Adaptive mutation F772S-enhanced p7-NS4A cooperation facilitates the assembly and release of hepatitis C virus and is associated with lipid droplet enlargement

Hepatitis C virus (HCV) infection is a major cause of chronic hepatitis and liver cancer worldwide. Adaptive mutations play important roles in the development of the HCV replicon and its infectious clones. We and others have previously identified the p7 mutation F772S and the co-presence of NS4A mutations in infectious HCV full-length clones and chimeric recombinants. However, the underlying mechanism of F772S function remains incompletely understood. Here, we investigated the functional role of F772S using an efficient JFH1-based reporter virus with Core-NS2 from genotype 2a strain J6, and we designated J6-p7/JFH1-4A according to the strain origin of the p7 and NS4A sequences. We found that replacing JFH1-4A with J6-4A (wild-type or mutated NS4A) or genotype 2b J8-4A severely attenuated the viability of J6-p7/JFH1-4A. However, passage-recovered viruses that contained J6-p7 all acquired F772S. Introduction of F772S efficiently rescued the viral spread and infectivity titers of J6-p7/J6-4A, which reached the levels of the original J6-p7/JFH1-4A and led to a concomitant increase in RNA replication, assembly and release of viruses with J6-specific p7 and NS4A. These data suggest that an isolate-specific cooperation existed between p7 and NS4A. NS4A exchange- or substitution-mediated viral attenuation was attributed to the RNA sequence, and no p7-NS4A protein interaction was detected. Moreover, we found that F772S-enhanced p7-NS4A cooperation was associated with the enlargement of intracellular lipid droplets. This study therefore provides new insights into the mechanisms of adaptive mutations and facilitates studies on the HCV life cycle and virus–host interaction.


Introduction
Hepatitis C virus (HCV) chronically infects 71 million people worldwide according to the estimation of World Health Organization 1 . HCV infection can lead to chronic hepatitis C, which increases the risk of developing liver fibrosis, cirrhosis, and hepatocellular carcinoma 2,3 . To date, no HCV vaccine is available. Recently, the use of direct-acting antiviral agents (DAAs) has revolutionized HCV therapy and cured ≥ 90% of patients 4 . However, pegylated interferon-α in combination with ribavirin (Peg-IFN/RBV) is still the standard of care for hepatitis C in many countries and/or regions 5 , which has unfavorable adverse effects and only cures~50% of patients 6 . Thus, challenges for hepatitis C treatment remain regarding the introduction of more-effective regimens, especially in DAA-based therapy, more patients who are in need and in optimizing regimens, limiting drug resistance, and ultimately providing a preventative vaccine.
HCV belongs to the hepacivirus genus of the Flaviviridae family. The HCV genome is a positive-sense, single-stranded RNA genome of~9600 nucleotides that consists of a 5′-untranslated region (UTR), an open reading frame (ORF), and a 3′UTR. The ORF is translated into a polyprotein of 3000 amino acids (aa), which is processed into 10 viral proteins by host peptidase and viral proteases, including three structural proteins (Core, E1, and E2), p7, and six nonstructural proteins (NS2, NS3, NS4A, NS4B, NS5A, and NS5B) 7,8 . The structural proteins constitute the viral particle, and the nonstructural proteins are critical for virus replication and other steps of the viral life cycle.
HCV p7 is a small viroporin of 63 amino acids and spans the endoplasmic reticulum (ER) membrane by two transmembrane domains connected by a cytoplasmic loop 9 . The p7 protein is not required for HCV RNA replication 10 , but is required for the entry, assembly, release, envelopment, and production of infectious virions; [11][12][13] it is also critical for cell-to-cell transmission of viral RNA or genomic material 14 . Nevertheless, the function of p7 in the HCV life cycle is not fully understood.
Here, we demonstrated that p7 cooperated with NS4A in an isolate-specific manner. F772S enhanced the cooperation and facilitated viral spread and assembly/release, which were associated with the enlargement of cellular lipid droplets (LD).

Amino-acid position
Recombinant specific 776 1667 1676 H77 reference (AF009606) 772 1663 1672 The nucleotide and amino-acid positions of the specific recombinant with mutations are given, with corresponding position of H77 sequence (AF009606). Shade, engineered mutations. Dominant/minor (capital/lowercase) nucleotides in sequencing reads are shown "." indicates identical to the original nucleotide, * similar results were observed in experiment 2. # , the p7 in J8 contained F768C substitution, important for the viability of the virus of F772S. Hence, A1663V was also an unfavorable change for J6-p7/J6-4A A1663V/A1672S , and F772S was able to compensate for the adverse effects of A1663V, suggesting that A1663 was important for the viability of the recombinant viruses.

