Potential conflict of interest: Nothing to report.
Serum amyloid A (SAA) is an acute phase protein produced by the liver. SAA concentration increases markedly in the serum following inflammation and infection. Large increases in SAA concentration during the acute phase response suggest that SAA has a beneficial role in host defense. This study sought to determine the effect of SAA on hepatitis C virus (HCV) infectivity using retroviral particles pseudotyped with HCV envelope glycoproteins (HCVpp) and the recently developed cell culture system for HCV (HCVcc). SAA inhibited HCVpp and HCVcc infection in a dose-dependent manner by affecting an early step of the virus life cycle. Further characterization with HCVpp indicated that SAA blocks virus entry by interacting with the viral particle. In addition, the antiviral activity of SAA was strongly reduced when high-density lipoproteins (HDL) were coincubated with SAA. However, HDL had only a slight effect on the antiviral activity of SAA when HCVpp was first preincubated with SAA. Furthermore, analyses of SAA in sera of chronic HCV patients revealed the presence of variable levels of SAA with abnormally elevated concentrations in some cases. However, no obvious clinical correlation was found between SAA levels and HCV viral loads. In conclusion, our data demonstrate an antiviral activity for SAA and suggest a tight relationship between SAA and HDL in modulating HCV infectivity. (HEPATOLOGY 2006;44:1626–1634.)
Hepatitis C virus (HCV) is a major cause of chronic liver disease, with approximately 170 million carriers worldwide. HCV-associated liver disease frequently progresses to cirrhosis and hepatocellular carcinoma. There is no vaccine to prevent HCV infection, and the current drug therapies are effective in only a fraction of patients. Moreover, these treatments are often poorly tolerated.
For a long time, it has been difficult to study the early steps of HCV life cycle because of the difficulties in propagating this virus in cell culture. A few years ago, a major step in investigating HCV entry was achieved by the development of pseudotyped particles (HCVpp), consisting of HCV glycoproteins E1 and E2 assembled onto retroviral core particles.1–3 More recently, several laboratories also reported the development of a cell culture system that leads to a relatively efficient amplification of HCV (HCVcc), which allows for the first time the study of HCV life cycle.4–6 Studying entry with HCVpp and HCVcc is shedding some light on some of the mechanisms as well as on the role of some cell surface molecules involved in virus entry.
The human scavenger receptor class B type I (SR-BI)7 has been shown to interact with E2,8 and it has been shown that SR-BI is involved in HCVpp entry (reviewed in Cocquerel et al.9). However, the exact role of SR-BI in HCV entry is not clear. Interestingly, when HCVpp is incubated in the presence of high-density lipoprotein (HDL), a physiological ligand of SR-BI, infectivity is increased rather than reduced.10–12 In addition, HDL-mediated facilitation of HCVpp entry depends on the lipid transfer property of SR-BI,10, 12 suggesting that HCV exploits the physiological activity of SR-BI for promoting its entry into target cells.
Recently, SR-BI has been shown to bind and internalize serum amyloid A (SAA).13, 14 SAA is an acute phase protein mainly produced by the liver immediately after infection, tissue damage or inflammation (reviewed in Uhlar et al.15). The high concentration of SAA during the acute phase response suggests that this protein has a beneficial role in host defense. However, the precise role of SAA in host defense during inflammation remains to be determined. Because SAA is a ligand of SR-BI and SR-BI is involved in HCV entry, we sought to determine the effect of SAA on HCV entry. In contrast to HDL, our data indicate that SAA inhibits HCV entry in a dose-dependent manner. However, analyses with HCVpp indicate that SAA inhibits entry by interacting with the particles instead of competing for a receptor. In addition, in some conditions, HDL modulated the antiviral activity of SAA, suggesting a tight relationship between SAA and HDL in modulating HCV infectivity. This report demonstrates an antiviral activity for SAA.
SAA, serum amyloid A; HCV, hepatitis C virus; HCVpp, retroviral particles pseudotyped with HCV envelope glycoproteins; HCVcc, HCV produced in cell culture; HDL, high-density lipoprotein; SR-BI, scavenger receptor class B type I; VSV-G, vesicular stomatitis virus G protein; LPDS, lipoprotein-depleted serum.
