In Vivo Determinants of Hepatitis C Virus Adaptation and Escape From Neutralizing Antibody AR5A

Abstract

Background & Aims

Mechanisms of hepatitis C virus (HCV) adaptation and escape from broadly neutralizing antibodies (bNAbs) have been primarily studied in vitro. Here, we used a previously developed in vivo adapted J6/JFH1_A876P virus and the highly bNAb sensitive hypervariable region 1 (HVR1) deleted variant, J6/JFH1_A876P,ΔHVR1, to study adaptation and bNAb AR5A escape in the HCV-permissive human-liver chimeric mouse model.

Methods

In vitro identified AR5A escape substitution, L665S, was introduced into J6/JFH1_A876P with or without HVR1. The infection of human liver chimeric mice with these recombinants revealed adaptive mutations, and the potential mechanism of adaptation was extensively characterized in vitro. Finally, we tested the barrier to resistance of AR5A in vivo by challenging passively immunized animals with HVR1-deleted viruses, either with or without the AR5A escape substitution, L665S.

Results

L665S was found to be an escape substitution in vivo. Furthermore, sequence analysis showed that the escape substitution L665S arose as early as 2 weeks post infection. At week 8, we also identified antibody escape substitutions as well as several potential in vivo adaptive substitutions in E2. For J6/JFH1_A876P, S449P and M702L increased cell-free particle infection and broadly affected antibody sensitivity for virus with HVR1. For J6/JFH1_A876P,ΔHVR1, N430D and M702L substitutions increased both cell-free particle mediated infection and cell-to-cell spread, whereas N430D also increased thermal stability at 37°C.

Conclusions

We show that L665S is an AR5A escape mutation in vivo, supporting the use of cost-effective vaccine escape studies in vitro. We also identify novel in vivo adaptive mutations and characterize their mechanism of action, thus facilitating interpretation of future HCV in vivo studies.

Graphical abstract

Summary.

By using the human-liver chimeric mouse model, we determined the barrier-to-resistance of the antibody AR5A in vivo, showing that this model can be used to test viral evolution in the presence of neutralizing antibodies.

What You Need to Know.

Background

The barrier-to-resistance of hepatitis C virus (HCV) against neutralizing antibodies (NAbs) has mostly been studied in vitro, and little is known about how this correlates to in vivo.

Impact

The results corroborate a strong correlation between in vitro and in vivo escape studies to establish the barrier-to-resistance of NAbs.

Future Direction

Our in-depth characterization of envelope protein evolution in humanized mice facilitates the use of this model to test viral evolution in the presence of vaccine-induced NAbs, thus helping predict the appearance of HCV resistance in the human population.

Hepatitis C virus (HCV) is one of the major causes of chronic liver diseases, including liver cirrhosis and hepatocellular carcinoma. The World Health Organization estimates that around 58 million people have chronic hepatitis C, including 3.8 million adolescents and children.¹ More than 1.5 million new infections occur every year, most of which will become chronic.¹ Decades of research have resulted in the development of HCV-specific direct-acting antivirals (DAAs) with cure rates of around 95%.^1,2 However, the inaccessibility of treatment in developing countries, the appearance of resistant variants, and the low rate of diagnosis severely limit DAA-mediated disease control.^1,3,4 Thus, the development of a protective vaccine is likely essential for global HCV elimination.

HCV is a positive-sense RNA virus, which is classified into 6 clinically relevant genotypes.⁵ The virus genome encodes a single polyprotein, which contains the structural proteins Core, HCV envelope glycoprotein E1 (E1), and HCV envelope glycoprotein E2 (E2) along with p7 and 6 nonstructural (NS) proteins. Early appearance and rapid induction of neutralizing antibodies (NAbs) during acute HCV infection is correlated with virus clearance.6, 7, 8, 9 Further, passive protection against HCV infection can be achieved in chimpanzees and human-liver chimeric mice pre-infused with anti-HCV antibodies,10, 11, 12, 13 showing that NAbs can play a key role in controlling HCV infection.

The E1/E2 complex forms an anti-parallel homodimer¹⁴ and is the principal target of NAbs and a preferable candidate for a B-cell based HCV vaccine.¹⁵ Vaccine studies have shown that the induction of antibodies against E1/E2 can prevent homologous infection in chimpanzees¹⁶ and protect human-liver chimeric mice against homologous and heterologous viral infection17, 18, 19 Despite these positive results, the high genetic heterogenicity of HCV and the incomplete understanding of the protective immune response are still principal challenges to overcome to design an effective B-cell HCV vaccine.^15,20

Immunogenic clusters on E1/E2 have been classified using several overlapping nomenclatures. Thus, antigenic sites (ASs) include the linear epitopes AS412 and AS434,²¹ antigenic regions (AR) include the conformational epitopes AR3, AR4, and AR5,^22,23 and 5 antigenic domains including the conformational epitopes A to D and the linear epitope E (same as AS412).24, 25, 26 Although several broad neutralizing antibodies (bNAbs) have been described targeting conserved epitopes, the high mutation rate of the virus has been shown to result in antibody escape in vitro and in vivo, such as observed when culturing naturally sensitive HCV isolates in the presence of NAbs targeting AS412 and domain B.^25,27, 28, 29, 30 Similarly, resistance substitutions localized in AS412 emerged after antibody treatment of HCV-infected chimpanzees and human patients,^10,31, 32, 33 showing the inherent capacity of HCV to develop resistance against NAbs.

The development of infectious HCV culture systems has facilitated escape studies to investigate the barrier to resistance of promising bNAbs.^34,35 By using highly antibody-sensitive viruses missing hypervariable region 1 (HVR1) in E2 (HVR1-deleted), we previously determined the barrier to resistance of the vaccine-relevant NAbs AR3A, AR4A, and AR5A.36, 37, 38 Although escape substitutions against AR3A and AR4A have a minor effect on virus resistance and/or greatly reduce infectivity, substitutions L665W and L665S/S680T greatly increased AR5A resistance without affecting infectivity.³⁶ Although our results led us to identify potential HCV vaccine-resistant variants, it remains unclear if escape substitutions identified in vitro will be similar in vivo, particularly for conformation-dependent bNAbs, like AR5A.

