell below the lowest standard of the assay (3.9 pg/mL). In subjects who progressed to chronicity, higher serum levels of IL15 were correlated with higher viral load at baseline (r = 0.77, P = 0.009) and liver inflammation (ALT) at month 1 of follow-up (r = 0.66, P = 0.039) (Supporting Fig. 2). Greater levels of IL15 were also correlated with higher baseline proportion of CD56bright NK-cells in patients who became chronic (r = 0.71, P = 0.021) (data not shown). These data may suggest that CXCL10 antagonism may play a role in inducing alterations in NK-cell development and differentiation.


This is the first report highlighting the role of CXCL10 antagonism in the clinical evolution of acute HCV infection. Elevated total CXCL10 levels are associated with poor treatment outcomes and the establishment of chronic infection after exposure to HCV.[6, 7, 11, 25, 26] However, there has been a paucity of understanding regarding the reasons for these paradoxical observations. The findings presented in this study provide the first body of evidence that the presence of CXCL10 antagonist is directly associated with the course and outcome of acute infection. By differentially quantifying the long and the truncated forms of CXCL10 and assessing antiviral T-cell and NK functions at standardized timepoints in all patients, we could demonstrate that the chronic evolution of the acute HCV infection is associated with consistently higher concentrations of the short CXCL10 antagonist and the failure of a robust antiviral immune response.

This truncated form of CXCL10 is generated by DPP4 and alterations of truncated CXCL10 could derive from upstream changes in the biological activity of soluble DPP4. Indeed, the observation of differential kinetics of DPP4 activity in our longitudinal plasma samples highlights a higher level of dysregulation imposed by HCV, whereby consistently high levels of viral replication and liver inflammation, as observed in patients who progress to chronicity, correlate with persistently high plasma DPP4 activity. This, in turn, may account for the increased truncation of CXCL10 and generation of the antagonist form found in patients who became chronically infected after the acute phase of infection. Interestingly, high baseline plasma concentrations of DPP4 were also recently associated with poorer treatment outcome and altered HCV-specific T-cell functionality in a cohort of patients with established cHCV treated with Peg-IFNα plus ribavirin.[27] Taken together, these data suggest a link between viral replication, liver inflammation, and regulation of the CXCL10 axis by DPP4. Studies have also described a role for DPP4 in the formation of lipid rafts during HCV replication and DPP4 knockdown in HCV replicon cells results in the suppression of HCV replication and a reduction in the production of viral particles.[28] These data suggest that therapeutic abrogation of DPP4 activity, by commercially available inhibitors such as sitagliptin,[29] may be a novel strategy to target both the virus and the host.

The main biological role of the chemokine CXCL10 is the recruitment and activation of host antiviral immune effectors such as T-cells and NK-cells[2-5, 18] to control the spread of infection. We therefore hypothesized that the different kinetics of DPP4 activity and magnitudes of CXCL10 antagonism observed between spontaneous resolvers and subjects who became chronically infected would be associated with differences in the antiviral activity of NK-cell and T-cell compartments. We show here that failure to clear acute HCV infection was associated with the absence of early virus-specific T-cell-mediated IFNγ immunity and subsequent development of anti-inflammatory IL10 responses. While patient numbers in this study were small, we speculate that the higher levels of truncated CXCL10 antagonist observed in patients who evolve to chronicity impede the early activation and development of favorable HCV-specific IFNγ T-cell responses. In the absence of antiviral IFNγ, the high viral replication and antigenemia rates would favor expansion of immunosuppressive anti-inflammatory IL10 responses,[30] subsequently culminating in an unfavorable host immune environment and persistence of infection.

Together with T-cell-mediated IFNγ responses, our study shows that a predominance of cytotoxic CD56dim NK-cells may also favor the spontaneous resolution of the acute HCV infection and that the greater the predominance of these cytotoxic NK-cells, the lower the levels of viral replication and liver inflammation. This finding is in line with the observation that higher proportions of cytotoxic CD56dim NK-cells are associated with sustained virological response in a prospective cohort of cHCV patients treated with Peg-IFNα plus ribavirin,[31] suggesting that the ability of NK-cells to kill HCV-infected hepatocytes is of greater importance than their ability to produce cytokines for the control of HCV replication. This is further supported by the notion that CD56bright NK-cells are stronger cytokine producers and are more involved in immune regulation, thus also having a possible role in limiting host-driven immunopathology.[20]

The observed correlations between CXCL10 antagonist and NK-cell subsets suggest that CXCL10 antagonism may have a role in the collapse of the host immunity favoring the persistence of HCV replication. Indeed, the altered equilibrium between NK-cell subsets observed in subjects who become chronic may be directly influenced by their defective CXCL10 signals. Unfortunately, we were unable to analyze the expression of the CXCL10-receptor (CXCR3) on NK and T-cell populations due to sample availability; however, it has been previously shown that CD56bright NK-cells express higher levels of CXCR3 and are therefore more sensitive, and show a stronger migratory response, to CXCL10 than CD56dim NK-cells.[20, 32, 33] In the presence of higher levels of CXCL10 antagonist, CD56bright NK-cells might fail to migrate to the infected liver and accumulate instead in the peripheral circulation, as we have observed in patients who developed chronicity. This might also result in a defective maturation of new functional CD56dim NK-cells at the site of infection, consequently reducing the NK-mediated elimination of HCV-infected hepatocytes. In fact, among the total circulating NK-cells, CD56bright cells are more immature NK precursors, whereas cytotoxic CD56dim cells are more mature NK effectors.[20, 21, 34]

IL15 plays a role as a differentiation factor for NK-cells[20, 21, 34, 35] and can induce the differentiation of CD56bright-like NK-cells from CD34+ progenitors.[21] In subjects who became persistently infected we observed higher levels of IL15 and higher baseline proportions of cytokine-producing CD56bright NK-cells, directly correlated with concentrations of CXCL10 antagonist and absence of a subsequent enrichment of the CD56dim subset, suggesting a developmental defect in the NK differentiation pathway.

The NK subset imbalance observed in subjects who failed to eradicate the infection, with low predominance of cytotoxic CD56dim NK-cells, has been described in various viral infections, including HIV, HBV, and also HCV,[22, 36, 37] together with the link between CD56dim NK-cells and serum concentrations of IL15 in established cHCV.[22] However, this is the first report to highlight a link between NK-cells and IL15 in the chronic evolution during the acute phase of infection and, importantly, to suggest that CXCL10 antagonism may favor defects in these differentiation and tissue homing pathways.

In summary, we believe that this study provides the first body of evidence in support for a role of chemokine antagonism in the evolution of an acute viral infection. We show that a dysfunctional DPP4-CXCL10 axis inhibits the development of favorable innate and adaptive host immunity to HCV and favors the evolution of viral persistence.

Author Contributions

Antonio Riva: Performed experiments, analyzed results, cowrote the article; Melissa Laird: Performed analyses and developed CXCL10 assay; Armanda Casrouge: Performed analyses and developed CXCL10 assay; Arvydas Ambrozaitis: Recruited patients, provided samples and study design; Roger Williams: Preparation of article, data analysis; Nikolai V. Naoumov: Experimental design, article preparation; Matthew L. Albert: Developed CXCL10 assay, performed data analysis and article preparation; Shilpa Chokshi. Designed and coordinated the study, performed experiments, analyzed results, cowrote the article.