|Home | About | Journals | Submit | Contact Us | Français|
An infectious cDNA clone of a genotype 3 strain of hepatitis E virus adapted to growth in HepG2/C3A human hepatoma cells was constructed. This virus was unusual in that the hypervariable region of the adapted virus contained a 171-nucleotide insertion that encoded 58 amino acids of human S17 ribosomal protein. Analyses of virus from six serial passages indicated that genomes with this insert, although initially rare, were selected during the first passage, suggesting it conferred a significant growth advantage. RNA transcripts from this cDNA and the viruses encoded by them were infectious for cells of both human and swine origin, the major host species for this zoonotic virus. Mutagenesis studies demonstrated that the S17 insert was a major factor in cell culture adaptation. Introduction of 54 synonymous mutations into the insert had no detectable effect, thus implicating protein, rather than RNA, as the important component. Truncation of the insert by 50% decreased the levels of successful transfection by ~3-fold. Substitution of the S17 sequence by a different ribosomal protein sequence or by GTPase-activating protein sequence resulted in a partial enhancement of transfection levels, whereas substitution with 58 amino acids of green fluorescent protein had no effect. Therefore, both the sequence length and the amino acid composition of the insert were important. The S17 sequence did not affect transfection of human hepatoma cells when inserted into the hypervariable region of a genotype 1 strain, but this chimeric genome acquired a dramatic ability to replicate in hamster cells.
Hepatitis E virus (HEV) is a small RNA virus that belongs to the genus Hepevirus in the family Hepeviridae (18). It causes hepatitis E, a disease clinically indistinguishable from hepatitis A (7). Initially, it was thought to occur only in underdeveloped regions of the world, where it caused waterborne epidemics and sporadic disease. However, in the past decade it has emerged as a sporadic disease in industrialized countries, including Great Britain, the European Union, and the United States (17, 23). This “emergence” is due to the recognition that there are not one, but four genotypes of HEV (genotypes 1 to 4) that infect humans and that genotypes 3 and 4 also routinely infect swine and occasionally other species (16). Genotypes 1 and 2 are found mainly in underdeveloped countries, where they are spread via contaminated water: in contrast, genotypes 3 and 4 are zoonotic and are found in industrialized countries, where they are spread mainly through eating undercooked pork or game products.
HEV infection was long thought to be an acute infection, lasting 2 to 7 weeks, that never progressed to chronicity. Recently, however, chronic HEV infection has been identified in immunosuppressed organ transplant patients or in AIDS patients (3, 10–12, 24, 25). Even more unexpectedly, some of these chronically ill patients have developed neurological symptoms (11, 12), and HEV has been isolated from cerebrospinal fluid (11). These chronic cases have been identified as genotype 3 infections.
The 7.2-kb genome of HEV (Fig. 1) is a single strand of positive-sense RNA with three overlapping reading frames (ORFs) (18). Approximately the first 5 kb serve as mRNA for the ORF1 polyprotein; it is not known whether the polyprotein is proteolytically processed. ORF1 contains regions encoding methyl transferase/guanylyltransferase, NTPase/helicase, RNA-dependent RNA polymerase and deubiquinating activities (13). In addition, ORF1 encodes a Y region and an X or macro region of unknown function and a hypervariable region (HVR) of approximately 250 nucleotides (nt) that is located near the middle of the ORF (1). The HVR varies in length and sequence among strains and genotypes: it tolerates small deletions, but replication levels of deletion mutants are severely depressed in cell culture (27). ORF2 and ORF3 are translated from a single bicistronic, subgenomic RNA to produce a 660-amino-acid (aa) capsid protein and a 113- to 114-aa multifunctional protein, respectively (9). The ORF3 protein is required for efficient release of virus particles from cultured cells and is required for the infection of macaques (5, 8, 33).
HEV studies were long hindered by the absence of an efficient cell culture system for any of the genotypes. That deficit was partially overcome when Okamoto and coworkers (22) succeeded in adapting a genotype 3 (29) and a genotype 4 (30) strain, isolated from acutely infected patients, to grow in cultured PLC/PRF/5 human hepatoma cells. These researchers also constructed an infectious cDNA clone of the genotype 3 virus and implicated point mutations as the basis for the adaptation (32). Shukla et al. (28) subsequently adapted the Kernow C-1 strain of genotype 3 virus, isolated from a chronically infected patient, to grow in cultured HepG2/C3A cells. This adapted virus differed strikingly from the two adapted acute strains by containing, in addition to point mutations, a 58-aa human S17 ribosomal protein fragment inserted into the HVR. In the present study, the adapted Kernow strain was cloned as an infectious cDNA, and mutagenesis studies were performed to determine whether the inserted S17 sequence played a role in cell culture adaptation and/or host range.