The human liver xenograft urokinase-type plasminogen activator/severe combined immunodeficiency (uPA-SCID) chimeric mouse model (human-liver chimeric mice) has been used to study HCV replication, antiviral therapies, and NAb protection in vivo.^39,40 These immunocompromised animals are humanized via transplantation of primary human hepatocytes to make them susceptible to HCV infection. This model has been used to test the prophylactic effect of the monoclonal NAbs, AR3A and AR4A, as well as other polyclonal and monoclonal antibodies isolated from chronically infected patients.^12,13,22,23,41, 42, 43, 44 Further, we have observed the appearance of resistance variants during prophylactic treatment of HCV-infected mice with polyclonal serum-derived Nabs,^11,41 indicating that this model can be used to study how HCV escapes immune surveillance in vivo. However, information on viral adaptation in vivo, particularly in response to the presence of NAbs, remains scarce. This knowledge is particularly important for unraveling dynamic changes in envelope proteins, as these are the key targets of NAbs and consequently important in B-cell vaccine designs to elicit NAbs.

In this study, we infected human liver chimeric mice with HCV J6/JFH1 variants in the absence or presence of the NAb, AR5A, and performed detailed molecular characterization of emergent envelope protein substitutions, which will be a useful resource in future studies using this model. Moreover, we show that the human-liver chimeric mouse model can be used to track the appearance of viral variants that emerge following NAb exposure and support the use of cost-effective in vitro escape studies. Our results are highly useful in the evaluation of the barrier to resistance of different neutralizing epitopes to inform HCV vaccine design.

Results

Inoculation of Human-liver Chimeric Mice With In Vivo Adapted AR5A-resistant HCV Recombinants

Previously, we showed that the substitution A876P increased infectivity and in vivo genetic stability of the HCV genotype 2a recombinant virus, J6/JFH1, with or without HVR1.⁴¹ To test the in vivo viability of J6/JFH1 with the AR5A-induced escape substitutions S680T and L665S,³⁶ we introduced A876P into the HCV recombinants harboring S680T or the combination L665S/S680T, in genomes either with or without HVR1. Viral genomic RNA was subsequently generated from the respective recombinants and used to transfect Huh7.5 cells. Supernatants were collected every 24 hours, and virus infectivity was measured in focus forming units (FFUs) (Figure 1A). Next, a first passage virus stock was generated for each J6/JFH1 recombinant, and their sensitivity to the NAb AR5A was determined (Figure 1B), as described.³⁶ As previously observed, the substitution A876P increased virus infectivity, but AR5A susceptibility of viruses harboring the escape substitutions (J6/JFH1_{A876P/L665S/S680T}, IC₅₀ = 240 μg/mL; J6/JFH1_{ΔHVR1/A876P/L665S/S680T} IC₅₀ = 14 μg/mL) was similar to findings in our AR5A in vitro escape study using the original J6/JFH1.³⁶

Figure 1.

Infection of human-liver chimeric mice with HCV recombinants harboring the AR5A in vitro escape substitutions S680T/L665S. (A) Huh 7.5 cells were transfected with the in vitro transcribed RNA of indicated recombinant viruses. The values in the graph represent the HCV titers collected after 72 hours post-transfection as the mean of 4 replicates ± SEM. (B) First passage stocks of the respective recombinant viruses were subjected to 4 replicates of 5-fold dilution from 250 to 0.038 μg/mL of AR5A for virus harboring HVR1 or 10-fold dilution from 50 to 5 × 10^-5 μg/mL of AR5A for HVR1-deleted viruses. The values in the graphs represent the IC₅₀ values, and the error bars represent SD. (C) Human-liver chimeric mice were infected with the culture-derived HCV recombinants. (D) J6/JFH1_A876P/S680T and (E) J6/JFH1_{ΔHVR1/A876P/S680T} with or without L665S (A876P is a previously described in vivo adaptive substitution in NS2). The graphs in (DandE) show the HCV RNA titers from individual animals of a total of 3 mice per recombinant virus monitored in plasma until 16 weeks post-infection. HCV RNA was further extracted from plasma collected during weeks 2 and 8, and the sequences of the amplified HCV Core-NS2 region was determined (Table 1 only for week 8). ^†Mouse was sacrificed because humane endpoints were met.

To test infectivity and genetic stability in vivo, we used cell-culture derived stocks of the viruses J6/JFH1_A876P/S680T, J6/JFH1_{A876P/L665S/S680T}, J6/JFH1_{ΔHVR1/A876P/S680T}, and J6/JFH1_{ΔHVR1/A876P/L665S/S680T} to infect human-liver chimeric mice (Figures 1C–E). Three mice were injected intrasplenically with 2000 FFUs per animal of each virus stock. HCV RNA titers in plasma were measured at week 1 and 2, and then every 2 weeks up to 16 weeks postinfection. Further, virus RNA at week 2 and 8 postinfection was recovered and amplified to identify potential changes in the E1/E2 amino acid sequence. Week 2 virus did not contain additional mutations, highlighting the genetic stability of J6/JFH1_A876P. However, we observed several coding mutations at week 8 (Table 1), of which many were specific to the corresponding viruses. M405V, S449P, and M702L (E2) were identified for J6/JFH1_{A876P/L665S/S680T}, whereas T542A (E2) was identified only in mice infected with J6/JFH1_A876P/S680T. Meanwhile, N430D (E2) was observed for in vivo adapted HVR1-deleted viruses harboring S680T or L665S/S680T and M702L (E2) was observed for J6/JFH1_{ΔHVR1/A876P/L665S/S680T}. Although additional substitutions were observed in E1/E2, they either were only present in 1 animal or did not represent dominant substitutions (Table 1).

Table 1.

