Potential conflict of interest: Nothing to report.
The role of the humoral immune response in the natural course of hepatitis C virus (HCV) infection is widely debated. Most chronically infected patients have immunoglobulin G (IgG) antibodies capable of neutralizing HCV pseudoparticles (HCVpp) in vitro. It is, however, not clear whether these IgG can prevent a de novo HCV infection in vivo and contribute to the control of viremia in infected individuals. We addressed this question with homologous in vivo protection studies in human liver–urokinase-type plasminogen activator (uPA)+/+ severe combined immune deficient (SCID) mice. Chimeric mice were loaded with chronic phase polyclonal IgG and challenged 3 days later with a 100% infectious dose of the acute phase H77C virus, both originating from patient H. Passive immunization induced sterilizing immunity in five of eight challenged animals. In the three nonprotected animals, the HCV infection was attenuated, as evidenced by altered viral kinetics in comparison with five control IgG-treated animals. Plasma samples obtained from the mice at viral challenge neutralized H77C-HCVpp at dilutions as high as 1/400. Infection was completely prevented when, before administration to naïve chimeric mice, the inoculum was pre-incubated in vitro at an IgG concentration normally observed in humans. Conclusion: Polyclonal IgG from a patient with a long-standing HCV infection not only displays neutralizing activity in vitro using the HCVpp system, but also conveys sterilizing immunity toward the ancestral HCV strain in vivo, using the human liver–chimeric mouse model. Both experimental systems will be useful tools to identify neutralizing antibodies for future clinical use. (HEPATOLOGY 2008.)
Recent attempts to prevent hepatitis C virus (HCV) infections in vivo by passive immunizations have been disappointing and challenge the existence of a neutralizing humoral immune response against HCV.1–4 Conversely, evidence for the neutralizing capacity of anti-HCV positive immunoglobulin G (IgG) was obtained after in vitro incubation with HCV pseudoparticles (HCVpp) or infectious patient H serum and subsequent challenge of Huh-7 cells or naïve chimpanzees, respectively.5 Using similar in vitro assays, neutralizing antibodies have been detected in the sera of patients chronically infected with HCV.6–8 Two recent reports studying the impact of neutralizing antibodies on the resolution of acute HCV in different cohorts of accidentally exposed subjects suggest that rapid induction of neutralizing antibodies during the early phase of infection may contribute to control of HCV infection.9, 10 However, in vitro HCVpp neutralization studies make use of heterogeneous, nonphysiological viral-like particles, and liver-derived tumor cell-lines as target cells.11, 12 Furthermore, several factors restrict the use of the chimpanzee model for in vivo passive immunization studies, not the least the requirement for large quantities of IgG. Therefore, the in vivo relevance of antibodies capable of neutralizing HCVpp remains to be defined.
The human liver–uPA–severe combined immune deficient (SCID) mouse harbors fully functional human hepatocytes, is susceptible to infection with native HCV, and subsequently produces highly infectious HCV virions of physiological density.13–15 Therefore, it represents an attractive preclinical model to perform passive immunization studies. We performed mouse protection studies by loading human liver–uPA–SCID mice with chronic phase polyclonal IgG and subsequently challenging them with a 100% infectious dose of the homologous H77C cloned virus, both isolated from patient H. We demonstrated that polyclonal antibodies derived from the serum of a patient can convey in vivo protection against the cloned ancestral HCV virus. The HCVpp system identified antibodies that also have in vivo neutralizing potential, but no quantitative correlate of in vivo protection could be defined. Our observations suggest that passive immunization against HCV infection may be achieved in a clinical setting. The HCVpp assay and the chimeric mouse model will be useful tools to identify monoclonal or polyclonal antibodies with optimal protective qualities.
HCV, hepatitis C virus; HCVpp, HCV pseudoparticles; IgG, immunoglobulin G; SCID, severe combined immune deficient; uPA, urokinase-type plasminogen activator.
Materials and Methods
Purification of Polyclonal Immunoglobulins from Human Plasma Samples.
Plasma samples were obtained from patient H, 26 years after his acute HCV infection,16 and from three healthy HCV-seronegative subjects. Subjects gave informed consent, and the protocol was approved by the institutional review boards of the National Institutes of Health and the Ghent University Hospital. After heat inactivation (56°C for 30 minutes), IgG was purified with a HiTrap Protein G column (GE Healthcare), as described previously.17
Generation and Viral Challenge of Hu-liv-uPA-SCID Mice.