F772S coordinated with NS4A by enhancing viral RNA replication
To explore the mechanisms underlying the compensatory effect of F772S, we performed a short-term transfection assay (72 h) to minimize and ignore the impact of mutations possibly emerged. We quantified the relative amount of intracellular HCV-positive RNA (+RNA) and negative RNA (−RNA; an indicator of active HCV replication) in three independent experiments and obtained similar results (Fig. 3). The +RNA levels at 4 h post transfection (p.t.) were set as the input baseline RNA and were close to the replication-deficient J6-p7/JFH1-4A_GND control (Fig. 3a). The +RNA and -RNA levels of recombinants at 72 h was relative to the GND +RNA level (no -RNA for GND) (Fig. 3b, c). At 72 h p.t., the +RNA levels of different recombinants were increased by 2.5-17-fold compared with GND, whereas the −RNA levels were increased by~2.5-fold or less (Fig. 3b, c). The ratios of ±RNAs were 2-9-fold, of which the ratio for J6-p7 F772S /J6-4A and J6-p7 F772S /JFH1-4A were similar, and both were higher than other recombinants (Fig. 3d). Addition of F772S significantly increased the +RNA level of J6-p7/J6-4A, as the +RNA level of J6-p7 F772S /J6-4A was approximately seven-fold higher than that of J6-p7/J6-4A (Fig. 3b).
Taken together, these data suggest that F772S promoted the replication of the viruses with isolate-specific NS4A. However, the enhancement effect of F772S was not observed or only minor at 72 h when NS4A contained A1672S or A1663V/A1672S. F772S enhanced the replication, assembly/release of HCV recombinants with J6-specific p7 and NS4A It is known that p7 is crucial for the assembly and release of infectious HCV particles [11][12][13] . To investigate whether F772S plays a role in these steps, we performed short-term transfection experiment. After 72 h of RNA (10 μg) transfections, we analyzed the infection rate, intracellular infectivity titers, core protein, extracellular infectivity, RNA titers, and specific infectivity (Fig. 4).
Taken together, the results from short-term assays demonstrated that F772S enhanced the assembly and release of HCV recombinants containing isolate-specific p7 and NS4A.
F772S-enhanced p7-NS4A cooperation was at the NS4A RNA sequence level and not at the protein level Next, we proceeded to investigate whether p7 physically interacts with NS4A at the protein level. We overexpressed HA-or Flag-tagged J6-p7 or J6-NS4A (with or b Western blotting of intracellular HCV Core. c The HCV infection rate was estimated by core immunostaining or NS5A-EGFP expression. d Extracellular infectivity titers (FFU/ml). e Extracellular RNA titers (copy/ml). f Specific infectivity (FFU/RNA ratio). See Fig. 3 legend for the detailed annotations without mutations). However, we did not observe a direction interaction of J6-p7 and J6-NS4A in immunoprecipitation, even after several experimental optimizations. Thus, we assumed that other viral proteins or host factors may be required to mediate p7-NS4A interaction in Huh7.5 cells during the complete HCV life cycle.
As replacing JFH1 NS4A into J6-NS4A attenuated the virus (Figs. 1 and 2), we further studied whether this attenuation was owing to differences in the NS4A protein.