Passage 1 Kernow-C1 virus RNA was extracted from the medium of infected HepG2/C3A cells with TRIzol LS reagent (Life Technologies). Reverse transcription-PCR (RT-PCR) was performed with SuperScript II reverse transcriptase (Life Technologies), Herculase HotStart Taq (Stratagene), and PrimeStar HS DNA polymerase (TaKaRa). A total of six overlapping fragments were joined by fusion PCR and ligation to generate the full-length genome. A unique XbaI restriction site and a T7 RNA polymerase core promoter were engineered into the 5′ end and a stretch of 36 adenosines, followed by a unique MluI site and a HindII site engineered into the 3′ end. The full-length genomic cDNA was ligated into pBlueScript SK(+) (Stratagene) between the XbaI and HindII sites of the polylinker to generate p1. p6 was synthesized using standard methods to replace restriction fragments of p1 with the corresponding fragments amplified by RT-PCR from the medium of passage 6 virus cultures. The GTPase sequence was amplified from a cDNA clone of the HVR region recovered from the feces of a U.S. patient (21). The green fluorescent protein (GFP) sequences were amplified from a modified genotype 1 HEV cDNA clone described previously (4). The gaussia luciferase gene was amplified from the pGLuc basic vector purchased from New England Biolabs. Unique NruI and AatII restriction sites introduced at the 5′ and 3′ termini of the S17 sequence were used to construct the S17 truncations. RNA folding was predicted with Mfold (34).
RNA was extracted with TRIzol LS (Invitrogen). The HVR with flanking regions (nt 1986 to 2762) was amplified by nested RT-PCR with a Qiagen Long-Range 2 Step RT-PCR kit. The visible product, including the neighboring regions above and below it, was eluted from an agarose gel and directly sequenced (consensus sequence) or cloned with the Zero Blunt TOPO PCR cloning kit (Invitrogen). RNA genomes in culture medium were quantified by real-time RT-PCR (TaqMan). Primer sequences and amplification conditions will be provided on request.
S10-3 cells are a subclone of human Huh-7 cells isolated in-house. Human HepG2/C3A (CRL-10741), swine LLC-PK1 (CL-101), and hamster BHK-21 (CCL-10) cells were purchased from the American Type Culture Collection; HepG2/C3A cells were grown on rat tail collagen type 1 (Millipore). Cells were propagated in Dulbecco modified Eagle medium supplemented with l-glutamine, penicillin-streptomycin, gentamicin, and 10% fetal bovine serum (Ultra-Low immunoglobulin G from Invitrogen).
Plasmids were linearized at a 3′ terminal MluI (Kernow-related) or BglII (Sar-related, GenBank accession no. AF444002) sites. Capped RNA transcripts were generated with a T7 riboprobe in vitro transcription system (Promega) and Anti-Reverse Cap Analog (Ambion) as described previously (6). For transfection of S10-3 and BHK-21 cells, 23 μl of RNA transcription mixture, 1 ml of Opti-MEM (Gibco), and 20 μl of DMRIE-C (Invitrogen) were mixed and added to cells in a T25 flask. After incubation with transfection mixture for 5 h at 34.5°C in a CO2 incubator, the transfection mixture was replaced with culture medium, and incubation was continued at 34.5°C.
HepG2/C3A and LLC-PK1 cells were killed by DMRIE-C, so they were transfected by electroporation using a Bio-Rad Gene Pulser II at settings of 240 V and 950 capacitance using Bio-Rad cuvette 165-2086. RNA transcripts from a 100-μl transcription mixture were extracted with TRIzol LS (Invitrogen), precipitated with isopropanol, washed with 75% ethanol, and resuspended in 50 μl of water. Confluent cells in a 100-mm dish were, treated with trypsin, mixed with an equal volume of 1% crystalline bovine serum albumin in phosphate-buffered saline (PBS), and pelleted at 525 × g at 4°C for 5 min. The cells were resuspended in 400 μl of Opti-MEM (Gibco), mixed with the RNA, pulsed, added to culture medium containing 20% fetal bovine serum, and incubated at 37°C (HepG2/C3A) or 34.5°C (LLC-PK1) overnight; HepG2/C3A electroporated cells in a T25 flask were supplemented with one-fourth of the untreated cells from a T25 flask in order to provide a dense enough culture to promote growth. The next morning, medium was replaced with fresh medium containing 10% serum, and the incubation was continued.
