Several groups have demonstrated that single amino acid substitutions in the HCV replicase can dramatically increase the efficiency with which subgenomic replicons initiate replication and persist in Huh-7 cells (5
). In this study we show that Huh-7 sublines, in particular Huh-7.5 cells, possess a cellular environment that is more permissive for the initiation of HCV RNA replication. For the replicon containing the highly adaptive NS5A S2204I mutation, at least 30% of the Huh-7.5 cells can be transduced to G418 resistance. A comparable fraction was positive for the NS3 antigen by FACS. Similarly, for the SG replicons lacking neo
(SG-5′HE in Fig. ), at least 50% initiation efficiency was achieved. It is likely that these are low estimates, since G418 transduction efficiency is based on the number of cells used for electroporation (and not all cells survive). Indeed, more sensitive FACS analysis suggests that >75% of the cells that survive the transfection procedure harbor replicating HCV RNAs (unpublished results). While it is unclear at present how these cells will compare to the natural host cell for HCV infection, liver resident hepatocytes, they nonetheless provide a useful substrate for future genetic and biochemical studies of HCV.
These highly permissive cells were obtained by curing replicon-containing cell clones with IFN-α. While we cannot rule out the possibility that IFN-α-mediated changes persist in the cellular environment, allowing Huh-7.5 cells to be more permissive for HCV replication, the fact that IFN-α treatment of parental Huh-7 cells did not alter the ability of HCV RNA to replicate suggests otherwise. Interestingly, the most highly permissive sublines (Huh-7.5 and Huh-7.8) were obtained from G418-selected clones that harbored replicons without adaptive changes in the NS3-5B region (at least at the population sequence level). These cells may represent a subpopulation of the original Huh-7 parental line that are permissive for replication of unmodified replicons as well as more permissive for replicons with adaptive mutations. Interestingly, curing of other replicon-containing cell lines did not always yield a cell population that was more permissive for the replicons tested. For instance, curing the Huh-7 population containing the SG-Neo (S2204I) replicon, yielded a cell substrate that was unchanged in its ability to support SG-Neo (S2204I) but less efficient (23-fold) for initiation of SG-Neo (5AΔ47) (Fig. ). These results suggest a complex interplay between the cellular environment and particular adaptive mutations to achieve productive replication.
The ability to study HCV replication directly after transfection, without the need for G418 selection, allowed us to examine replication of subgenomic replicons lacking neo
as well as full-length HCV genome RNAs. Our initial attempt to create a monocistronic replicon by fusing cellular ubiquitin in-frame between the first 12 amino acids of core and NS3 [SG-5′Ub-NS3 (S2204I)] (Fig. ) was unsuccessful. A bicistronic derivative with ubiquitin fused to NS3-5B was viable, suggesting that the failure of SG-5′Ub-NS3 (S2204I) to replicate was not due to a defect in processing at the ubiquitin/NS3 junction (data not shown). Rather, the fusion of the ubiquitin-coding sequence near the HCV 5′ NTR may have interfered with translation due to the formation of deleterious RNA secondary structures (20
) or RNA replication, by disrupting RNA elements that lie in the HCV 5′ NTR or its complement (7
). Fusion of the HCV 5′ NTR to the EMCV IRES yielded a subgenomic replicon (SG-5′HE) that replicated better than SG-Neo replicons (Fig. ). Why deletion of the first cistron (sequences encoding C-Neo fusion) from the bicistronic SG-Neo (S2204I) stimulated replication is unknown, but the stimulation may result from enhanced translation of the replicase due to eliminated binding of the 40S ribosomal subunit to the usual HCV translation initiation site and diminished competition between the EMCV and HCV IRES elements (J. Marcotrigiano and C. M. Rice, unpublished results).