Breeding and genotyping of alb/uPA-CBySmn.CB-17 Prkdcscid (uPA-SCID) mice have been described previously.14, 18 SCID mice homozygous for the uPA-transgene were transplanted with cryopreserved human hepatocytes from a single, HCV-uninfected female donor, as described previously.14 Only animals with a high repopulation grade, defined by a plasma human albumin content of 1 mg/mL or greater, were used. Passive immunization studies were performed by administering human polyclonal IgG to human liver–uPA+/+SCID mice (0.2 mg/g or 1 mg/g mouse body weight), 3 days before viral challenge with a 100% infectious mH77C dose (104 IU/mouse). In vitro neutralization of the viral inoculum was done by mixing the same challenge dose with human polyclonal IgG at a final concentration of 10 mg/mL IgG. This virus/IgG mixture was incubated at 37°C for 1 hour and subsequently injected into HCV-naïve chimeric animals harboring hepatocytes from the same human donor. All injections were intraperitoneal. The study protocol was approved by the animal ethics committee of the Ghent University Faculty of Medicine and Health Sciences.
Mouse plasma samples were stored at −80°C until analysis. Human polyclonal IgG were measured by an in-house enzyme-linked immunosorbent assay as described previously.19 Human albumin levels were quantified with the Human Albumin ELISA Quantitation kit (Bethyl Laboratories Inc., Montgomery, TX). HCV RNA levels were quantified using the COBAS Ampliprep TaqMan HCV Assay (Roche Diagnostics, Mannheim, Germany) at a 1/50 dilution, resulting in a lower limit of detection of 750 IU/mL. Sequence analysis of HCV recovered from infected animals was performed by polymerase chain reaction amplification of overlapping regions covering either the entire open reading frame or the envelope regions only and sequencing directly to determine the consensus sequence.20, 21
HCVpp Neutralization Studies.
The production and infectivity assays of HCV pseudoparticles were previously described.12, 22 Briefly, 293T cells were transfected with expression vectors encoding the E1E2 glycoproteins from the H77C strain, retroviral core proteins, and packaging-competent green fluorescent protein–containing retroviral transfer vectors. Serial twofold dilutions of samples to be tested for neutralization were mixed with H77C-HCVpp, pre-incubated at room temperature for 1 hour, and added to Huh-7 cells at 37°C. After 6 hours, the supernatants were removed and the Huh-7 cells were incubated in Dulbecco's modified Eagle's medium/10% fetal bovine serum for 72 hours. Green fluorescent protein–positive cells were quantified by fluorescence-activated cell sorting analysis. The neutralization titer was defined as the last dilution of the sample that conveyed a 50% or more reduction in the number of green fluorescing protein–positive cells compared with a blank (for mouse plasma samples) or an equivalent dilution of purified irrelevant IgG (for IgG samples).7
Statistical analysis was performed with GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA). IgG levels are expressed as mean ± standard deviation. The Mann-Whitney U test assessed statistical significance of differences, with two-sided probability values (P < 0.05). HCV RNA levels are expressed on a log10 scale, using the geometric mean viremia.
Defining the Optimal Experimental Conditions for In Vivo Intraperitoneal Homologous Viral Protection Studies.
To examine the optimal timeframe for intraperitoneal injection of IgG and viral challenge, the IgG pharmacokinetics was examined in uPA+/−SCID mice (n = 8) after intraperitoneal administration. As described in the online supplementary material on the HEPATOLOGY website (available at http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html), 3 days after intraperitoneal injection only a small fraction of the IgG load remained in the peritoneum, whereas plasma IgG levels were still high. These observations guided us to perform the passive immunoprophylaxis in human liver–uPA+/+SCID mice, as follows: (1) on day −3, peritoneal injection of polyclonal IgG at a dose of 0.2 mg/g mouse body weight; and (2) on day 0, intraperitoneal viral challenge with a 100% infectious dose in a large (300 μL) volume, to achieve maximal dilution of the residual antibodies in the peritoneum.