Approximately 100,000 cells/well were seeded onto eight-well Lab-Tek II CC2 slides (Nunc) a day before infection. Virus samples were diluted in Opti-MEM or cell culture medium, and 100 μl of the diluted virus was added to each well, followed by incubation for 5 h at 34.5°C in a CO2 incubator. The virus mixture was removed, and cell culture medium was added, followed by incubation at 34.5°C.
Transfected or infected cells on chamber slides were washed with PBS and fixed and permeabilized with acetone. ORF2 and ORF3 proteins were detected with a mixture of HEV ORF2-specific hyperimmune plasma from an HEV-infected chimpanzee (Ch1313) (6), and rabbit anti-ORF3 peptide antibody (4): the chimpanzee plasma was preadsorbed on the respective cells to minimize background staining. Secondary antibodies were a mixture of Alexa Fluor 488-conjugated goat anti-human IgG (Molecular Probes) and Alexa Fluor 568-conjugated goat anti-rabbit IgG (Molecular Probes). Stained cells were overlaid with Vectashield mounting medium with DAPI (4′,6′-diamidino-2-phenylindole; Vector Laboratories) and visualized at 40× with a Zeiss Axioscope 2 Plus fluorescence microscope. Positive cells or foci were manually counted.
Trypsinized cells were pelleted at 525 × g, mixed with 1 ml of methanol for 15 min at 4°C, and stored at −80°C until all of the samples from one experiment were stained in parallel. The cells were pelleted out of methanol, washed once with 5 ml of PBS, and resuspended in 100 μl of blocking solution (0.5% skim milk, 0.5% crystalline bovine serum albumin, and 0.1% Triton X-100 in PBS) at room temperature for 30 min before the addition of 100 μl of 2× chimp 1313 preadsorbed serum; the cells were then washed with 10 ml of PBS and resuspended in 100 μl of anti-human Alexa Fluor 488-conjugated antibody. After 30 min, the cells were washed with 10 ml of PBS, resuspended in ~0.5 ml of PBS, and analyzed using a FACScan flow cytometer (Becton Dickinson). A total of 20,000 to 40,000 events were acquired for each sample, and the data were analyzed using BD CellQuest software.
Every 24 h posttransfection, all of the medium was removed from the culture, filtered through a 0.45-μm-pore-size Millex HV filter (Millipore), and stored at −80°C. Fresh medium was then added to the culture, and incubation was continued. Media from the entire experiment were thawed together, and 20 μl was assayed for luciferase activity using the Renilla luciferase assay system (Promega) according to the manufacturer's protocol. Briefly, 20 μl of culture medium was added per well of a 96-well black, flat-bottom microplate (Corning), followed by the addition of Renilla luciferase assay substrate and the detection of luminescence using a Synergy 2 Multi-Mode microplate reader (Bio-Tek, Winooski, VT). The microplate reader was set to dispense 50 μl of substrate, followed by shaking for 2 s and reading for 5 s. Samples were assayed in triplicate and read sequentially.
Sequences have been deposited in GenBank under accession numbers JQ679013 to JQ679025.
The fact that the Kernow virus adapted by six serial passages in HepG2/C3A cells was a recombinant virus containing part of a human S17 gene was not discovered until the passage 6 virus was sequenced (28). Although viral genomes containing the 171-nt S17 sequence could be detected in the feces by nested RT-PCR with virus-human primer pairs (25), they constituted such a minor quasispecies that they were not represented in 120 cDNA clones of the HVR region of viruses in the fecal inoculum (data not shown). In order to determine when in the passage series the virus containing this insert first emerged and when it became the dominant species, the HVR region of viruses in the medium at each of the six cell culture passage levels was amplified by RT-PCR, cloned, and sequenced. Two of eleven clones from the first passage already contained the S17 sequence and from passage 2 onward; it was present in the majority of clones (Table 1). Amazingly, a different mammalian gene insert, 114 nt long, was present in 5 other of the 11 clones from the first cell culture passage and, in this case, an almost identical sequence was found in 2 of the 120 clones from the feces (Fig. 1). This 114-nt sequence lacked 10 nt in the middle of the GTPase-activating protein gene sequence (GenBank accession no. AB384614.1) and consisted of a rearranged gene segment in which GTPase nt 3009 to 3105 were followed by GTPase nt 2981 to 3008, and the reading frame was changed so that the sequence, as inserted, encoded an unrelated amino acid sequence that did not match anything when BLAST analyzed against all known nonredundant protein databases. However, this insert was not detected in any of the clones from subsequent passages 2 through 6.