Materials and Methods
Recombinant human SAA corresponding to SAA1α (except for the presence of an N-terminal methionine and substitution of asparagine for aspartic acid at position 60 and arginine for histidine at position 71) was purchased from Biodesign International Inc. (Saco, ME). Anti-HCV monoclonal antibodies (mAbs) A4 (anti-E1)16 and 3/11 (anti-E2; kindly provided by J. McKeating, University of Birmingham, UK)17 and the anti-VSV-G mAb P5D418 have been described. We purchased mAbs against SR-BI and actin from BD Biosciences and Santa Cruz Biotechnology, respectively. The anti-SAA mAb and the SAA ELISA kit were obtained from Anogen (Mississauga, Canada).
Lipoproteins and Lipoprotein-Depleted Serum Preparation.
Human HDL3 (density: 1.13-1.18 g/mL) fractions from fresh human plasma and lipoprotein-depleted serum (LPDS) (density: >1.25 g/mL) from fetal bovine serum were isolated by KBr density gradient ultracentrifugation as previously described.19
293T human embryo kidney cells (HEK293T), Huh-7 human hepatoma cells20 were grown in Dulbecco's modified essential medium (DMEM; Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum.
Production of HCVpp and HCVcc.
HCVpp was produced as described,1, 21 with plasmids kindly provided by B. Bartosch and F. L. Cosset (INSERM U758, Lyon, France). Plasmids encoding HCV envelope glycoproteins of genotypes 1b (UKN1B-5.23), 2b (UKN2B-1.1), 3a (UKN3A-1.28), and 4 (UKN4-11.1) were kindly provided by J. Ball (University of Nottingham, UK).22 The genotype 1a plasmid (strain H) has been previously described,1 and the genotype 2a plasmid (strain JFH-1) was kindly provided by T. Pietschamnn and R. Bartenschlager (University of Heidelberg, Germany). For the production of VSV-Gpp and RD114-pp, plasmids encoding the vesicular stomatitis virus G (VSV-G) protein and the feline endogenous virus RD114 glycoprotein23 were used. The firefly luciferase was used as a reporter gene to monitor virus entry as described.21
The plasmid pJFH1,5 containing the full-length cDNA of JFH1 isolate and kindly provided by T. Wakita (Tokyo Metropolitan Institute for Neuroscience, Japan) was used to generate HCVcc as described.24 HCVcc stock used in our experiments was produced in DMEM containing 5% LPDS.
To analyze SAA-HCVpp interactions, HCVpp were preincubated in the presence or absence of SAA (15 μg/mL) for 3 hours at 37°C. SAA was then immunoprecipitated with the anti-SAA mAb. Coimmunoprecipitated proteins as well as SAA were revealed by Western blotting with specific mAbs and by enhanced chemiluminescence (ECL) detection (GE Healthcare–Amersham Biosciences, Uppsala, Sweden).
HCVpp Binding Assay.
Huh-7 cells treated with 0.1% sodium azide were incubated for 2 hours at 37°C with HCVpp. Cells were then washed with phosphate-buffered saline (PBS), and cell-bound particles were cross-linked to the cell surface for 30 minutes at room temperature with 2 mmol/L DTSSP in PBS. Addition of a solution of 50 mmol/L Tris-HCl at pH 7.5 stopped the cross-linking reaction. Cells were then detached by scraping and lysed with 1% Triton X-100 in PBS. The presence of HCVpp bound to the cell surface was determined by Western blotting.
SAA Inhibits HCV Entry Into Huh-7 Cells.
To avoid any interference with lipoproteins, the infections were performed in the presence of LPDS. Interestingly, the presence of SAA during infection reduced HCVpp infectivity in a dose-dependent manner (Fig. 1A). Infectivity was reduced to about 10% at a concentration of 20 μg/mL; however, higher concentrations did not further reduce HCVpp infectivity (data not shown). Inhibition of HCVpp entry was not genotype-specific because similar levels of inhibition were observed for other HCV genotypes (Fig. 1B). Importantly, this inhibition was dependent on the presence of HCV envelope glycoproteins because SAA did not reduce the infectivity of pseudotyped particles containing the envelope glycoproteins of the feline endogenous virus RD114 (RD114pp) or VSV-G (VSV-Gpp) (Fig. 1A). In contrast to HCVpp, the level of VSV-Gpp infectivity was slightly increased in the presence of SAA. Together, these data indicate that SAA specifically inhibits HCVpp entry.
