Potential conflict of interest: Nothing to report.
The existence of hepatitis C virus (HCV) proteins encoded by alternate reading frames overlapping the core-encoding region has been suggested. Several mechanisms of production have been postulated, and the functions of these proteins in the HCV life cycle remain unknown. We analyzed cases of seroconversion to an alternate reading frame protein in a group of 17 patients infected by one of the two HCV genotype 1b strains during an outbreak in a hemodialysis unit. Three patients seroconverted, and antibodies were transiently detected in another patient. Three of these patients were infected by one of the two HCV strains, whereas the strain infecting the remaining patient could not be identified. Quasispecies sequence analysis of the core-coding region showed no differences in the core or +1 reading frame sequences that could explain alternate reading frame protein seroconversion in some but not all of the patients infected by one of the HCV strains, and no such difference was found between the two strains. Because differences in the structure of RNA elements could play a role in frameshift events, we conducted a predictive analysis of RNA folding. No difference was found between the patients who did and did not seroconvert to alternate reading frame protein. Conclusion: Our findings prove that alternate reading frame proteins can be produced during acute HCV infection. However, seroconversion does not occur in all patients for unknown reasons. Alternate reading frame protein could be generated by minority quasispecies variants or variants that occur transiently. (HEPATOLOGY 2009.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Hepatitis C virus (HCV), a member of the Flaviviridae family, has a single-stranded positive RNA genome of approximately 9,600 nucleotides. The genome is composed of a 5′ noncoding region, a long open reading frame (ORF) encoding the precursor polyprotein of about 3,000 amino acids, and a 3′ noncoding region. The ORF encodes 10 proteins that are generated through processing of the precursor polyprotein by cellular and viral proteases. HCV proteins comprise structural proteins (core protein and the two envelope glycoproteins E1 and E2) and nonstructural proteins (NS2, NS3, NS4A, NS4B, NS5A, and NS5B) that have various functions in the HCV lifecycle.1, 2
In 2001, Walewski et al.3 used computerized sequence analysis to map dual-use regions in HCV coding sequences. The results strongly suggested the existence of a new HCV antigen encoded, at least in part, by an alternate, +1 reading frame overlapping the core encoding region. This was further supported by the observation of a single stop codon in the +1 reading frame of reference HCV strains at positions that depended on the HCV genotype.4 The detection of specific antibodies targeting this protein in patients with chronic hepatitis C via both enzyme immunoassay and western blotting strongly suggests that alternate reading frame proteins (ARFPs) are produced during HCV infection.3, 5
Since the first report suggesting translation of an ARFP,5 several mechanisms of ARFP production have been postulated.6 The first is a +1 programmed ribosomal frameshift promoted by a cluster of 10 adenines at codons 8–11 of the HCV genotype 1a ORF, leading to the formation of a 16- to 17-kDa protein (ARFP/F) including the first 10 amino acids of the core protein.3, 5 The adenosine-rich sequence could also induce a −1 programmed ribosomal frameshift.7 Another potential mechanism in HCV genotype 1b infection is a double-frameshift event. The first (+1) frameshift would occur at codon 42, located in stem loop VI of the HCV genome; the second (−1) would occur at codon 144 in the +1 ORF, close to the terminal stem loops.8 Rephasing back to translation in the HCV ORF leads to production of ARFP–double frameshift core, whereas interruption of translation after the stop codon leads to the production of ARFP-1b.8 Internal initiation near codons 85–88 or at codon 26 in the core-encoding sequence has also been suggested to generate ARFPs in the +1 reading frame (ARFP-86 and ARFP-26, respectively).9, 10
ARFP/F can be expressed in vitro. It is present as a highly basic protein in the cytoplasm of transfected cultured cells, sometimes with a perinuclear distribution.11 ARFP/F is unstable, however, and is degraded by the proteasome complex rapidly after its synthesis (half-life, ≈10 minutes).5, 11 There is some evidence that ARFPs are produced during HCV infection. Indeed, anti-ARFP antibodies have been detected with specific assays in up to 40% of patients with chronic HCV infection, and ARFP-specific T and B cell–mediated responses have been documented.5, 8, 12–14 However, the frequency and timing of ARFP production during infection, as well as the role of ARFPs in the HCV lifecycle and pathogenicity, are unknown. ARFP does not appear to play a role in viral replication, because subgenomic replicons lacking the full structural region of the genome successfully replicate in vitro.15 The basic nature and cytoplasmic localization of ARFP are compatible with an interaction with HCV RNA. The possible role of ARFP in the establishment of acute and chronic HCV infection is unknown.