The medium of cultured cells should contain the members of a virus quasispecies that are best able to infect and complete a replication cycle in these cells. Therefore, the first full-length cDNA clone of the Kernow virus was constructed from uncloned cDNA fragments amplified from the medium (passage 1 virus) of HepG2/C3A cells that had been inoculated 111 days previously with a stool suspension containing the original Kernow strain. This Kernow passage 1 cDNA clone, p1 (GenBank accession no. JQ679014), lacked the S17 insert and differed from the consensus sequence of virus in the feces (GenBank HQ389543) by 16 aa, including an extra proline (Table 2). It was transfected into S10-3 hepatoma cells, which were monitored 5 to 6 days later by immunofluorescence microscopy for cells stained for ORF2 protein. Less than 2% of the S10-3 cells transfected with in vitro transcripts of the passage 1 clone produced detectable ORF2 protein, suggesting that this virus genome, although infectious, lacked elements essential for robust replication. Incorporation of the S17 insert into the cDNA clone to yield p1/S17 increased the number of cells successfully transfected, but the levels remained below ~10%. In an attempt to derive a more robust virus and to identify regions that contributed to cell culture adaptation, convenient restriction fragments of p1/S17 cDNAs were sequentially replaced with the quasispecies of uncloned PCR product amplified from passage 6, cell-culture-adapted virus (Table 2). Transcripts from multiple clones of these new full-length genomes were transfected into S10-3 cells and examined for ORF2 production by immunofluorescence microscopy. The clone producing the highest percentage of transfected cells was used as the backbone for the next substitution, and this process was repeated four more times. Finally, all clones were compared by flow cytometry in the same experiment (Fig. 2). The first three sequential fragment substitutions had introduced mutations into the 3′ ORF2 and noncoding regions [nt 6812-poly(A)], into the 3′ ORF1 and ORF2/ORF3 overlap (nt 4608 to 6812), and into the 5′ third of ORF1 (nt 671 to 2182): of the three fragments, only the 6812-An substitution significantly increased the level of successful transfections (Fig. 2). Unexpectedly, of the passage 6 PCR amplicons spanning nt 4608 to 6812, the sequence that boosted successful transfection levels the most contained mutations that eliminated the only two methionine codons (aa 1 and 69) in ORF3 (Table 2); immunofluorescence microscopy confirmed that viruses from this cDNA clone and the three subsequent cDNA clones did not produce ORF3 protein (data not shown). The passage 6 fragment with the greatest enhancing effect spanned nt 2182 to 3063 and contained three naturally occurring amino acid mutations in the X domain and a single proline deletion in the HVR: additionally, four proline codons in this fragment were changed by site-directed mutagenesis to CCA codons in order to preserve the amino acid sequence while disrupting a cluster of C residues in the HVR that greatly hindered PCR and sequence analyses. The fifth fragment substitution (nt 3063 to 4608) contained a highly conserved region of the helicase and polymerase genes, did not introduce any amino acid changes, and had no obvious effect (P = 0.067). Finally, the methionine initiation codon of ORF3 was restored so that ORF3 protein could be produced by the p6 virus. The presence or absence of this methionine codon had no apparent effect on levels of successful transfection of S10-3 cells (compare the p6/ORF3-null and p6 transfection levels in Fig. 1 [P = 1.0]). Western blots of cell lysates detected ORF3 protein in p6 transfected cells but not in p6/ORF3 null transfected cells (data not shown). The p6 clone (GenBank accession no. JQ679013), excluding the insert, differed from the stool consensus sequence (GenBank accession no. HQ389543) by 16 aa and from p1 by 25 aa but from the passage 6 consensus sequence (GenBank accession no. HQ709170) by only 2 aa (aa 598 = R to C in ORF1 and aa 593 = T to A in ORF2). Transcripts of the final p6 clone routinely transfected 10 to 45% of S10-3 cells.
Because the function of the X domain is unknown and the C-to-A changes in proline codons of the HVR were engineered rather than natural, it was important to determine whether the three mutations in the X domain or the C-to-A synonymous mutations in the HVR in fragment from nt 2182 to 3063 were the most important for enhancing transfection. Because back-mutation of the proline codons would recreate the sequencing problems, the amino acid codons in the X domain were chosen for back mutation. All three mutations in the X domain were simultaneously back-mutated to the original codons present in the p1 cDNA clone, and the level of successful transfections was quantified by flow cytometry at day 6 posttransfection (Fig. 3). Transcripts from the clone containing the three reverted X domain mutations were significantly (P = 0.0006) less efficient than those from the p6 cDNA clone in successfully transfecting S10-3 cells and not significantly different (P = 0.12) from the 671-2182 clone, which lacked both the HVR proline mutations and the X domain mutations, suggesting that the engineered changes that interrupted the poly(C) tract had a minimal effect on successful transfections, whereas one or more of the three mutations in the X region played a critical role.