HCV In Vivo Adaptive Substitutions Identified at Week 8 Postinfection

HCV gene	E1	E1	E2	E2	E2	E2	E2	E2	E2	E2	E2	E2	E2	E2	E2	NS2
Nucleotide number
J6 position	1218	1274	1553	1568	1628	1674	1685	1965	1970	2232	2346	2367	2390	2439	2456	2978
H77 abs. ref. (AF009606) position	1219	1275	1554	1569	1629	1675	1689	1960	1965	2221	2335	2356	2379	2428	2445	2967
Nucleotide change	T	A	A	A	A	A	T	A	A	T	T	G	T	T	A	G
Mutant
J6/JFH1A876P/S680T		A/G					T/c		A/G				A			C
J6/JFH1A876P/S680T				A/G				A/C/g	A/g				A			C
J6/JFH1A876P/L665S/S680T							C		G		a		A			C
J6/JFH1A876P/L665S/S680T			G				C				C		A		T	C
J6/JFH1A876P/L665S/S680T	C		G				C				C		A		T	C
J6/JFH1ΔHVR1/A876P/S680T					a/G							G/T	A			C
J6/JFH1ΔHVR1/A876P/S680T					G					C			A	t/C		C
J6/JFH1ΔHVR1/A876P/L665S/S680T					G						C		A		T	C
J6/JFH1ΔHVR1/A876P/L665S/S680T					G	G					C		A		T	C
Amino acid
Polyprotein position in J6	293	312	405	409	430	445	449	542	544	631	669	676	684	700	706	880
H77 abs. ref. (AF009606) position	293	312	405	409	430	445	449	540	542	627	665	672	680	696	702	876
Amino acid change	V>A	T>A	M>V	K>E	N>D	H>R	S>P	N>T>s	T>A	F>S	L>S	L>W	S>T	I>T	M>L	A>P

NOTE. Substitutions were identified by direct sequence analysis of RT-PCR from recovered viruses present in mice plasma at 8 week post-infection. Italicized columns denote substitutions already present in the virus stocks before in vivo adaptation.

E1, HCV envelope glycoprotein E1; E2, HCV envelope glycoprotein E2; HCV, hepatitis C virus; NS, nonstructural.

^aSubstitution reverted to the parental amino acid.

E2 Substitutions N430D and M702L Influence Cell-to-cell and Cell-free Particle Infectivity, Whereas N430D Also Improves Virus Thermal Stability

To study the mechanisms of observed in vivo adaptation of the HVR1-deleted viruses, we introduced substitutions N430D and M702L alone or in combination with S680T or L665S/S680T, which we used to generate viral genomic RNA and transfect Huh7.5 cells (Figure 2A). Release of viral particles (cell-free particles) and their infectivity was monitored every 24 hours post-transfection. Although N430D alone reduced the infectivity of J6/JFH1_ΔHVR1/A876P, we did not observe any measurable effect when tested in combination with S680T or L665S/S680T (Figure 2A). Similarly, M702L substantially reduced the infectivity of J6/JFH1_ΔHVR1/A876P and slightly reduced J6/JFH1_{ΔHVR1/A876P/S680T} infectivity; however, infectivity increased when we tested M702L in combination with L665S/S680T, suggesting a compensatory effect in this background (Figure 2A). Next, we measured cell-to-cell infectivity by mixing naive cells with cells transfected with genomic viral RNA of the respective recombinant viruses and cultured them in the presence of high levels of polyclonal NAb, C211⁴⁵ (Figure 2B) or monoclonal NAb AR3A (Figure 2C), to abrogate cell-free particle infection. By counting the number of infected cells per FFU, we found that viruses harboring N430D and M702L had increased cell-to-cell spread by ∼2 times compared with J6/JFH1_{ΔHVR1/A876P/L665S/S680T} (Figure 2B and C).

Figure 2.

Effect of in vivo derived substitutions N430D or M702L with or without known AR5A escape substitutions in virus infectivity, antibody susceptibility, and particle stability in 37°C in the HCV recombinant J6/JFH1_ΔHVR1/A876P. (A) Huh7.5 cells were transfected with in vitro-transcribed RNA of the respective HCV recombinants. The values in the graph represents the HCV titers in collected supernatant after 72 hours post-transfection as the mean of 4 replicates ± SEM. (B–C) Huh7.5 cells were transfected with RNA, and transfected cells were mixed with naive cells and plated in 96-well plates and treated with 50 μg/mL C211 polyclonal antibody (B) or 10 μg/mL of AR3A (C) to avoid cell-free particle infection. The values represent the fold increase of single spots per FFU in 24 hours calculated in 6 replicates per each condition using the BioSpot software (Cellular technology Lmtd) following HCV-specific staining at 24- and 48-hours post-transfection. Error bar represents ± SEM. (D–E) First passage stocks were generated, and the indicated HCV recombinants were subjected to 4 replicates of 10-fold dilution series from 50 to 5 × 10^-5 μg/mL of AR5A (D) or AR3A (E). The values in the graphs represent the IC₅₀ values and the error bars represent SD. (F) The temperature stability was tested by incubating the viruses at 37°C for 0.5, 1, 2, 3, 4, 6, and 8 hours prior to infection in Huh7.5 cells, and the half-life was calculated for each specific recombinant. (G–I) Huh7.5 cells were incubated with a dilution series of an antibody that block the interaction of HCV with cellular receptor CD81 (G), SR-BI (H), and LDLr (I) in quadruplicates with 8 replicates of only medium prior to infection with the indicated recombinants. Values in the graph represent the percentage of infectivity as the mean of 4 replicates ± SEM. The data were analyzed by using 1-phase exponential decay curve fitting. Statistical analysis was carried out by 1-way ANOVA or 2-way ANOVA, significance is indicated as: ∗P < .05; ∗∗P < .005; ∗∗∗P < .0005; and ∗∗∗∗P < .0001.

Next, we generated first passage virus stocks of the various recombinants and tested whether N430D and M702L affected sensitivity to AR5A in dose-response FFU reduction assays, which was not observed for either substitution (Figure 2D). In addition, we probed if these substitutions affected sensitivity to neutralization by the unrelated NAb, AR3A, in similar dose-response assays. Although we observed a slight increase in AR3A susceptibility for virus harboring N430D, we did not observe any effect for M702L (Figure 2E).