To further confirm the relevance of our observation, we analyzed the effect of SAA on HCVcc, produced by the recently developed cell culture system for HCV.4–6 HCVcc infectivity was reduced in a dose-dependent manner (Fig. 1C). Furthermore, SAA inhibited HCVcc infection by affecting an early step of the life cycle. Indeed, removal of SAA after virus adsorption did not reduce its inhibitory effect, and adding SAA after virus adsorption did not alter HCVcc infectivity (Fig. 1D). Finally, apolipoprotein A-I and apolipoprotein E had no effect on HCVcc and HCVpp infectivity (Supplementary Fig. 1; Supplementary material for this article can be found on the HEPATOLOGY website (http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html), indicating that the anti-HCV effect is specific to SAA.
Altogether, these data demonstrate the inhibitory effect of SAA on HCV entry. In addition, the data obtained with HCVcc validate the HCVpp system to further study the role of SAA in HCV entry.
Antiviral Activity of SAA Is Not Due to Its Action on Target Cells.
Since it has recently been shown that SAA can be internalized in an intracellular compartment via SR-BI,13 we analyzed whether SAA might potentially modify the level of SR-BI expression at the plasma membrane. However, as determined by cell surface biotinylation, we did not detect any change in cell surface expression of SR-BI after preincubation with SAA (data not shown), indicating that SAA does not inhibit HCV entry by reducing the pool of SR-BI present at the plasma membrane.
It has also recently been reported that SAA can promote cholesterol efflux mediated by SR-BI.25 Because the lipid-transfer function of SR-BI has been shown to affect HCVpp infectivity in the presence of HDL,10, 12 we wondered whether SAA-mediated cholesterol efflux might play a role in inhibition of HCV entry. If SAA-mediated cholesterol efflux is responsible for entry inhibition, treating target cells with BLT1, a drug that inhibits SR-BI-dependent cholesterol efflux,26 should therefore restore HCVpp infectivity when infection is performed in the presence of SAA. However, in the presence of SAA, HCVpp infectivity was strongly inhibited despite BLT1 addition to the cells (data not shown), indicating that BLT1 treatment does not restore HCVpp infectivity. These data therefore suggest that the antiviral activity of SAA is not related to its ability to induce lipid exchanges.
Although SAA had no effect on the level of expression of SR-BI at the plasma membrane and it did not seem to inhibit HCVpp entry by promoting cholesterol efflux, we could not exclude an effect on other properties of the target cells. To investigate whether SAA inhibits entry by affecting target cells, Huh-7 cells were preincubated with SAA before infection at 37°C and SAA was removed during virus adsorption. To increase the chances of maintaining the SAA effect during virus entry, adsorption was reduced to 15 minutes in this experiment. As shown in Fig. 2A, preincubation of Huh-7 cells with SAA for 3 hours at 37°C had no effect on HCVpp entry, suggesting that SAA does not affect the sensitivity of target cells to infection by HCVpp. Similar results were obtained with HCVcc (Supplementary Fig. 2).
Because SR-BI plays a role in HCVpp entry and SAA is a ligand for SR-BI, it was important to determine whether SAA would have antiviral activity by competing with HCVpp for binding to the plasma membrane. To test this hypothesis, Huh-7 cells were preincubated with SAA at 4°C, a temperature at which only binding takes place. After SAA binding, the cells were washed and HCVpp was adsorbed at 4°C. Infected cells were then incubated at 37°C for 2 days. In these conditions, infectivity should be reduced if SAA competes with HCVpp for binding to SR-BI. However, when cells were preincubated with SAA, we did not observe any reduction in HCVpp infectivity, suggesting that SAA does not compete with HCVpp for binding to Huh-7 cells (Fig. 2B, compare −SAA with +SAA before HCVpp binding). Interestingly, similar results were obtained with HCVcc (Supplementary Fig. 2). Because both SAA and HCV glycoprotein E2 have been shown to bind to SR-BI, we also tested whether SAA would inhibit the binding of soluble E2 to SR-BI. However, our experiments did not show any inhibition of E2 binding to SR-BI in the presence of SAA (Supplementary Fig. 3).