In order to determine whether the effect of the S17 sequence was limited to the modest increase in transfection levels observed following its insertion into the p1 cDNA clone, the S17 sequence was selectively removed from the p6 cDNA clone containing all of the point mutations to yield p6delS17. Flow cytometry confirmed that addition of S17 sequence to p1 virus genomes significantly increased successful transfections by these genomes, although the levels did not approach those attained by the recombinant p6 genomes (Fig. 4). Surprisingly, removal of the S17 sequence from the p6 cell culture-adapted cDNA clone dramatically decreased successful transfections by the genome transcripts to levels only 3-fold better than those of the p1delS17 cDNA clone (Fig. 4). This result suggested that the point mutations responsible for the incremental improvement in successful transfections by the serial clones were mostly ineffective in the absence of S17 sequence.
The flow cytometry analyses based on ORF2 protein immunostaining revealed the percentage of cells that produced detectable ORF2 protein, but they did not provide a quantitative comparison of the amount of ORF2 protein produced or of the duration of ORF2 synthesis. In order to confirm and extend the flow cytometry data, the 5′ portion of ORF2 was replaced with the in-frame gaussia luciferase reporter gene to yield p6/luc: this luciferase has a signal sequence that results in its secretion and accumulation in the cell culture medium. Therefore, sequential time points can be obtained from the same culture.
The luciferase system was validated by measuring the amount of luciferase secreted into the medium by p6/luc virus containing either a functional polymerase or a mutated, nonfunctional polymerase that could not synthesize viral RNA. Whereas the luciferase signal in medium from S10-3 cells transfected with the p6/luc defective polymerase mutant or from untransfected S10-3 cells was less than 111 U/24 h at its peak on day 2, that in the medium of cells transfected with p6/luc rose from 2,163 U/24 h on day 1 to over 36 million units/24 h on days 4 through 6 (data not shown).
Therefore, luciferase production requires viral RNA synthesis, as predicted based on the ORF2 location of the luciferase gene in the subgenomic mRNA. Luciferase production by p6/luc virus was then compared to that by the p6/luc virus mutated to either delete the S17 insert or to eliminate the three X gene mutations. The production of luciferase by p6/luc virus, either with or without the S17 insert, peaked on day 6 posttransfection, but the ratio of p6/luc units to p6/luc(del S17) units steadily increased and reached 52-fold at day 7, thus confirming that the S17 insert conferred a significant growth advantage and served as a cell culture-adaptive mutation (Fig. 5A). In addition, the luciferase data for the three X gene back-mutations was consistent with that from the flow cytometry analyses (Fig. 5B); cultures transfected with the p6/luc X gene revertant produced less luciferase than those transfected with the p6/luc virus. It is interesting that, as shown in both Fig. 5A and B, the luciferase values on day 9 decreased substantially for the p6/luc virus but remained near plateau levels for the mutants.
The enhancing effect of the S17 insert could be due to either the RNA or the protein sequence. In an attempt to distinguish between these possibilities, the third base in 93 and 70% of the 58 codons in S17 was changed (purine to purine and pyrimidine to pyrimidine) in two clones without altering the encoded AA except for two M-to-I changes. The number of cells successfully transfected by each of these two clones did not differ significantly from that of p6 (Fig. 6), even though the predicted RNA structures and ΔG (Gibbs free energy) differed from those of p6 (ΔG = −120.38) and each other (ΔG = −102.99 and 98.97 for clones 1 and 2, respectively). Therefore, it appeared that the enhancing effect probably occurred at the protein level.
The previous studies demonstrated the importance of the S17 insert for growth of the Kernow virus in cell culture but did not provide any insights into how it functioned. Since the first passage of the stool inoculum had provided evidence for a possible enhancing effect of the GTPase insert on growth of the Kernow strain in cell culture, this insert was substituted for that of the S17 insert in the p6 clone to evaluate its effect. Although the 114-nt GTPase insert increased the number of successfully transfected cells, it was only about half as effective as the 171-nt S17 insert (Fig. 7A). In order to determine whether length of the insert per se was a factor, sequence encoding the N-terminal or C-terminal 58 aa (174 nt) of GFP was substituted for the S17 sequence. GFP was chosen because it has been shown to be relatively benign when expressed as a fusion protein with many partners in many cell types. Indeed, fluorescence microscopy indicated that GFP was produced when the entire coding region was fused in-frame to the 3′ terminus of the S17 insert (data not shown); however, neither of the 174 nt encoding 58 aa of the N-terminal or C-terminal amino acids of GFP had a detectable effect on the levels of transfection of S10-3 cells (Fig. 7A and B). Therefore, the number of nucleotides and/or amino acids in itself is not a determining factor. The effect of size was tested also by removing approximately half of the nucleotides from the S17 inserted sequence in p6 to yield 87 to 90 nt of sequence encoding the N-terminal half, the C-terminal half, or the middle portion of the S17 insert. All three constructs successfully transfected cells to a similar extent that averaged 2- to 6-fold less than if the entire insert was present and 2.5-fold more than if it was absent (Fig. 7C). Finally, 117 nt of another mammalian gene sequence (S19 ribosomal protein) which we discovered inserted into the HVR of another genotype 3 strain from a different chronically infected hepatitis E patient (21), was substituted for the S17 sequence in the p6 clone. Although genomes carrying the S19 sequence in this different genotype 3 strain had been selected during culture in HepG2/C3A cells, much as had the S17-containing Kernow genomes (21), transfer of this sequence from that genotype 3 strain to the Kernow strain resulted only in a modest enhancement (Fig. 7D).