To further characterize the effects of N430D and M702L on J6/JFH1_{ΔHVR1/A876P/L665S/S680T}, we tested receptor dependency and particle stability at 37°C. To test receptor dependency, we performed dose-response entry blocking assays using specific antibodies against receptors cluster of differentiation 81(CD81), scavenger receptor class B type I (SR-BI), and low-density lipoprotein receptor (LDLr). However, the mutations did not alter receptor dependency (Figure 2G–I). Next, we tested virus stability by incubating the respective virus stocks for 0.5, 1, 2, 3, 4, 6, or 8 hours at 37°C prior to infecting Huh7.5 cells (Figure 2F). We observed that the half-life of virus harboring M702L was similar to that of J6/JFH1_ΔHVR1/A876P and J6/JFH1_{ΔHVR1/A876P/L665S/S680T}. Conversely, N430D increased virus half-life from 1.7h in J6/JFH1_{ΔHVR1/A876P/L665S/S680T} to 4.0h in J6/JFH1_{ΔHVR1/A876P/L665S/S680T/N430D}. Thus, our results suggested that N430D and M702L were fixed during the in vivo infection due to their capacity to increase virus stability and/or cell-to-cell infectivity.

Substitutions S449P and M702L Increase HCV Cell-free Particle Infectivity and Affect Broad Sensitivity to Antibodies in Viruses Harboring HVR1

Similarly to the HVR1-deleted viruses, viruses harboring HVR1 were genetically stable at 2 weeks post infection in mice. However, we observed several amino acid changes in virus isolated at week 8 postinfection, including the previously observed, M702L (Table 1). To study the effect of these substitutions on virus infectivity, we selected dominant substitutions that appeared in 2 or more animals. We introduced the substitutions M405V, S449P, and M702L (E2) alone or in combination with S680T or L665S/S680T in the parental virus J6/JFH1_A876P (Figure 3A). The viral titers obtained after transfection of Huh7.5 cells with recombinant viral RNA showed that, although all the substitutions reduced the titers of the parental virus J6/JFH1_A876P, M405V increased titers in combination with S680T (Figure 3A) and S449P did not affect cell-free particle infectivity in combination with S680T or L665S/S680T (Figure 3A). Remarkably, virus harboring M702L in combination with L665S/S680T showed higher titers than the parental J6/JFH1_A876P (Figure 3A), suggesting that M702L had a compensatory effect specific to virus harboring L665S/S680T, as also observed for HVR1-deleted viruses (Figure 2A). Although we also tested the effect of T542A in cell-free particle infectivity, this substitution was only observed in J6/JFH1_A876P/S680T and greatly reduced viability of the recombinant J6/JFH1_{A876P/L665S/S680T} (Table 1 and data not shown); thus, it was not included in further experiments. In addition, by testing cell-to-cell spread, we did not observe any effect by M405V and M702L, whereas cell-to-cell spread was reduced by substitution S449P (Figure 3B and C).

Figure 3.

Effect of in vivo derived substitutions or M405V, S449P, and M702L with or without known AR5A escape substitutions in virus infectivity, antibody susceptibility, particle stability in 37°C, and SR-BI dependency in the HCV recombinant J6/JFH1_A876P. (A) Huh7.5 cells were transfected with in vitro transcribed RNA of the respective HCV recombinants. The values in the graph represents the HCV titers in collected supernatant after 72 hours post-transfection as the mean of 4 replicates ± SEM. (B–C) Huh7.5 cells were transfected with RNA and transfected cells were mixed with naive cells and plated in 96-well plates and treated with 1000 μg/mL C211 polyclonal antibody (B) or 250 μg/mL of AR3A (C) to avoid cell-free particle infection. The values represent the fold increase of single spots per FFU in 24 hours calculated in 6 replicates per each condition using the BioSpot software (Cellular Technology Lmtd) following HCV-specific staining at 24- and 48-hours post-transfection. Error bar represents ± SEM. (D–E) First passage stocks were generated, and the indicated HCV recombinants were subjected to 4 replicates of 10-fold dilution series from 100 to 2.5 × 10^-4 μg/mL of AR5A (D) or AR3A (E). The values in the graphs represent the IC₅₀ values, and the error bars represent SD. (F) The temperature stability was tested by incubating the viruses at 37°C for 0.5, 1, 2, 3, 4, 6, and 8 hours prior to infection in Huh7.5 cells, and the half-life was calculated for each recombinant. Values in the graph represent the percentage of infectivity as the mean of 4 replicates ± SEM. The data were analyzed by using 1-phase exponential decay curve fitting. (G–I) Huh7.5 cells were incubated with a dilution series of an antibody that block the interaction of HCV with cellular receptors CD81 (G), SR-BI (H), and LDLr (I) in quadruplicates with 8 replicates of only medium prior to infection with the indicated recombinants. The values in the graph represent the percentage of HCV infection inhibition reached at 12.5 μg/mL of antibody ± SEM. Statistical analysis was carried out by 1-way ANOVA or 2-way ANOVA; significance is indicated as: ∗P < .05; ∗∗P < .005; ∗∗∗P < .0005; and ∗∗∗∗P < .0001.

We generated first passage stocks of J6/JFH1_{A876P/L665S/S680T} harboring the substitutions M405V, S449P, or M702L. We next tested AR5A (Figure 3D) and AR3A (Figure 3E) susceptibility, stability at 37°C (Figure 3F), and receptor dependency (Figure 3G–I). M405V did not affect virus sensitivity to NAbs nor did it affect receptor dependency, whereas S449P increased viral sensitivity to AR5A (Figure 3D) and AR3A (Figure 3E) neutralization and reduced SR-BI dependency (Figure 3H). In contrast, M702L broadly reduced NAb neutralization (Figure 3D and E) and increased SR-BI dependency when in combination with L665S/S680T (Figure 3H). In addition, M405V, S449P, and M702L did not affect CD81 and LDLr dependency (Figure 3G and I) or particle stability at 37°C (Figure 3F). This suggests that S449P and M702L were selected in vivo due to their capacity to increase cell-free particle mediated infectivity, whereas the effect of M405V remains obscure.