Altogether, these data indicate that the antiviral activity of SAA is not due to its action on target cells.
Antiviral Activity of SAA Is Due to Its Action on the Viral Particle.
To test whether the anti-HCV activity of SAA is due to an effect on viral particles, SAA and HCVpp were incubated before contact with target cells. In these conditions, SAA should have more time to bind to HCVpp and hence have a stronger antiviral activity. In contrast, the antiviral activity of SAA should not be modified in these conditions if SAA does not interact with the particles. The anti-HCV activity of SAA was greater when SAA was preincubated with HCVpp (Fig. 3A), suggesting that SAA interacts directly with HCVpp and thereby reduces their infectivity. Interestingly, similar results were obtained with HCVcc (Fig. 3B). This observation is an argument in favor of SAA acting on the particles instead of the target cells. When comparing Figs. 3A and 1A, the antiviral activity of SAA was slightly lower in Fig. 3A. This is likely due to differences in the experimental conditions. The preincubation of SAA at 37°C can indeed potentially lead to some degradation or aggregation of SAA.
To further investigate the action of SAA on HCVpp, we sought to determine whether SAA interacts with HCVpp in coimmunoprecipitation experiments. After immunoprecipitation with an anti-SAA antibody, the presence of HCVpp in the immunoprecipitates was analyzed by Western blotting with HCV-specific mAbs. HCV envelope glycoproteins were detected after immunoprecipitation with the anti-SAA antibody (Fig. 4), indicating that HCVpp can be coimmunoprecipitated with SAA. These data are in favor of an interaction between HCVpp and SAA. Moreover, this interaction was specific of the presence of HCV envelope glycoproteins in the pseudoparticles, because particles containing the envelope glycoprotein G of VSV (VSV-Gpp) were not coimmunoprecipitated with SAA (Fig. 4). This suggests that SAA interacts directly with HCV envelope glycoproteins. However, when HCVpp were pretreated with Triton X-100 before immunoprecipitation with the anti-SAA antibody, no precipitation of HCV envelope glycoproteins was detected (data not shown), suggesting that mild detergents disrupt the interactions between SAA and HCV envelope glycoproteins. It is worth noting that only a small fraction of HCV envelope glycoproteins was coimmunoprecipitated with SAA (Fig. 4), suggesting that the affinity of SAA for HCV envelope glycoproteins is weak. Alternatively, as discussed below, only a small fraction of HCVpp might be competent for infectivity and only this fraction might interact with SAA. Due to the low titers of virus produced in cell culture, it was not possible to study the interaction between SAA and HCVcc in a biochemical assay. It is also important to note that the current approaches to study viral particles isolated from patients do not allow unequivocal analysis of a physical interaction between native particles and SAA in a biochemical assay.
SAA bound to HCVpp might inhibit virus attachment to the cell surface. To test this effect, Huh-7 cells were coincubated with HCVpp and SAA for 90 minutes at 4°C and, after washing, the cells were incubated at 37°C for 2 days. Interestingly, in these conditions, HCVpp infectivity was reduced to ≈40% (Fig. 2B, compare −SAA with +SAA during HCVpp binding). Because SAA addition before adsorption did not affect HCVpp infectivity (Fig. 2B, compare −SAA with +SAA before HCVpp binding), this observation suggests that SAA binds to HCVpp and inhibits viral attachment to target cells.
To further study the effect of SAA on HCVpp binding, we analyzed the binding of HCVpp in the presence or absence of SAA. To determine the binding of HCVpp to target cells, pseudotyped particles were incubated with Huh-7 cells for 2 hours at 37°C in the presence of sodium azide, an inhibitor of endocytosis. After virus attachment, cells were washed and the presence of bound virus was revealed by Western blotting with anti-HCV mAbs. The amount of HCV glycoproteins associated with the target cells was reduced to ≈30% in the presence of SAA (Fig. 5). The presence of SAA did not affect the binding of VSV-Gpp to Huh-7 cells (data not shown), indicating that SAA-mediated reduction of binding is specific to the presence of HCV glycoproteins in the particles. It is worth noting that only a small fraction of the input virus was bound to the Huh-7 cells (Fig. 5), suggesting that a large proportion of HCVpp might not be functional for binding. This is in agreement with the large heterogeneity of HCV glycoproteins incorporated into HCVpp.27 These data indicate that SAA reduces HCVpp binding to Huh-7 cells.