For homologous HCV viral protection studies, a high amount of both challenge virus and IgG from a single chronically infected patient are needed. Sera from patient H represent the ideal source, because a monoclonal H77C viral challenge pool has been generated in a chimpanzee after its inoculation with an infectious HCV clone that was derived from the acute phase plasma of patient H's posttransfusion hepatitis C virus infection.16, 20 A mouse H77C pool (referred to as mH77C) was generated by expansion of the monoclonal H77C pool in naïve chimeric mice. We found that all four human liver–uPA+/+SCID mice were infected after intraperitoneal inoculation with 104 IU HCV RNA of mH77C (100% infectious dose), as described in the online supplementary material on the HEPATOLOGY website.
Because increasing H77-HCVpp neutralizing IgG titers have been measured in patient H's plasma during the course of his chronic HCV infection,6, 7, 23 IgG was purified from a patient H plasma sample obtained in 2003, 26 years after his acute infection. The purified IgG retained its H77Cpp neutralizing capacities up to a dilution of 1/1600. Although HCV RNA was detected in one of two concentrated IgG preparations derived from the 2003 plasma sample (414 IU/mL and <150 IU/mL, respectively), the heat-inactivation step would have eliminated any residual infectivity.8 Similarly, IgG was isolated and pooled from three healthy HCV-seronegative subjects and further used as irrelevant, control IgG.
Low-Dose Passive Immunization Studies with Polyclonal Human Antibodies.
In a first study, 0.2 mg/g mouse body weight of patient H IgG and control IgG was injected in four and three chimeric animals, respectively. Animals were challenged 3 days later with 104 IU mH77C and monitored until 4 weeks after inoculation (Fig. 1). The human albumin levels remained well above the arbitrary limit of 1 mg/mL in each animal, indicating that the human hepatocytes graft was intact and functional. At challenge, adequate IgG concentrations were obtained in all animals (755.9 ± 123.3 μg/mL; n = 7). All mice immunized with control antibodies became HCV RNA positive within 2 weeks and had geometric mean viremia levels of 7.25 log10 IU/mL (n = 3) (Fig. 1A). Two of four animals receiving patient H IgG were completely protected (HCV RNA < 750 IU/mL; Fig. 1B), and two showed evidence of HCV infection (Fig. 1C). An early and high-titered viral breakthrough with HCV RNA levels of 5.64 log10 IU/mL at week 2 was observed in mouse K249-L. A late and low-titered viral breakthrough with HCV RNA levels of 3.76 log10 IU/mL at week 4 was demonstrated in mouse K313-. In these infected animals, IgG concentrations at challenge were 858.8 μg/mL and 885.3 μg/mL, respectively, indicating that the failure to prevent infection was not attributable to an insufficient IgG exposure. The course of these HCV infections was monitored by weekly plasma samples for 10 weeks after inoculation. The infection in K249-L reached a maximal HCV RNA titer of 6.77 log10 IU/mL, which slowly decreased to 5.64 log10 IU/mL at week 10. In animal K313- HCV RNA levels became undetectable (<750 IU/mL) from week 7 on. This could not be attributed to hepatocyte graft failure or persistent neutralizing activity of patient H IgG, because human albumin levels remained stable throughout the observation period, and IgG concentration became undetectable (<2.5 μg/mL) from week 6 on (Fig. 1C).
High-Dose Passive Immunization Studies.
To examine whether the in vivo neutralization was dose-dependent and could be improved by increasing the IgG loading dose, we injected another four and two chimeric mice with fivefold higher doses (1 mg/g body weight, referred to as “high dose IgG”) of patient H or control IgG, respectively. The same viral challenge was injected 3 days later. Because we identified the importance of the early and late viral kinetics during the “low dose IgG” studies, we sampled challenged animals weekly and followed all infected mice until 10 weeks after infection (Fig. 2). Human albumin levels indicated stable and functional hepatocyte grafts in all animals throughout the observation period. In comparison with the “low dose IgG” study, the plasma IgG levels at viral challenge were on average 2.8-fold higher (P = 0.0012). A wide range of IgG levels (1418.8- 2821.1 μg/mL) was observed in patient H IgG-treated animals. After challenge, both control IgG-treated animals displayed fast viral kinetics, with HCV RNA levels peaking at 7.69 log10 IU/mL within 3 weeks, followed by a steady decline to a geometric mean 5.76 log10 IU/mL at week 10 (Fig. 2A). Two animals treated with patient H IgG were completely protected (HCV RNA < 750 IU/mL; Fig. 2B) until week 4. In a third animal, the HCV RNA levels remained below the detection limit, until its spontaneous death on day 17 (K350-; Fig. 2B). In one (K358-R; Fig. 2C) of the four animals treated with patient H IgG, an HCV infection appeared at week 3 with an HCV viremia of 5.00 log10 IU/mL. Viremia increased until a peak value of 7.60 log10 IU/mL was reached at week 8. Infection occurred even though the plasma IgG level at viral challenge appeared to be adequate (1915.3 μg/mL). This animal died spontaneously on day 66.