AR5A Escape of HVR1-deleted Viruses In Vivo

Due to the inherent resistance of J6/JFH1 viruses to NAb neutralization, we tested if L665S provides reduction of sensitivity to AR5A neutralization in vivo by using the antibody-sensitive HVR1-deleted variant to infect human-liver chimeric mice. We inoculated mice intraperitoneally with phosphate buffered saline (PBS) or 0.2 mg/mice of the AR5A antibody. After 24 hours, mice were challenged by intrasplenic injection with 500 FFUs per animal of the recombinant viruses J6/JFH1_{ΔHVR1/A876P/S680T} or J6/JFH1_{ΔHVR1/A876P/L665S/S680T} (Figure 4). Viral RNA in plasma was measured at weeks 1, 2 ,4, 6, 8, and 17 post-challenge. HCV RNA was extracted from plasma, and the E1/E2 HCV sequence was determined for weeks 2 and 8 postinfection. For J6/JFH1_{ΔHVR1/A876P/L665S/S680T}, it readily induced high-titer viremia in the presence of AR5A (Figure 4C), indicating that L665S conferred AR5A resistance in vivo as also observed in vitro.³⁶ For J6/JFH1_{ΔHVR1/A876P/S680T}, we observed protection by AR5A in 2 of 4 animals. However, low HCV RNA titers were detected in 2 animals week 1 post-challenge, and the titers increased to the level of unprotected animals at week 2 (Figure 4B). Sequence analysis of virus from these 2 animals revealed that, for J6/JFH1_{ΔHVR1/A876P/S680T}, they had both acquired the substitution, L665S, as early as 2 weeks postinfection, thus demonstrating in vivo escape caused by the same substitution as previously observed during in vitro escape studies³⁶ (Table 2). Additionally, virus sequences obtained at week 8 postinfection showed that most of the viruses developed the previously observed substitution N430D, whereas M702L was only present in combination with L665S/S680T substitutions (Table 3). However, our in-depth in vitro studies of these substitutions permitted us to rule out their role in AR5A escape, instead further implicating them in in vivo adaptation.

Figure 4.

AR5A protection studies of human-liver chimeric mice against HCV. (A) Human-liver chimeric mice were inoculated intraperitoneally with 0.2 mg of AR5A before being challenged intrasplenically with the indicated HVR1-deleted recombinant, and HCV RNA titers were monitored in plasma until 17 weeks post-challenge. HCV RNA was further extracted from plasma during weeks 2 and 8, and the sequence of the HCV core-NS2 sequence was determined (Table 2 and 3). (B) AR5A inoculation protected animals against infection with virus without L665S; however, virus escape was observed in 2 animals. Sequence analysis of recovered viruses showed that they acquired L665S as early as week 2 postinfection. (C) Virus harboring the escape substitution, L665S, were able to replicate in the presence of AR5A. Dashed line indicates the assay detection limit. ^†Mouse was sacrificed because humane endpoints were met.

Table 2.

AR5A Treatment of Humanized Mice Infected With HVR1-deleted HCV-resistant Variants Sequenced at Week 2 Postinfection

HCV gene		E1	E1	E1	E2	E2	E2	E2	E2	E2	NS2
Nucleotide number
J6 position		1274	1370	1460	1628	1650	2235	2258	2346	2390	2978
H77 abs. ref. (AF009606) position		1275	1371	1461	1629	1651	2224	2247	2335	2379	2967
Nucleotide change		A	A	A	A	T	A	G	T	T	G
Mutant	Treatment
J6/JFH1ΔHVR1/A876P/S680T	NT									A	C
J6/JFH1ΔHVR1/A876P/S680T	NT									A	C
J6/JFH1ΔHVR1/A876P/S680T	NT									A	C
J6/JFH1ΔHVR1/A876P/S680Ta	AR5A 0,2mg	G							Ca	A	C
J6/JFH1ΔHVR1/A876P/S680Ta	AR5A 0,2mg								Ca	A	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	NT								C	A	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	NT				A/G				C	A	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	NT								C	A	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	NT				A/G				C	A	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	AR5A 0.2 mg			A/G	A/G			G/c	C	A	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	AR5A 0.2 mg					T/c	C		C	A	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	AR5A 0.2 mg		A/g						C	A	C
Amino acid
Polyprotein position in J6		312	344	374	430	437	632	640	669	684	880
H77 abs. ref. (AF009606) position		312	344	374	430	437	628	636	665	680	876
Amino acid change		T>A	I>V	I>V	N>D	F>S	K>T	V>L	L>S	S>T	A>P

NOTE. Substitutions were identified by direct sequence analysis of RT-PCR from recovered viruses present in mouse plasma after 2 week postinfection. Italicized columns denote substitutions already present in the virus stocks before in vivo adaptation.

AR5, antigenic region 5; E1, HCV envelope glycoprotein E1; E2, HCV envelope glycoprotein E2; HCV, hepatitis C virus; NS, nonstructural.

^aVirus escaped AR5A treatment by acquiring the substitution L665S.

Table 3.

AR5A Treatment of Humanized Mice Infected With HVR1-deleted HCV-resistant Variants Sequenced at Week 8 Postinfection

HCV gene		E1	E2	E2	E2	E2	E2	E2	E2	E2	E2	E2	E2	NS2
Nucleotide number
J6 position		1274	1628	1940	1941	2235	2258	2337	2346	2367	2390	2456	2580	2978
H77 abs. ref. (AF009606) position		1275	1629	1935	1936	2224	2247	2326	2335	2356	2379	2445	2569	2967
Nucleotide change		A	A	A	A	A	G	T	T	G	T	A	A	G
Mutant	Treatment
J6/JFH1ΔHVR1/A876P/S680T	NT		G								a			C
J6/JFH1ΔHVR1/A876P/S680T	NT	A/G	A/G								a			C
J6/JFH1ΔHVR1/A876P/S680T	NT	a/G	G								A			C
J6/JFH1ΔHVR1/A876P/S680Tb	AR5A 0,2mg	a/G	G		A/C				Cb		A			C
J6/JFH1ΔHVR1/A876P/S680Tb	AR5A 0,2mg		G						Cb	G	A	T		C
J6/JFH1ΔHVR1/A876P/L665S/S680T	NT		G						C		A	T		C
J6/JFH1ΔHVR1/A876P/L665S/S680T	NT	A/G				A/C		T/A	C		A		A/G	C
J6/JFH1ΔHVR1/A876P/L665S/S680T	NT		G	G					C		A	T		C
J6/JFH1ΔHVR1/A876P/L665S/S680T	AR5A 0,2mg		G						C		A	T		C
J6/JFH1ΔHVR1/A876P/L665S/S680T	AR5A 0,2mg		G						C		A	C		C
J6/JFH1ΔHVR1/A876P/L665S/S680T	AR5A 0,2mg		G				G/A		C		A			C
Amino acid
Polyprotein position in J6		312	430	534	534	632	640	666	669	676	684	706	747	880
H77 abs. ref. (AF009606) position		312	430	532	532	628	636	662	665	672	680	702	743	876
Amino acid change		T>A	N>D	N>D	N>T	K>T	V>I	L>Q	L>S	L>W	S>T	M>L	Q>R	A>P