We analyzed cases of seroconversion to ARFP in a group of patients infected by one of two HCV genotype 1b strains during an outbreak in a hemodialysis unit. Our findings prove that ARFP can be produced during acute HCV infection, but not all patients or even all patients infected by a given strain are able to mount a humoral response against it if produced.
Six groups of patients were studied (Table 1). Group I consisted of a cohort of 17 patients (6 males and 11 females; mean age, 64 years) who developed acute hepatitis C during an outbreak in a Greek hemodialysis unit.16 The route of transmission was not identified. The patients were infected with one of two distinct HCV genotype 1b strains, nine patients with strain A and seven with strain B (one patient's strain could not be genotyped).16 The patients were monitored without treatment during the first 6 months, and blood samples were taken frequently (on average, 15 ± 4 samples per patient) and stored at −80°C. Alanine aminotransferase (ALT) and HCV RNA levels were measured in all samples. The seroconversion profiles were characterized by using a semiquantitative line immunoassay (Inno-LIA HCV IV prototype assay, Innogenetics, Ghent, Belgium) that semiquantifies antibodies against eight HCV epitopes present in the core, E1, E2, NS3, NS4, and NS5 protein sequences.
Table 1. Study Groups
Acute HCV infection with one of the two HCV genotype 1b strains (strain A and strain B) during an HCV outbreak in a hemodialysis unit
HCV-negative acute hepatitis A (n = 9) or B (n = 15)
Chronic HCV genotype 1 infection
Patients with no markers of common chronic viral infections
Patients infected with human immunodeficiency virus
Patients with chronic HBV infection (hepatitis B surface antigen–positive)
Group II consisted of 24 patients with acute viral hepatitis unrelated to HCV: nine patients had acute hepatitis A characterized by the presence of anti–hepatitis A virus immunoglobulin M; 15 patients had acute hepatitis B characterized by a seroconversion to hepatitis B surface antigen (including seven patients who presented with symptoms of acute hepatitis and eight regular blood donors who seroconverted between two donations). Group III consisted of a cohort of 216 patients chronically infected with HCV genotype 1 strains and managed in the Hepatology Department of Henri Mondor Hospital, France.
Three control groups of subjects with no markers of HCV infection were used to assess the specificity of our anti-ARFP enzyme-linked immunosorbent assay (ELISA). Group IV consisted of 187 patients managed in medical departments of Henri Mondor Hospital who had no serological markers of common viral infections. Group V consisted of 122 patients infected with human immunodeficiency virus. Group VI consisted of 93 patients infected with hepatitis B virus.
We developed and characterized a sensitive ELISA method for anti-ARFP (43–141) antibody detection. It was used to detect anti-ARFP antibodies in all available serum samples, including serial samples from group I patients and single samples from group II and III patients. The nucleotide and deduced amino acid sequences of the HCV core region were determined in all group I patients. Multiple clones were generated to characterize the quasispecies distribution of the analyzed region. The ARFP sequences were deduced from the core protein-coding sequences and compared between the patients who did and did not seroconvert to ARFP. The core protein-coding sequence contains highly conserved structural RNA elements, including two consecutive stem loops (V and VI) and two consecutive terminal stem loops.17 In order to determine whether the production of anti-ARFP antibodies during acute HCV infection in some patients but not in others was related to structural differences in the core protein-coding sequence of the corresponding HCV strains, we examined sequence variations and predicted their impact on the secondary structure of the core protein-coding region. The prevalence of anti-ARFP antibodies was also measured via ELISA in the 24 group II patients with acute hepatitis A or B in order to assess the specificity of anti-ARFP antibody detection for HCV-related acute hepatitis and in the 216 group III patients with chronic HCV genotype 1 infection in order to compare it with the incidence of anti-ARFP seroconversion during acute infection. The specificity of anti-ARFP antibody detection was assessed by testing the patients in groups IV, V, and VI.
Anti-ARFP (43–141) Antibody ELISA.