Altogether, these data indicate that the antiviral activity of SAA is due to its effect on the particle.
Antiviral Activity of SAA Can Be Attenuated in the Presence of HDL.
SAA is present in human sera, so we wanted to determine whether serum would modulate the activity of SAA. The antiviral activity of SAA was therefore tested in the presence of normal human serum. As described,10 the addition of 1% normal human serum during infection resulted in approximately a 3-fold increase of HCVpp infectivity (Fig. 6A).Although the addition of SAA at a concentration of 15 μg/mL inhibited HCVpp infectivity by more than 90% in the presence of LPDS, this antiviral effect was strongly attenuated in the presence of human serum. In these conditions, infectivity was indeed only reduced by 30%. Thus, the antiviral activity of SAA seems to be partly compensated by component(s) of the human serum when both SAA and human serum are coincubated with HCVpp during virus adsorption.
Because HDL has been shown to be the major serum component responsible for enhancement of HCVpp infectivity,10–12 we also tested the antiviral activity of SAA in the presence of purified HDL. The inhibitory effect of SAA was strongly attenuated in the presence of this lipoprotein during infection (Fig. 6B). At the concentration of 20 μg/mL of SAA, infectivity was only reduced by ≈50% in the presence of HDL. These data indicate that the antiviral activity of SAA is reduced in the presence of HDL in human serum. However, when HCVpp were preincubated with SAA before infection in the presence of HDL, the antiviral effect of SAA was much less affected (Fig. 6B), suggesting that SAA has a dominant inhibitory effect when added before HDL.
SAA Is Detected in Sera of Some Patients Chronically Infected With HCV.
Because SAA has antiviral activity against HCV, it was important to determine whether SAA is produced during HCV infection. We therefore tested SAA concentration in the sera of a series of chronically infected patients. We tested 94 patients (51 ± 1.5 years, male:female 53:41) chronically infected with HCV for the presence of SAA in their sera (Supplementary Table 1). The HCV genotype was determined by direct sequencing of the 5′ noncoding region (1: n = 61, 2: n = 10, 3: n = 17, 4: n = 3, 5: n = 3), and the viral loads were determined by an in-house TaqMan assay.28 They were expressed as the logarithmic values of IU/mL of HCV RNA (6.193 ± 0.085 log IU/mL). The usual concentration of SAA in the sera of normal patients is estimated to be below 5 μg/mL. In patients chronically infected by HCV, the SAA concentration was higher than 5 μg/mL in 16 of them. Statistical analyses were performed using the Statview software. Linear regression analyses did not show any statistically significant correlation between SAA values and viral loads considering all HCV-infected patients or only those infected by HCV genotype 1. Moreover, SAA was investigated in sera of 28 patients with chronic HBV infection (45 ± 2.9 years, male:female 20:8) (Supplementary Table 2). The viral load was determined by Cobas TaqMan HBV test (Roche Diagnostics, Meylan, France). The SAA values obtained from patients with chronic HBV infection did not differ significantly from those of HCV-infected patients, and no correlation was found between the SAA values and the HBV viral load in HBV-infected patients.
The acute phase response is the immediate set of host inflammatory reactions that counteract challenges such as tissue injury, infection, and trauma.15 One of the systemic responses to an acute inflammatory stimulus is the alteration in the hepatic biosynthetic profile of acute phase proteins.29, 30 SAA belongs to this group of acute phase proteins whose concentration increases by as much as 1000-fold during inflammation.15 However, the precise role of SAA in inflammation remains to be clarified. Here, we show that SAA has an antiviral activity against HCV by affecting virus entry. In addition, in some conditions, HDL modulated the antiviral activity of SAA, suggesting a tight relationship between SAA and HDL in modulating HCV infectivity. Furthermore, analyses of SAA in sera of chronic HCV patients revealed the presence of variable levels of SAA with abnormally elevated concentrations in some cases, suggesting that, in vivo, HCV entry might be affected by the inhibitory effect of SAA.