NOTE. Substitutions were identified by direct sequence analysis of RT-PCR from recovered viruses present in mouse plasma at 8 weeks postinfection. Blue columns denote substitutions already present in the virus stocks before in vivo adaptation.

AR5, antigenic region 5; E1, HCV envelope glycoprotein E1; E2, HCV envelope glycoprotein E2; HCV, hepatitis C virus; HVR1, hypervariable region 1; NS, nonstructural.

^aSubstitution T680 reverted to S680.

^bVirus escaped AR5A treatment by acquiring the substitution L665S.

Discussion

Several studies have investigated the barrier-to-resistance of HCV NAbs in vitro, but the applicability of these studies to a clinical and vaccine-relevant scenario remain largely unclear. Encouragingly, escape substitutions at positions N415 and N417 have been observed both in humans,^32,33 in experimentally infected chimpanzees treated with the human NAb, HCV1, targeting the linear epitope, AS412,¹⁰ and in vitro in Huh7.5 cell culture treated with AS412-specific antibodies,^29,30 suggesting that in vitro escape studies can recapitulate the appearance of vaccine-resistant variants in vivo. To further establish if translation between model systems in HCV escape studies is possible, particularly for more complex conformational epitopes, we used the HCV-specific cross-genotype-reactive bNAb, AR5A, to test the ability of HCV harboring in vitro-derived resistance substitutions to escape NAb protection in human-liver chimeric mice.

We previously reported high-level AR5A escape caused by the substitutions L665S/S680T identified in vitro. Here, we find that these substitutions are retained in vivo and enabled HCV antibody escape in mice. Furthermore, loading animals with AR5A prior to challenge with wt virus resulted either in protection or rapid development of the escape substitution, L665S, as observed in previous in vitro escape studies.³⁶ Although L665S appeared after 23 days postinfection in vitro,³⁶ we identified L665S 2 weeks postinfection in mice, suggesting that HCV escape may occur even quicker in vivo. As L665S was observed alone at week 2, our data supports focusing on early timepoints to primarily identify antibody-related escape mutations. However, it is possible that the presence of the substitution S680T in the control viruses facilitated the appearance and fixation of L665S by the compensatory effect of S680T on infectivity.³⁶ Sequencing of later infection timepoints revealed multiple other envelope substitutions that we found to increase cell-free infectivity, cell-to-cell spread, and thermal stability.

To analyze the barrier to AR5A resistance in vivo, we used highly neutralization-sensitive HVR1-deleted HCV viruses (>3500-fold more sensitive to AR5A³⁶), which allow us to reach full passive protection in human-liver chimeric mice with lower antibody doses. Due to the natural NAb resistance of HCV in general, and J6/JFH1 in particular, the study of the barrier to resistance of NAbs in cell culture is limited to highly sensitive HCV isolates.³⁴ We have shown that by using HVR1-deleted viruses, we can overcome this limitation and readily identify escape substitutions.36, 37, 38 By challenging with the recombinant, J6/JFH1, we reached only partial protection in human-liver chimeric mice passively immunized with either NAbs AR3A, AR4A, or the polyclonal serum, H06.^23,41 However, full protection against HVR1-deleted J6/JFH1 was achieved with H06, demonstrating the usefulness of highly neutralization-sensitive variants in vivo.⁴¹ We previously described how the substitution A876P, in the HCV protein NS2, increased infectivity and genomic stability of HCV recombinants J6/JFH1 and J6/JFH1_ΔHVR1 in vitro and in vivo.⁴¹ Here, we used A876P to reduce in vivo selective pressure on culture-derived J6/JFH1 viruses. However, despite this, we observed many novel envelope protein substitutions at week 8 postinfection. Thus, although human-liver chimeric mice infected with virus harboring A876A showed higher RNA titers (>10⁶ IU/mL) without the need for additional substitutions during the first 2 weeks postinfection (Figure 1), these viruses may indeed adapt further during extended infection. This suggests that in vivo escape studies should focus on sequencing of early timepoints.

In line with this, we identified the in vivo adaptive substitutions, N430D and M702L, for HVR1-deleted viruses as well as S449P and M702L in viruses with HVR1. Our results suggest that envelope substitutions can enhance HCV in vivo viability by increasing cell-free particle-mediated infectivity (N430D, S449P, and M702L), cell-to-cell infectivity (N430D and M702L), and/or thermal stability at 37°C (N430D). Additionally, N430D abolished the third N-linked glycosylation motif (N3) in E2,^45,46 and S449P substitution also disrupts the consensus sequence (N-X-S/T) of the N448 glycosylation site (N5). Although the loss of glycosylation sites has been observed during in vitro adaptation of HCV,47, 48, 49 our results suggest that glycan loss can also occur in vivo to increase infectivity. Moreover, we have previously observed that HCV substitutions close to the glycan sites N430D (H434R) and N448D (F447V) can appear in mice infected with J6/JFH1_ΔHVR1/A876P,⁴¹ supporting a role of these sites for HCV in vivo adaptation. Lastly, M702L is localized in the stem region of E2 close to S680T. Because the reduction of infectivity induced by L665S is specifically compensated by S680T, it would suggest that the AR5 epitope involves the stem region of E2. In addition, M702 has been shown to be involved in the stability of the E2-E2 homodimer, suggesting that mutations in this position could induce conformational changes in the E1/E2 homodimer.¹⁴ Thus, our data not only suggests that early infection timepoints are the most informative when assessing escape from NAbs in vivo, it also characterizes the effect of common envelope substitutions observed in vivo, which will further facilitate the interpretation of future in vivo escape studies using the in vivo-adapted J6/JFH1 HCV recombinant harboring A876P.