F772S-enhanced viral assembly/release was associated with an enlargement of LDs in HCV-infected cells
It is known that LDs are important organelles associated with HCV assembly and production of infectious virus particles 35 . AS p7-NS4A cooperation enhanced viral assembly, we set out to determine whether this effect is associated with the LDs. We performed Oil Red O staining coupled with immunostaining procedures to analyze the size and number of LDs following HCV infection. To minimize the bias of the infection rate on the lipid production, we analyzed the lipid content when the culture reached peak infection (≥90% cells positive for HCV at day 7 post infection) (Fig. 6). The results showed that infectivity titers of HCV recombinants were similar (<0.5 log 10 FFU/ml) (Fig. 6a). We randomly determined the size of the LDs in 10-15 cells and found that the LDs in non-infected control and J6-p7/JFH1-4A-infected Huh7.5 cells were 0.17 μm 2 and 0.51 μm 2 , respectively. Replacement of J6-4A severely reduced the size of LDs (0.13 μm 2 ) (Fig. 6a, b). The size of LDs for J6-p7 F772S /J6-4A infection (0.49 μm 2 ) was comparable to that infected with J6-p7/JFH1-4A (0.51 μm 2 ). Addition of F772S partially restored the viability of the viruses with isolate-specific p7 and NS4A; however, F772S did not significantly alter the LDs of J6-p7/JFH1-4A (J6-p7/JFH1-4A vs J6-p7 F772S /JFH1-4A). Substitution of A1672S and A1663V/A1672S reduced the size of LDs (0.28 μm 2 and 0.14 μm 2 , respectively), and this size reduction could not be rescued by addition of F772S, indicating that the co-presence of S772/A1672 or S772/A1663/A1672 was important for the growth of the LDs (Fig. 6b).
To clarify whether the difference in the enlargement of LDs was the consequence of difference in HCV protein levels, we quantitatively analyzed the viral proteins in the infected cells, in which the size and numbers of LDs were determined. No correlation was identified between the amount of HCV protein and the size and number of LDs (data not shown). These observations were further confirmed by randomly analyzing 26 HCV-positive cells. Indeed, no correlation was found (Fig. 6e-h). For example, J6-p7/J6-4A and J6-p7 F772S /J6-4A associated with a significant difference in the size and numbers of LDs, the correlation coefficient (r value) was 0.1392 (p = 0.4885), thus suggesting a non-correlation between HCV protein levels and the changes in LDs (Fig. 6g). In addition, taking into account no difference in the LDs between HCVinfected (e.g., J6-p7/J6-4A and other NS4A-mutated viruses) and non-infected cells (Fig. 6c, d), the presence of HCV protein or difference in HCV protein levels did not affect the formation of LDs. Therefore, the changes in the LDs were most likely the consequence of the F772S enhanced p7-NS4A cooperation.