Viral Kinetics During Passive Immunization Studies and Viral Sequence Analysis.
The viral replication in the control animals was not influenced by the presence of irrelevant IgG, because similar HCV RNA levels were reached within 2 weeks in animals without pretreatment (Fig. 3A). Furthermore, the entire open reading frame sequence of virus genomes from a week 4 plasma sample from one control mouse (K302-R) was identical to the sequence previously reported for the chimpanzee H77C plasma pool (Fig. 3B).24 Conversely, passive immunization with IgG from patient H protected five of eight animals until the end of follow-up or spontaneous death. Slower and less vigorous viral kinetics in the three animals with viral breakthrough infection were indicative of a nonsterilizing protection as compared with animals receiving control IgG (Fig. 3A, B). Sequence analysis revealed that the viral breakthroughs in the low-dose IgG studies were failures of neutralization, because the recovered viruses had envelope sequences identical to those of the challenge virus. In the virus causing the infection that occurred after the high-dose IgG administration, two nucleotide changes in the envelope genes (G1033A and C1730T) were detected; one was a silent mutation and the other resulted in an amino acid change (R231H) in the E1 protein (Fig. 3B).
Correlation of Neutralization of mH77C In Vivo and H77C-HCVpp Ex Vivo.
To examine whether the outcome of the in vivo experiment could be predicted from the ex vivo HCVpp neutralization titer, we incubated HCV pseudovirions carrying the envelope glycoproteins of the H77C clone with dilutions of plasma samples obtained immediately before viral challenge. Plasma samples from all animals treated with irrelevant antibodies did not contain H77Cpp neutralizing activity (Table 1). Plasma samples from animals treated with patient H IgG displayed neutralizing antibody titers in the H77Cpp system that ranged between 1/100 and 1/400 and correlated with increasing IgG levels. However, no clear correlation existed between in vivo protection and HCVpp neutralization titers, because viral breakthroughs were observed in animals with reciprocal neutralization titers of 200 and 400, whereas one protected animal had a titer of 100 (Table 1). Because one viral breakthrough contained a mutation in the E1 envelope protein, we have examined whether this represented a neutralization escape by comparing the neutralization titers of purified IgG (patient H) and pre-infection plasma (day 0) from 5 mice toward the mutant H77Cpp containing the R231H mutation with that against the wild-type H77Cpp. Neutralization titers toward both mutant and wild-type pseudoparticles were identical (data not shown), which indicates that the R231H mutation does not affect neutralization in the HCVpp system. Therefore, we assume that all breakthrough infections observed were most likely attributable to neutralization failures.
Table 1. Correlation of In Vivo and In Vitro Neutralization Data
Like humans, mice have the characteristic of an increased IgG catabolism when plasma levels of IgG are elevated. This IgG concentration–catabolism phenomenon has also been demonstrated in SCID mice injected with high amounts of human IgGs.25 Because loading chimeric mice with a fivefold higher amount of human IgG resulted in only 2.8-fold higher concentrations at the time of viral challenge, we assume that it would be difficult to obtain higher IgG concentrations in subsequent passive immunization studies. To expose mH77C viruses to an IgG concentration that is normally found in human sera, we have adapted the experimental design of the chimpanzee study by Farci et al.8 Therefore, viruses were incubated with 10 mg/mL IgG at 37°C for 1 hour both for comparison with the HCVpp system and to mimic in vivo circumstances. Mixtures of virus and IgG of patient H or control subjects were injected into three naïve chimeric animals each and HCV RNA was measured in weekly plasma samples. Surprisingly, control animals showed divergent viral kinetics compared with previous experiments, even though the animals had intact hepatocyte grafts based on human albumin levels (Fig. 4A). The earliest time at which all control animals became HCV RNA positive was 3 weeks after viral challenge, with HCV RNA levels of 3.27, 4.59, and 6.49 log10 IU/mL. Because of these delayed viral kinetics, viral outcome was compared at week 3 (Fig. 5A) and week 5 after challenge (Fig. 5B). HCV RNA levels remained undetectable in all animals receiving the patient H mixture. Unfortunately one of these died on day 21 and therefore could not be regarded as definitively protected from mH77C infection (Fig. 4B). The control animal with the lowest viremia (K489) spontaneously died on day 23. Taken together, two animals were protected from infection for at least 5 weeks after challenge with a mixture of mH77C and 10 mg/mL patient H IgG (Fig. 5B).