In addition to the effect on virus infectivity, S449P and M702L also broadly affected sensitivity to NAbs and SR-BI dependency. Although S449P increased broad NAb sensitivity and reduced SR-BI dependency, M702L reduced broad NAb sensitivity and increased SR-BI dependency. Further, S449P disrupts the glycosylation site N448, which we have previously shown to broadly increase sensitivity to Nabs.⁴⁵ In contrast, the broad NAb resistance conferred by M702L has been observed for other in vivo-derived substitutions such as D476G, which was identified in sera from a human-liver chimeric mice that was first infused with polyclonal anti-HCV antibody and then infected with J6/JFH1,⁴¹ or L448V, which is a naturally occurring polymorphism identified in an HCV genotype 1-infected patient.⁵⁰ Previously, we and others have shown that changes in HCV broad sensitivity to NAb and SR-BI dependency are due to substitutions that alter the global conformational dynamics of E1/E2.^45,50, 51, 52, 53, 54 Our results suggest that, although S449P induced a Nab-sensitive “open” conformation of E1/E2, M702 induced a less Nab-sensitive “closed” conformation of E1/E2.

In vitro adaptive substitutions have been related to high sensitivity to Nabs, whereas in vitro resistance substitutions are frequently associated with reduced infectivity.^34,47,55,56 In contrast, our results suggest that the appearance of broadly neutralization-resistant variants is likely linked to adaptation in vivo and that epitope disruption in HCV can be acquired without loss of fitness during a natural infection. Furthermore, because broad antibody resistance is associated with an increase in SR-BI dependency, it is possible that the interaction between HCV and SR-BI is more critical in vivo than in vitro. Accordingly, antibodies targeting SR-BI were able to fully protect human-liver chimeric mice from HCV infection^57,58 and prevented on-therapy viral breakthrough of resistant viral variants during DAA therapy.⁵⁹ Thus, further studies are needed to determine the effect of SR-BI dependency on HCV infectivity in vivo.

Initial studies in HCV vaccine development included non-human primates challenge models to estimate the effectiveness and safety of vaccine candidates60, 61, 62; however, due to practical, financial, and ethical considerations, B-cell vaccines studies have shifted to the use of small animal models such as human-liver chimeric mice to evaluate the protection capacity of vaccine-induced Nabs.^17,39,62 Thus, it is important to expand our understanding of virus evolution and adaptation during passive immunization in human-liver chimeric mice. Virus variants with broad resistance to NAbs have been identified after vaccination against hepatitis B virus, as well as SARS-CoV-2 responsible for the global COVID-19 pandemic,63, 64, 65, 66 underscoring the importance of studying virus escape mutations in vaccine development. By expanding our knowledge of in vivo adaptation of HCV under antibody surveillance, we confirm the effects of known in vitro escape mutations for AR5A and characterize more broadly in vivo adaptive mutations, which will be a useful resource in future studies using this model. Thus, our results support the use of cost-effective in vitro models while also simplifying the use of in vivo models to analyze the barrier-to-resistance of NAbs. Using the human-liver chimeric mouse model to study the appearance of HCV vaccine-resistant variants may thus define antibodies conferring rigorous immune surveillance and antigenic determinants capable of eliciting pan-genotypic neutralizing antibodies impervious to viral evolution-mediated immune escape.

Materials and Methods

Antibodies and Reagents

Human monoclonal antibodies, AR3A and AR5A, were produced as described previously.^22,23 Antibody against NS5A, 9E10,⁶⁷ was kindly provided by Charles Rice. Antibodies against the HCV receptors were anti-CD81 (BD Pharminogen, Cat. JS81), anti-SR-BI, C16-17,^51,68 and polyclonal anti-LDLr (R&D Systems, Cat. AF2148). Control antibodies for receptor blocking were antibody 55344 for CD81 (BD Pharminogen), the antibody D for SR-BI,⁶⁸ and a goat IgG (R&D Systems, Cat. AB-108C) for LDLr. Polyclonal antibody IgG, C211, was purified from an HCV genotype 1a-infected patient.⁴⁵ Plasmids with the core-NS2 sequence from genotypes 2a (J6), untranslated regions (UTRs), as well as NS3-NS5B region from genotype 2a (JFH1) either with or without HVR1, were described previously.^49,67 All J6/JFH1 plasmids used in this study contains the NS2 in vivo adaptive substitution A876P.⁴¹ Standard cloning techniques were used to generate plasmids with point substitutions. The sequence of each plasmid was confirmed by Sanger sequencing of the DNA preparation (Macrogen). All nucleotides and amino acids positions in the manuscript are relative to the H77 reference genome, Genbank accession number AF009606.

Cell Culture of Huh7.5 Cells

Huh7.5 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Gibco/Invitrogen Corporation) supplemented with 10% of heat-inactivated fetal bovine serum (FBS), penicillin 100 U/mL, and streptomycin 100 μg/mL (Gibco/Invitrogen Corporation) in 5% of CO₂ at 37°C. Cells were split every 48 to 72 hours unless otherwise indicated.

Transfection of Huh7.5 Cells

Huh7.5 cells were plated at 4 × 10⁵ per well in 6-well plates 24 hours prior to transfection. Template plasmids were linearized by XbaI restriction (New England BioLabs). RNA was generated by T7-mediated in vitro transcription and transfected into Huh7.5 cells using lipofectamine 2000 (Invitrogen). After 6 hours, the cells from each 6 wells were dislodged by trypsin and replated into 4 wells in a 24-well plate at a density of 8 × 10⁴ cells per well. Supernatants containing infectious particles were collected every 24 hours until 96 hours post-transfection, and the percent of infected cells was estimated at each timepoint by reaction with mouse anti-HCV NS5A 9E10 antibody and visualization with Alexa488 goat anti-mouse IgG (H+L) (Invitrogen) secondary antibody, as described previously.⁶⁹ The supernatant virus titers were determined as described previously.⁶⁹ Two-way analysis of variance (ANOVA) was conducted to compare the virus titer of each recombinant against their respective parental viruses at 72 hours post-transfection. A P value < .05 was considered statistically significant. Data and statistical analysis were done in GraphPad Prism 10.