A similar picture was observed for replication of the full-length constructs containing the NS5A S2204I adaptive change (Fig. ). In Huh-7.5 cells, the bicistronic construct containing the C-Neo cistron [FL-Neo (S2204I)] (Fig. ) initiated replication less efficiently than the RNA with the HCV 5′ NTR fused to the EMCV IRES [FL-5′HE (S2204I)] (Fig. ). Interestingly, the FL construct with the unmodified HCV genome (except for the S2204I substitution in NS5A) was better at initiating replication than RNA whose translation was mediated by the EMCV IRES (Fig. ), demonstrating that EMCV IRES-driven translation is not required for HCV replication in Huh-7.5 cells, thus allowing the study of unmodified HCV genomic RNAs. This point could certainly impact the ability of HCV RNAs to be packaged into infectious particles. However, in our hands (Blight and Rice, unpublished results) and in a recent report (22
) selective packaging of these unmodified FL RNAs was not observed in Huh-7 cells. Interestingly, full-length HCV RNAs were less efficient at establishing replication than the corresponding adapted subgenomic replicons, suggesting that addition of the structural-NS2 coding region inhibits HCV replication initiation. Whether this is due to the encoded proteins or RNA elements that lie in this region (or both) (31
) is currently unclear.
In an attempt to further enhance HCV replication in cell culture, we also examined the effect of other amino acid substitutions at position 2204 in NS5A. Efficient subgenomic RNA replication was observed for Ile and Val and, to a much lesser extent, Ala at position 2204 (Fig. ). Val or Ile are small, β-branched, nonpolar residues, whereas Ala has similar properties but is not β-branched. In contrast, polar residues such as Tyr, Glu, Thr, Ser, or Asp at position 2204 severely impaired HCV replication (Fig. ), suggesting that replication favors nonpolar residues at this locus. It is interesting that Ser is found naturally at position 2204 for this genotype 1b isolate (18
) and is conserved between other HCV genotypes (30
), suggesting that this residue may be important for HCV replication and/or pathogenesis in vivo.
Combining NS5A adaptive mutations resulted in replicons that were either impaired (A2199T + S2204I) or unable to replicate (S2197P + A2199T + S2204I) in Huh-7.5 cells (Fig. ). Incompatibility of adaptive mutations elsewhere in the HCV NS coding region has been previously described (17
). For example, combining an adaptive mutation in NS5B (R2884G) with either NS4B (P1936S) or NS5A (E2163G) drastically reduced the efficiency of G418-resistant colony formation. On the other hand, combining certain NS3 and NS5A adaptive mutations can increase replication efficiency (15
). However, despite the observation that mutations E1202G and T1280I in NS3 act synergistically with S2197P in NS5A to increase the replication efficiency (15
; Blight and Rice, unpublished results), engineering these NS3 changes into SG-5′HE (S2204I) did not enhance replication in our system (Fig. ). These results again underscore the empirical nature of optimizing adaptive mutations with different Huh-7 cellular environments.
The phosphorylation of NS5A is conserved among divergent HCV genotypes (25
), suggesting that it plays an important role in the virus life cycle. We previously showed that NS5A hyperphosphorylation is not essential for HCV replication (5
). Following the recent identification of S2194 as a major phosphate acceptor site for subtype 1b (12
), we substituted Ala or Asp at this position and examined the effect on HCV replication in the context of the S2204I adaptive mutation. Given the incompatibilities observed when combining NS5A mutations, the absolute replication efficiencies of the different mutants could not be evaluated; however, replicating RNAs were recovered that harbored these substitutions at the 2194 locus. These results show that phosphorylation at S2194 is not an absolute requirement for replication of this subtype 1b isolate. While this observation deserves further study in the context of adaptive mutations outside of NS5A, it should be noted that NS5A phosphorylation is complex and only a few of the potential serine acceptor sites have been identified (12
). Hence, additional phosphorylation sites need to be defined and coupled with mutagenesis of these sites as well as a thorough mutational analysis of the NS5A protein.
In conclusion, we have isolated a Huh-7 subline (Huh-7.5) that is highly permissive for replication of subgenomic and full-length HCV RNAs. These cells provide a valuable substrate for future genetic studies on HCV proteins and RNA elements. Finally, Huh-7.5 and other cured Huh-7 cells that differ in their ability to support HCV replication may prove useful for defining cellular parameters that affect the efficiency of HCV RNA replication initiation by gene array and other approaches.