Discussion
Adaptive mutations significantly contribute to the development of infectious culture systems that recapitulate the complete life cycle of HCV. In this study, we demonstrated that p7 cooperated with NS4A in an isolate-specific manner. F772S enhanced the cooperation and facilitated viral spread and assembly/release, which are associated with the enlargement of cellular LDs. Understanding the molecular mechanisms of cultureadaptive mutations will facilitate future studies on HCVhost interactions and the development of HCV infectious cell culture systems.
The p7 protein forms oligomers with ion-/proton channel activity 36 , contributing to virion infectivity at the post-assembly step 37 . p7 was found to be essential for infectivity and may interact with other genomic regions in a genotype-specific manner. The amino-and/or carboxylterminal intraluminal tails of p7 contain sequences with genotype-specific functions 38 . However, specific regions potentially involved in interactions with p7 were not pursued in the study. Here, by replacing the JFH1 NS4A sequence of p7-J6/JFH1-4A with another genotype 2a (strain J6) sequence or simultaneously replacing both p7 and NS4A into genotype 2b (J8 strain) sequences, we observed increased virus production (Fig. 1). Virus production was affected by introducing F772S and/or A1663V and A1672S, thus providing a direct evidence for the genotype-specific cooperation of p7 and NS4A. As nonstructural proteins from NS3 to NS5B are sufficient for HCV RNA replication 10 and p7 plays a role in steps of virus replication and assembly 12 , it is likely that the genotype-specific cooperation effects of p7 and NS4A were attributed to the better compatibility of nonstructural proteins in the replication complex and assembly machinery. Nevertheless, no direct physical interactions between p7 and NS4A proteins were identified here in the transient co-expressions. We also demonstrated that the attenuation of NS4A was mediated, at least partially, by J6-4A at RNA level other than the protein level ( Fig. 5b-g). Therefore, these data also suggest that the cooperation of p7 and NS4A involves viral RNAs, which warrant further investigations.
p7 substitutions have been identified in several studies and have been shown to increase the infectivity titers of the JFH1 strain [39][40][41][42] . F772S was identified in several infectious HCV cell culture systems 23,24,33 and proven to enhance virus replication. F772S was mapped within the first transmembrane domain of p7 (ref. 9 ) and was critical for the translocation of viral proteins into the ER membrane. An alanine-scanning assay suggested that F772S enhances the interaction of p7 and Core protein 33 . However, F772W was found to decrease the infectivity titer of J6/JFH1, and F772 might have multiple interactions within p7 and play a role in stabilizing the protein 43 . The co-presence of F772S and V1663A was also identified in several infectious culture systems of J6/JFH1 with genotypes 1-6 NS4A 23 , indicating the importance of the co-presence of S772 and A1663 for virus production.
Our previous studies found that F772S was important for the viability of J6-JFH1 recombinants and J6 fulllength clones 24 . A1672S was recently found to rescue virus replication and assembly defects caused by other NS4A mutations by strengthening the dimerization of the transmembrane domain 44 . However, we found that A1672 was required for the viability of the viruses, since introduction of A1672S or A1663V/A1672S attenuated the J6/ JFH1-EGFPΔ40-based recombinants; In addition, F772Senhanced cooperation of p7-NS4A required A1663 and A1672 (Fig. 3b-f; Fig. 5b-g). These discrepancies might be difficult to explain with a common mechanism because the interactions among HCV RNAs, proteins, and RNAproteins in the complete HCV life cycle remain largely unknown. It is a possibility that the difference in the genome organization of J6cc and the J6-JFH1 chimeras caused the discrepancies. This does not exclude that the attenuation effect of A1672S here might be attributed to the lack of other adaptive mutations, e.g., those identified together with A1672S in the J6cc 24 . It is highly possible that A1672S work together with the L and G mutations, as LSG mutations had been the foundation for the culture adaptation of full-length infectious HCV clones of genotypes 1a, 2a, and 2b, as well as the 5′UTR-NS5A recombinants of 1-6 genotypes [24][25][26][27]45,46 . Our data showed that A1663 was required for F772S-enhanced p7-NS4A cooperation; this is in line with the previous study, in which the co-presence of F772S and V1663A was required for culture adaptation of J6/JFH1-based NS4A recombinants 23 .
Although it was reported that p7 was not required for viral RNA replication 10 , it is indispensable for viral assembly, release and the production of infectious virions [11][12][13] . p7 was found to be involved in the interaction with NS2, which is crucial for the production of infectious HCV particles 47 . Along with NS5B, it has been shown to promote viral assembly 48,49 and is required for colocalization with the core proteins, E2, NS2, NS3, and NS5A 37,50 . NS4A functions as a cofactor for the NS3 protease in completing the chymotrypsin-like folding of NS3 and cleaving MAVS to regulate the innate immune response 17,19 . Moreover, p7 and NS3/4A are components of the NS2 complex and are responsible for HCV particle assembly 51,52 . Our study provides direct evidence for isolate-specific p7-NS4A cooperation important for viral assembly and its association with lipid metabolism.
LDs are recognized as organelles that are essential for HCV particle production, mainly because HCV core proteins recruit nonstructural proteins (NS3 and NS4A) and replication complexes (NS4B, NS5A, and NS5B) to lipid droplet-associated membranes 35 . p7 and NS2 have been shown to be determinants in governing the subcellular localization of the HCV Core to LDs and the ER, (see figure on previous page) Fig. 6 F772S-enhanced viral assembly/release was associated with the enlargement of LDs in HCV-infected cells. Huh7.5 cells were infected with HCV (MOI = 0.1), and~90% of cells were HCV positive at day 7. The cells were fixed and stained for HCV and LDs. a The supernatant infectivity titers at day 7 post infection. b Images of HCV antigens (Core/NS5A-EGFP) (green), nuclei (blue), LD (red), and the merged images. Bar, 10 μm. c The average size of LDs in HCV positive Huh7.5 cells (10-15 cells). d The average numbers of LDs per Huh7.5 cell positive for HCV (by ImageJ software). See Fig. 3 legend for the annotations of statistical analysis. e-h The correlation analysis of the total HCV protein and the formation of LDs in 26 HCVinfected cells using ImageJ software. Correlation coefficient (r value) and p value are shown. p ≤ 0.05 indicates a statistically correlation, in which the higher r value suggests the stronger correlation which is essential for the initiation of the early stage of virus assembly 39 . We found that F772S rescued the size and numbers of LDs that decreased by NS4A replacement (Fig. 6). Moreover, it has been reported that HCV activates NLRP3 inflammasome, which then activates SREBP-1 (lipid-producing transcription factor) 53 , and p7 plays an important role in the activation of NLRP3 (ref. 54 ). Our data showed that the difference in the enlargement of LDs was not due to the difference or presence of HCV protein levels but was due to the F772S-enhanced p7-NS4A cooperation (Fig. 6c-h). Recently, another adaptive mutation, C762Y in p7, was found to increase the size of LDs in JFH1-infected cells 48 . Thus, although they were identified in different HCV genomes, the adaptive p7 mutations F772S and C762Y may share a common mechanism for increasing the size and number of LDs, likely also through the p7-NS4A cooperation. Therefore, our data demonstrate that F772S played a primary role in increasing the size of LDs in cells infected with the HCV recombinant with genotype-specific p7 and NS4A.
In conclusion, our study provided the first evidence of the isolate-specific cooperation of p7 and NS4A and the mode of action of adaptive mutation F772S in the complete HCV life cycle. Taken together, these data suggest that p7 coordinates with NS4A to regulate lipid synthesis for efficient virus RNA replication, viral assembly, and release. Small LDs polymerized to form a large droplet to facilitate viral replication and the assembly of infectious particles.
Plasmids pJ6/JFH1-EGFP△40 contains J6 Core-NS2 and JFH1 5′ UTR/NS3-3′UTR, with enhanced green fluorescent protein (EGFP) insertion in NS5A domain III and a 40-aa-deletion (△40) in domain II (ref. 34 ), was used as backbone plasmid for cloning. It was designated J6-p7/JFH1-4A in this manuscript, according to the strain origin of the p7 and NS4A sequences. All plasmids were confirmed by sequencing. The primers used in this study are listed in Supplementary Table 1. Transfection of plasmids was performed using Lipofectamine 2000 (Life Technologies).