Both the S10-3 cells used for transfection and the HepG2/C3A cells to which the passage 6 virus was adapted are human hepatoma cells, so it was important to determine whether the p6 virus retained the ability of the original fecal inoculum to grow also in swine cells (28). Transcripts of p6 and p6del S17 were electroporated into LLC-PK1 swine kidney cells, which were assayed by flow cytometry 5 days later. ORF2 protein was produced in swine cells by both constructs, demonstrating that both the negative strand genomes and the subgenomic mRNAs had been synthesized by each. More than 31% of the swine cells were transfected by the p6 clone compared to 12% by the p6 clone missing the S17 insert, thus demonstrating that the S17 sequence enhanced transfection of swine cells as it did human cells (Fig. 8A). Next, the p6 virus itself was tested for the ability to infect swine cells. Two different lots of p6 virus grown in HepG2/C3A cells were titered in parallel on HepG2/C3A cells and LLC-PK cells (Fig. 8B). In both cases, the infectivity titer was higher on the swine cells than on the human cells; although the difference varied for the two preparations and reached significance in only one case, the p6 cDNA clone clearly did encode a virus which could infect cultured cells originating from each of the two major host species for genotype 3 HEV.
Transcripts from a genotype 1 cDNA clone, Sar55, transfect S10-3 cells readily, but extensive attempts to adapt the virus to grow in cell culture have thus far failed (S. Emerson, unpublished data). A previous experiment had demonstrated that recombinant Sar55 genomes containing the S17 sequence from p6 virus in their HVR (Sar55/S17) were able to transfect S10-3 cells, but a quantitative comparison with Sar55 genomes lacking the insert was not performed (28). Therefore, Sar55 and Sar55/S17 transcripts were transfected into S10-3 cells, which were subjected to flow cytometry 5 days later. Both sets of transcripts produced a similar number of ORF2-positive cells, suggesting that the S17 sequence neither enhanced nor diminished successful transfection Sar55 genomes in this system (Fig. 9A).
Since the Kernow virus had displayed such a diverse host range previously (28), p6 transcripts were tested for the ability to transfect hamster BHK-21 cells and were found to produce ORF2-positive cells, although with low efficiency (3.8% for BHK-21 compared to 30.1% for S10-3). Therefore, the Sar55 and Sar55/S17 transcripts also were tested by flow cytometry for the ability to transfect BHK-21 cells, even though these cells were an unlikely host given the restricted host range of genotype 1 viruses. Amazingly, not only were the hamster cells transfected by the Sar55 genomes, the number of transfected cells was boosted almost 7-fold by inclusion of the S17 insert (Fig. 8B, P < 0.0001). The marked enhancement of transfection by the S17 insert was confirmed by immunofluorescence microscopy in an independent experiment (Fig. 9C).