In this study, we examined the ability of polyclonal IgG isolated from patient H to neutralize in vivo the parental infecting hepatitis C virus in the only small animal model currently susceptible to HCV infection. We demonstrate that (1) the human liver uPA+/+SCID mouse is useful for passive immunization studies against HCV; (2) polyclonal IgG from a patient chronically infected with HCV can convey in vivo sterilizing immunity against a homologous nonmutated ancestral hepatitis C virus; and (3) the chimeric mouse model validates the HCV pseudoparticle system as being capable of detecting antibodies with in vivo neutralizing potential.
All animals used herein received hepatocytes from the same human donor and were challenged with mH77C within 9 weeks after transplantation. Differences in viral outcome are therefore most likely attributable to the amount and specificity of the injected IgG. Although passive immunization with patient H IgG protected five of eight animals, pre-incubation of the same viral inoculum with 10 mg/mL of the identical IgG conveyed a complete viral protection in two chimeras. The divergent viral kinetics in the “pre-incubated” control group contrasted with the highly reproducible viral kinetics in all control animals receiving an identical “fresh” inoculum (week 2 geometric mean viremia titer of 7.14 ± 0.40 log10 IU/mL; n = 12; Vanwolleghem T and Meuleman P, unpublished data). This might suggest that the pretreatment interferes with the infectivity of the inoculum. Therefore, the higher viral protection rate after pre-incubation (100%) compared with passive in vivo immunization experiments (66% and 50%) cannot solely be ascribed to higher IgG concentrations.
Because the lower detection limit of the HCV RNA quantification method was 750 IU/mL, it cannot be excluded that low titered (<750 IU/mL HCV RNA) viral replication was present in “protected” animals. However, this seems highly unlikely. First, infection studies with similar genome titered inocula of all HCV genotypes tested until now (1a, 1b, 2a, and 4a) have revealed very rapid viral kinetics in chimeric mice and high-level viremia within 2 weeks. Furthermore, lower genome titered inocula have never resulted in positive HCV RNA titers later than 3 weeks after injection (Vanwolleghem T, Meuleman P, unpublished data). The high viral replication in animals receiving control antibodies supports this observation. Second, we have previously demonstrated that a high load of HCV virions in a nonsusceptible uPA+/−SCID mouse becomes undetectable within 3 days, confirming previous reports of the short in vivo half-life of HCV particles.15, 26 Finally, 4 weeks after challenge, human IgG levels have dropped more than 10-fold (4 half-lives, supplementary data) and were not able to neutralize H77Cpp at the lowest dilution in samples from two patient H IgG-treated animals (H77Cpp titer <25 in K311- and K249-). Therefore, in the absence of functional murine T-cells and B-cells and with gradually decreasing levels of patient H IgG, an HCV infection should have become productive and detectable within a 4-week follow-up period.
Indeed, HCV breakthrough infections were observed at week 2 (K249-L), week 3 (K358-R), and week 4 (K313-) after challenge. Of these, only the viral replication in K313- did not mimic the course observed in control animals (Fig. 1; Table 1). In animal K313-, no envelope mutations were identified in recovered viruses, and patient H IgG levels had already dropped substantially. We speculate that the infected hepatocyte cell mass in K313- may have been small enough to be controlled by the innate antiviral interferon response. In the other animals this control mechanism may have failed because a larger hepatocyte cell mass may have been infected from the very first moment.27–29 It is noteworthy that even animals with high level breakthrough infections showed a clear delay in viral replication compared with control IgG-treated animals, indicative of a nonsterilizing protection by patient H's IgG (Fig. 3).