Transcription and RNA transfection
Ten micrograms of HCV plasmids were used for linearization, in vitro transcription, and RNA transfection by following the procedures described previously 24,26,55 . The transfected cultures were left for~16 h and sub-cultured every 2-3 days. The supernatant was collected, filtered (0.45 μm), and stored at −80°C.

Determination of HCV infection, FFUs, and sequences
HCV infection in the culture were determined by immunostaining of anti-Core C7-50 or NS5A-EGFP expression, as previously described 24,26,34 . HCV infectivity titers were determined by the FFU assay 24,26,34 . Sequence analysis of the recovered HCV was performed according to the procedures previously described 24,56 , using primers listed in Supplementary Table 1.

Quantification of intracellular and extracellular HCV RNA
The cells and supernatant were collected at time points as indicated after transfection of equal amount of RNA. J6/ JFH1-EGFP△40-GND was the replication-deficient control. Total intracellular RNA was extracted using TRIzol (Life Technologies), and specific RNA was quantitated by real-time RT-PCR (Supplementary Table 1). The PCR results were analyzed using the equation 2 -ΔΔCq = (Cq Target -Cq Actin ) Sample/X hour −(Cq Target −Cq Actin ) Control/4 h to obtain the fold-change in expression via the 2 -ΔΔCq method. The extracellular level of HCV RNA was quantitated using the COBAS AmpliPrep system 57 .

Immunofluorescence confocal microscopy
Cells were fixed by 4% paraformaldehyde in phosphatebuffered saline (PBS) for 30 min at room temperature, permeabilized with 0.2% TritonX-100 in PBS for 10 min, and blocked with 3% BSA in PBS. Slides were stained with anti-Core C7-50 (1:200 dilution) at 4°C overnight, washed three times with PBS, and stained with IgG-Alexa Fluor-488 (1:250 dilution) (Life Technologies) at room temperature for 2 h. After washing three times with PBS, Oil Red O (Sigma, USA) staining was performed as previously described 58 . The slides were mounted by Prolong Antifade (Life Technologies). Images were collected using a confocal microscope (Zeiss, LSM710) and processed using Adobe-Photoshop-CS5 software.

Statistical analysis
GraphPad Prism software was used to make graphs (GraphPad Software), and mean ±SEM (the standard deviations of the mean) (n = 3 or as indicated) was determined. Student's unpaired t test was used for statistical analysis, and statistical significance was indicated by asterisk(s) (*, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001).

Data availability
We have constructed a number of plasmids in the present study, and the request of plasmids should be addressed to: Yi-Ping Li (lyiping@mail.sysu.edu.cn).