Since the p6 cDNA genome was derived from virus adapted to grow in HepG2/C3A cells, the virus encoded by this cDNA clone was predicted to replicate and spread efficiently in cultures of these cells: in contrast, previous studies implicating ORF3 protein in virus egress (5, 33) suggested that a p6/ORF3-null virus genome incapable of producing ORF3 might transfect as many cells as did p6 genomes but that virus would not spread to other cells. p6 virus genomes and p6/ORF3-null genomes were electroporated into HepG2/C3A cells, and virus production and spread were monitored by flow cytometry. The p6 virus and the ORF3-null mutant displayed surprisingly similar patterns, and both appeared to replicate and spread efficiently throughout the culture: in both cases, the percentage of ORF2 protein-positive cells increased from ca. 15% on day 5 to more than 70% on day 14 (Fig. 10). An independent experiment produced similar results with the percentage of positive cells increasing from 12.4% ± 1.96% to 59.7% ± 0.87% for p6 virus and from 13.3% ± 0.31% to 67.8% ± 5.57% for the ORF3-null mutant between days 5 to 15. Although these results demonstrated that the p6 clone did indeed encode a cell-culture-adapted virus, the similar levels of cell-to-cell spread for the two viruses was puzzling because both our laboratory (5) and that of Okamoto (33) had published data, demonstrating that efficient viral egress required functional ORF3 protein; in these reports, virus release in the absence of ORF3 protein was only ca. 10% as much as that in its presence. Sequence analysis of the ORF3 region of the null mutant genomes amplified by RT-PCR from the day 9 medium confirmed that no methionine codons were present, and ORF3 protein was not detected by immunofluorescence microscopy of the cells (data not shown). However, an infectious focus assay performed with the medium from the two cultures identified an average of 11630 FFU of p6 virus/ml and twice as many, 23,200 FFU/ml, of the ORF3-null mutant (Table 3). Determination by real-time RT-PCR of the number of viral genomes in the medium was most revealing: there were indeed ~10-fold fewer viral genomes released into the medium for the ORF3-null mutant compared to the p6 virus (Table 3). Calculations of the number of viral genomes per FFU indicated that the mean specific infectivity of the ORF3-null mutant virus was ~25-fold higher than that of p6 virus itself. Therefore, the decrease in egress from cells due to a lack of ORF3 was more than offset by the increase in infectivity, thus enabling the null mutant to spread through the culture as efficiently as the parent p6 virus.
A nonhomologous recombination event involving the HEV genome and a human RNA molecule was ultimately responsible for the ability of the Kernow strain of HEV to flourish in cell culture. Such a dramatic change in virus phenotype following virus-host RNA recombination is rare, but a few cases have been reported previously. Thus, a poliovirus lethal mutation was pseudoreverted by introduction of 15 host nucleotides into a mutated cleavage site during replication in cell culture (2). Similarly, introduction, by recombination, of 54 host nucleotides into the hemagglutinin cleavage site of an apathogenic influenza virus produced a virus variant with increased pathogenicity (14). Even more strikingly, a rare recombination event which inserted 228 or more nucleotides of host ubiquitin gene sequence into a noncytopathogenic variant of bovine viral diarrhea virus rendered the virus cytopathogenic (19).
The infectious genotype 3 cDNA clone we constructed provides an additional tool for HEV research. Since the liver is the target organ for this virus, the ability to transfect or infect human liver (HepG2/C3A) cells and to produce large quantities of viable virus may provide a more authentic model system in which to revisit numerous, well-executed studies that produced intriguing data but were limited by their reliance on overexpression of single viral proteins out of context. In addition, the ability of p6 virus to infect both human and swine cells may prove useful for identifying parameters that restrict the host range of genotype 1 and 2 strains to humans and nonhuman primates. The luciferase replicon we developed should be especially useful for some studies since it permits convenient sequential sampling and is exquisitely sensitive: since the luciferase gene is located on the subgenomic mRNA, luciferase production can act as an indirect indicator of subgenomic RNA synthesis and stability. This new model system has already provided the first evidence that the previously uncharacterized X gene region has a function in viral replication since three mutations in it contributed substantially to establishment of the infected state following transfection (Fig. 3 and and55).
HEV is not noted for recombination and intergenotypic recombination has been reported only rarely (31). In retrospect, this might reflect the different transmission pathways and localized geographic distribution of the four human genotypes resulting in a low number of coinfections with two or more readily distinguishable genomes; intragenotypic recombination might not be noticed unless specifically searched for. However, our discovery of three different human sequences embedded in HEV genomes from the only two patients examined suggests that HEV may undergo recombination more frequently than realized. For instance, a genotype 3 virus from France was reported to have an insert of unidentified origin of ~90 nt in the HVR (15). Additional studies are required to determine whether insertion of ribosomal protein genes occurred by chance or reflected some unknown aspect of HEV replication.