In Vitro HCV Neutralization Assay

Neutralization assays were done as described previously.⁷⁰ Briefly, virus corresponding to a readout of 50 to 200 FFUs was incubated either with serially diluted human monoclonal antibody AR3A or AR5A, for 1 hour at 37°C. Following this preincubation, antibody-virus mixes or virus only were added to 7 × 10³ Huh7.5 cells plated the day before in 96-well plates. Following a 4-hour incubation at 37°C, the cells were washed, fed fresh medium, and maintained further for a total infection time of 48 hours, methanol-fixed, and stained with anti-HCV NS5A antibody. The data were normalized to 8 replicates of virus only. Dose-response neutralization experiments were further analyzed using 3- or 4-parameter curve fitting, and the IC₅₀ was calculated in GraphPad Prism 10.

Virus Stability Assay

Virus supernatants were thawed and incubated at 37°C for 0.5, 1, 2, 3, 4, 6, or 8 hours prior to infectivity titration on 7 × 10³ Huh7.5 cells plated the day before in 96-well plates. Virus titers were determined as described previously.⁶⁹ Two-way ANOVA was conducted to compare of each HCV recombinant values against the respective parental virus. A P value < .05 was considered statistically significant. Data analysis, determination of half-life, and statistical analysis were done in GraphPad Prism 10.

Receptor Blocking

For assessment of receptor blocking, 7 × 10³ Huh7.5 cell were plated per well in 96-well plates and incubated for 24 hours. The next day, Huh7.5 cells were incubated for 1 hour in 4 replicates with a dilution series of antibodies against CD81, SR-BI, or LDLr⁵¹ with concentration from 2.5 to 0.00016 μg/mL for CD81 and SR-BI and 12.5 to 0.0008 μg/mL for LDLr with 4 replicates of the respective control antibodies at the highest concentrations used in the assay. Next, virus stock corresponding to a read-out of 50 to 200 FFUs/well was added to the cell-antibody mix and incubated for 4 hours at 37°C. Cells were washed, fresh medium was added, and the cultures were incubated for a total infection time of 48 hours. Cells were fixed and stained with 9E10 antibody, and FFUs were counted as described for the neutralization assays. The data were normalized to 8 replicates of virus only and analyzed using 3-parameter curve fitting in GraphPad Prism 10. In each graph, 1-way ANOVA was conducted to compare the maximum inhibition between the respective recombinants and the parental virus. A P value < .05 was considered statistically significant.

Cell-to-cell Infectivity Assay

For cell-to-cell infection assessment, 4 × 10⁵ Huh7.5 cells were plated in 6-well plates, and after 24 hours, the cells were transfected with HCV RNA transcripts as described in the section on Huh7.5 cell transfection. The following day, cells were trypsinized and mixed with naive trypsinized Huh7.5 cells prior to plating at 12,000 Huh7.5 cells per well in 2 96-well plates. Six wells per plate for each virus condition were seeded along with 12 wells per plate of only naive Huh7.5 cells to estimate background staining on each plate. For cells transfected with RNA transcript from HVR1-deleted HCV viruses, a ratio of transfected/naive cells of 1:150 were plated in standard medium, and a ratio of 1:30 was used for cells plated in standard medium in the presence of 10 μg/mL of the NAbs AR3A, or 50 μg/mL of polyclonal antibody C211. For cells transfected with RNA transcript from HCV variants retaining HVR1 a ratio of transfected/naive cells of 1:150 were plated in standard medium, and a similar ratio (1:150) was plated in standard medium supplemented with 250 μg/mL of NAb AR3A or 1000 μg/mL of polyclonal antibody C211. The first plate was fixed in methanol after 24 hours post-plating while the second plate was fixed after 48 hours post-plating. The cells were subsequently stained for infection using the NS5A-specific antibody, 9E10, and the number of single infected cells and number and size of FFUs were counted and calculated using adapted BioSpot software (Cellular Technology Lmtd). Data were analyzed and cell-to-cell infectivity in 24 hours was calculated using GraphPad Prism 10. One-way ANOVA was conducted to compare each HCV recombinant value against the parental virus. A P value < .05 was considered statistically significant.

In Vivo Infectivity Assays

Animal studies were carried out according to local and European legislation, with protocols approved by the Animal Ethics Committee of the Ghent University Faculty of Medicine and Health Sciences (approvals ECD 17-93, 19-74 and 22-46). Mice were generated as described.^71,72 Briefly, homozygous uPA-SCID mice were transplanted with cryopreserved primary human hepatocytes (donor 191501; Lonza Bioscience) via intrasplenic injection under isoflurane anesthesia. Six weeks after transplantation, engraftment was assessed by quantifying human albumin in mouse plasma (Bethyl Laboratories Inc). Human-liver chimeric mice were inoculated by intrasplenic injection of 2 × 10³ FFUs of culture-derived HCV. Blood was sampled at week 1 and 2 and every second week until week 16 post-inoculation, according to approved guidelines for blood sampling, processed for EDTA plasma, and stored at −80°C. HCV RNA titers in mouse samples were measured by in-house quantitative polymerase chain reaction (qPCR).

In Vivo Antibody Protection Assays

Human-liver chimeric mice were passively immunized through intraperitoneal (IP) administration of 0.2 mg of NAb AR5A, 24 hours prior to intrasplenic injection with 500 FFUs of indicated culture-derived HCV variants listed in Tables 2 and 3. Blood was sampled at weeks 1, 2, 6, 8, and 17 post-inoculation according to approved guidelines for blood sampling, processed for plasma and stored at −80°C. HCV RNA titers in mouse plasma were measured by in-house qPCR.