The identical amino acid sequence of the envelope glycoproteins in both the H77Cpp and our challenge virus allowed a direct comparison of in vivo neutralizing capacities and the in vitro (HCVpp) neutralizing titer present in mouse plasma on viral challenge. The absence of in vitro neutralization at the lowest dilution correlated perfectly with the viral kinetics observed in control animals (Table 1). More importantly, all animals treated with patient H IgG had in vitro neutralizing titers ranging between 1/100 and 1/400. Because viral replication was consistently suppressed in the latter animals, the HCVpp system could specifically detect antibodies with in vivo neutralizing potential. However, even with an in vitro H77Cpp neutralization titer of 1/200 (K249-L) and 1/400 (K313- and K358-R) at viral challenge, in vivo viral breakthroughs occurred. In one of these animals (K358-R), rapidly replicating breakthrough viruses had an amino acid change in the E1 envelope glycoprotein (Fig. 3B). Because both mutant and wild-type H77Cpp were equally well neutralized by day 0 plasma samples and purified patient H IgG, there is no evidence that this represented an escape mutation. A correlation between in vitro neutralization and in vivo protection has been described for several viruses, with protection generally requiring in vitro titers in the order of magnitude as observed herein.30 However, because of the limited range of H77Cpp titers measured, a quantitative in vitro correlate of viral protection could not be defined in this study.
When comparing the in vivo and in vitro neutralization systems, important differences that relate to the infectious particles as well as to the target cells should be highlighted. The production of HCV particles is dependent on secretion and assembly of very-low-density lipoproteins.31 The HCV generated in chimeric mice is assembled in fully functional human hepatocytes that are able to produce several human apolipoproteins.14 The physiological buoyant density and infectivity for Huh 7.5 cells of these viral particles are indistinguishable from chimpanzee-derived virions.15 Furthermore, electrophoresis analysis of chimeric mouse plasma reproducibly showed a clear lipoprotein band comigrating with human very-low-density lipoproteins, which was absent in plasma of nontransplanted uPA+/+SCID mice (Vanwolleghem T, unpublished data). The 293T human embryonic kidney cells, used to produce HCVpp, are unable to secrete very-low-density lipoproteins.31 Likewise, it is not clear whether the heterogeneity and density of envelope glycoproteins in HCVpp resemble that of native HCV virions.11 Finally, in contrast to the chimeric mouse model, the HCVpp target hepatoma cell lines do not support the replication of native HCV particles (except for the JFH-1 strain).14 Therefore, challenge studies in the chimeric mouse model will more closely reflect the normal human clinical situation than the HCVpp system.
In contrast to our results, recent trials of passive immunizations with polyclonal and human anti-E2 monoclonal antibodies have not been able to prevent experimental HCV infections in chimpanzees or re-infections of the liver allograft after transplantation in humans.1–4 Although low antibody titers in some patients have been held responsible for the poor outcomes, the quality and quantity of the viral challenge should not be disregarded. Compared with the massive number of viral particles passing through the liver allograft on reperfusion, our chimeric mice received a relatively small inoculum (104 IU HCV RNA). Whereas in natural HCV infections the inoculum consists of a highly complex quasispecies, the challenge virus used in the current study was highly homogeneous. Finally, the diversity of HCV genotypes, subtypes, and strains in clinical trials require a successful broad cross-neutralizing activity of antibodies, whereas our results are limited to a homologous in vivo neutralization. Heterologous challenge studies are needed to estimate the HCV cross-neutralizing activities of polyclonal IgG. Still, even in the absence of cross-reactivity, our results open perspectives for passive immunizations with a cocktail of isolate-specific or strain-specific monoclonal antibodies.
In conclusion, polyclonal IgG from a patient with a long-standing HCV infection conveys sterilizing protection against a homologous viral challenge in human liver–uPA+/+SCID mice. The HCVpp system is able to specifically detect antibodies with in vivo neutralizing potential. Both experimental systems will be helpful in identifying antibodies for future clinical use.
The authors thank Kristina Faulk, Manuela Cousin, and Lieven Verhoye for excellent technical assistance.