During the acute phase response, SAA synthesis can account for as much as 2.5% of total protein production in the liver and plasma levels can reach concentrations of more than 1 mg/mL. Such a high concentration during the acute phase response suggests that SAA has a beneficial effect in host defense. However, the precise role of SAA in host defense during inflammation remains poorly understood. SAA might play a role in host defense by being involved in lipid metabolism (reviewed in Uhlar et al.15). Indeed, SAA shares many structural features with other apolipoproteins, including amphipathic alpha helices that may be responsible for its binding to HDL. SAA association with HDL suggests that this protein might play a role in regulating HDL reverse cholesterol transport, by promoting cholesterol removal from sites of tissue damage.31 Because SAA is able to bind cholesterol, it has also been suggested that SAA might promote cholesterol delivery to cells during tissue repair.15
In addition to its role in lipid metabolism, SAA might also have a direct effect on pathogens. Indeed, SAA has been shown to bind to the outer membrane protein A of a large number of Gram-negative bacteria and to be involved in opsonization of these bacteria.32, 33 The observations that SAA binds to HCV and has an antiviral activity against this virus provide experimental evidence that SAA can have a direct effect against pathogens. This observation is in favor of a role for SAA in the innate immune response against HCV. However, the antiviral activity of SAA is probably restricted to HCV or to a limited number of viruses. Indeed, we did not detect any antiviral activity against several other positive-stranded RNA viruses like yellow fever virus, Sindbis virus, and bovine viral diarrhea virus (data not shown). Furthermore, pseudotyped particles containing the envelope protein of VSV or RD114 were not affected by SAA. Therefore, we cannot exclude that the interaction between SAA and HCV might be the result of an adaptative evolution of this virus to reduce its infectivity in the environment of the liver. This might potentially reduce HCV pathogenicity and keep its host in healthy conditions for a longer time. Some viruses have indeed developed strategies that reduce their virulence or that control an inappropriate immune response which could be harmful for the host, as shown for vaccinia virus.34
The antiviral activity of SAA can be attenuated by the presence of HDL in human serum. SAA is a ligand of SR-BI, which can potentially inhibit HDL binding, selective uptake, and cholesterol efflux.14, 25 Moreover, HDL-mediated facilitation of HCVpp entry depends on the lipid transfer properties of SR-BI. Therefore, one would have expected that SAA would inhibit the HDL effect on HCVpp entry. On the other hand, SAA is an apolipoprotein and we cannot exclude that a proportion of SAA would associate with HDL, which would attenuate the inhibitory effect of SAA on HCVpp infectivity. In addition, SAA associated with HDL might not be able to attenuate HDL-mediated facilitation of HCVpp entry. However, when HCVpp were preincubated with SAA before infection in the presence of HDL, the antiviral effect of SAA was maintained, suggesting that SAA has a strong anti-HCV activity if it encounters an HCV particle before coming in contact with HDL. Because SAA and HCV particles are produced in the hepatocyte, it is likely that SAA has the opportunity to be in contact with HCV before being affected by HDL.
Analyses of SAA in sera of chronic HCV patients revealed the presence of variable levels of SAA with abnormally elevated concentrations in some cases. However, no obvious clinical correlation was found between the SAA levels and the HCV viral loads. This suggests that SAA does not play a major role in the control of the viral infection in chronically infected patients. However, one cannot exclude a role for SAA during the acute phase of HCV infection, probably in association with other factors associated with the early innate immune responses. Further investigations will therefore be needed to determine the role of SAA during the early phases of HCV infection.
In conclusion, SAA has an antiviral activity against HCV by affecting virus entry. In addition, the potential attenuation of the antiviral activity of SAA by HDL suggests a tight relationship between SAA and HDL in modulating HCV infectivity.
We thank Laurence Cocquerel for helpful comments on the manuscript, Yves Rouillé for help in the experiments concerning non-HCV viruses, and André Pillez and Sophana Ung for technical assistance. We are grateful to T. Wakita, J. McKeating, J. Ball, B. Bartosch, FL. Cosset, T. Pietschamnn, and R. Bartenschlager for providing us with reagents.