The discovery of the human S17 gene sequence embedded in the HEV genome (28) was especially surprising since it indicated (i) that the virus genome had recombined with host RNA and (ii) that this event had apparently imparted properties that resulted in selection of this extremely minor quasispecies virus in cell culture. This scenario was subsequently repeated with a genotype 3 strain from another chronically infected patient (21), suggesting that illegitimate recombination by HEV is not necessarily a rare event. In the present study, we demonstrated that this recombinant virus emerged as soon as the first passage in cell culture (Table 1): its dominance in all passages thereafter strongly suggested that the insert played a critical role in cell culture adaptation. Mutagenesis studies of the infectious cDNA clone demonstrated unequivocally that the insert was a major factor in enabling efficient virus propagation in cell culture and therefore was an adaptive mutation. However, since the stepwise cloning strategy demonstrated that mutations other than the S17 insert also contributed to adaptation (Fig. 2), it was surprising to find an almost total elimination of enhancement of transfection by point mutations upon removal of the S17 insert from the final construct (Fig. 4). One possible explanation is that the inserted S17 sequence enhanced the stability/translatability of the RNA or aided the folding/processing/stability of ORF1 protein. The question of proteolytic processing has not yet been resolved for HEV: however, since introduction of synonymous mutations into 24 to 32% of the nucleotide positions in the S17 insert did not appreciably affect the level of transfection (Fig. 6), it seems unlikely that the viral RNA is the critical factor, but rather suggests that the effect is at the protein level. Deletion experiments have shown that decreasing the size of the standard HVR could decrease the virulence of HEV or reduce its replication in cell culture (27). Our database analysis of the 50% truncations of the S17 insert demonstrated that the size of the insert, and hence of the HVR, matters, but the experiments substituting GFP, GTPase, or S19 gene fragments (7A, B, and D) suggested that the amino acid composition of both the insert and the genomic background must also contribute to enhancement. This conclusion is in agreement with data showing decreased replication in vitro when the HVR of a genotype 1 and a genotype 3 strain were swapped (26).
Very few HEV strains have been successfully propagated in cell culture, and the selection of the S17 recombinant in cell culture raised the hope that insertion of this sequence into other genotypes and strains might increase the success rate. Certainly, the ability of Sar55-S17 to transfect human or hamster cells was promising. Unfortunately, in toto, the data suggested that insertion of foreign sequences into the HVR of an HEV genome will not have a predictable impact and that subjection to long-term selective pressure in culture may be the only way to obtain culturable strains. On the other hand, the fact that the S17 sequence increased the ability of Sar55 genomes to replicate in cells from such an unlikely species as the hamster leads one to speculate that new syndromes, such as neurological disorders recently associated with HEV infections, may reflect the ability to infect new cell types because of changes in the HVR. Indeed, Kamar et al. have reported that genomes in the cerebrospinal fluid of a chronically infected patient had sequence differences from those that circulated in the serum (11). Certainly, this possibility merits further exploration.
Transfection and infection experiments with human HepG2/C3A and swine LLC-PK1 cells demonstrated that the p6 virus clone retained the ability of the fecal virus quasispecies to cross species boundaries and displayed a slight preference for swine cells. In contrast, the titer of the fecal inoculum was previously reported to be up to 13-fold higher on swine cells compared to human cells (28), which suggests that there might be other members of the fecal quasispecies that either had mutations favorable for infection of swine cells or detrimental for infection of human cells (Fig. 8). It is not known whether receptors or other factors determine host range. Between the p6 cloned virus and the consensus sequence of viruses in the feces, there are 4 aa differences in the capsid protein which might affect receptor interactions. Two of the four mutations were also present in the p1 virus clone which represented the first selection step for HepG2/C3A cells, so it will be interesting to determine whether reversion of any of these mutations to the consensus sequence in the feces will increase the relative titer on swine cells.
Although both p6 virus and ORF3-null virus eventually spread and infected the majority of HepG2/C3A cells in a culture, they did so relatively slowly, and the percentage of infected cells did not begin to increase until after day 7 (Fig. 10). In contrast, luciferase expression was detected in the culture medium as soon as day 1 posttransfection (2,163 U) and had jumped 38-fold by day 2 (Fig. 5A). Since the luciferase is translated from the subgenomic mRNA, viral negative-strand and subgenomic RNA synthesis must have been greatest between days 0 and 2 in this experiment, suggesting that synthesis of viral RNA and/or proteins is probably not rate limiting but rather that assembly, maturation, and/or excretion are responsible for the relatively slow production of infectious HEV virions. It is worth noting that since the luciferase construct lacks a capsid gene, it cannot spread, so the data in Fig. 5B suggested that translation of p6 subgenomic mRNA continued at peak rates through day 7 or 8 before declining.
Perhaps the most confounding result was the discovery that a virus unable to make ORF3 protein spread throughout the culture as efficiently as one synthesizing ORF3 protein. This result poses more questions than answers. The observed difference in specific infectivities provides an explanation of why it happened, but the question of why the specific infectivities differed remains. Okamoto has proposed that HEV egress from PLC/PRF/5 cells depends on ORF3 protein interaction with cellular protein Tsg101 (20). So, is the null mutant exiting the HepG2/C3A cells by a different pathway than p6 uses? Is a different pathway used in different cell lines? Are any of the available culture systems truly reliable models for infection in vivo? It now may be possible to address some of these questions.
This research was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Allergy and Infectious Diseases.
Published ahead of print 7 March 2012