For ELISA detection of anti-ARFP (43-141) antibodies, we used the synthetic peptide G97A, composed of 99 amino acids encoded by codons 43–141 and derived from the +1 ORF of the core protein sequence of HCV genotype 1b isolate HCV-JA.12 The wells of microtiter plates were coated with 200 ng per well of purified G97A peptide (10 μg/mL) via overnight incubation at room temperature in sodium carbonate buffer (50 mmol [pH 9.5]). Anti-ARFP (43–141) monoclonal antibodies raised in mice immunized with the same peptide were used as a positive control. The wells were washed three times with PBS-Tween and incubated with 200 μL of blocking buffer (200 mmol/L Tris, 200 mmol/L maleic acid, 0.9 g/L sodium azide, 5 g/L bovine serumalbumin, 1% goat serum [pH 6.2]) for 2 hours at 37°C. After washing, coated plates were incubated with 200 μL of human serum diluted to 1:10 with Diasorin sample diluent (Saluggia, Italy) at 37°C for 1 hour. After washing, the wells were incubated for 1 hour at 37°C with 200 μL of a peroxidase-conjugated anti-human immunoglobulin G (Diasorin) or anti-mouse immunoglobulin G antibody for the clinical specimens or positive controls, respectively. After washing, the reaction was developed by adding a tetramethylbenzidine/hydrogen peroxide mix and incubating the plate at 37°C for 40 minutes in the dark. After adding 100 μL of 0.4 N sulfuric acid, absorbance was read at 450/620 nm. The cutoff value was determined as the mean optical density value obtained with three sera from HCV-uninfected blood donors plus three standard deviations plus 0.1. A gray zone of ±20% was defined around the cutoff value.
Core and ARFP Quasispecies Sequence Analysis.
In order to generate core and ARFP quasispecies sequences, viral RNA was extracted from 200 μL of serum with the High Pure kit (Roche Applied Sciences, Indianapolis, IN) and recovered in 50 μL of RNase-free water. After 15 minutes of denaturation at 75°C, 5 μL of viral RNA was reverse-transcribed at 53°C for 1 hour using 10 pmol of antisense primer CORE-AS1 (5′-CCGAACACAGGGCACGCACC-3′) and 200 U of Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA). Samples were heated for 15 minutes at 70°C to inactivate the reaction. The core protein-coding region was amplified by means of nested polymerase chain reaction (PCR) with outer primers CORE-S1 (sense, 5′-CTAGCCGAGTAGTGTTGGG-3′) and CORE-AS1 (antisense) and inner primers CORE-S2 (sense, 5′-CTGATAGGGTGCTTGCGAG-3′) and CORE-AS2 (antisense, 5′-TCCGCTGCCTCATACACAATGC-3′). The two PCR rounds were performed with the Advantage 2 Polymerase Mix (Becton-Dickinson, Palo Alto, California). The first round consisted of 5 minutes of denaturation at 95°C, followed by 40 PCR cycles with denaturation at 95°C for 30 seconds, annealing at 55°C for 30 seconds, and extension at 68°C for 45 seconds, and a final extension step at 68°C for 5 minutes. Two microliters of the first-round PCR product was used as a template for the second round, which consisted of 5 minutes of denaturation at 95°C, followed by 35 PCR cycles with denaturation at 95°C for 30 seconds, annealing at 60°C for 30 seconds, and extension at 68°C for 45 seconds, and a final extension step at 68°C for 5 minutes.
PCR amplicons were purified before ligation into the pCR4.0 plasmid (Invitrogen). On average, 24 clones per sample were sequenced. Nucleotide and amino acid sequences were aligned with the CLUSTAL W program.18 ARFP sequences were deduced from core protein sequence alignments. Sequence data from this study have been deposited in the EMBL Nucleotide Sequence Database under accession numbers AM285684 to AM286184 and AM397933 to AM397949.
Predictive Analysis of Conserved Secondary Structures in Core/ARFP Coding Sequences.
All analyses were performed using the European HCV Database Web site facilities (http://euhcvdb.ibcp.fr),19 the Clustal W program,18 and the Vienna RNA secondary structure server (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi).20 We predicted the impact of core protein-coding sequence variations on the RNA secondary structure by using the RNAfold program. Two HCV genotype 1b reference genome sequences from the European HCV database were used (numbered D90208 and M58335). The regions selected to generate stem loops V and VI and the terminal stem loops spanned nucleotides 43 to 169 and 438 to 518, respectively, using the A of the AUG initiation codon as position 1. Only nonredundant sequences were analyzed in each patient. In order to eliminate misincorporation errors at the PCR stage, sequences harboring a nucleotide substitution at a position that was fully conserved in all the other sequences were excluded.
Performance of the Anti-ARFP (43–141) ELISA.
We evaluated the analytical sensitivity of our anti-ARFP (43–141) ELISA by testing several dilutions of an anti-ARFP (43–141) monoclonal antibody produced in mice immunized with the G97A peptide (initial concentration, 2.8 mg/mL; range of dilutions, 1/1,000 to 1/128,000). ELISA detected anti-ARFP (43–141) at a concentration of 2.2 × 10−2 ng/μL. By comparison, commercial anti-HCV assays can detect anti-HCV antibody concentrations of about 10−3 ng/μL.
The specificity of our anti-ARFP (43–141) ELISA was 99.4% (95% confidence interval 99.3%–99.5%) when tested on samples from group IV patients, who had no markers of common viral infections (one of the 187 patients fell in the gray zone) (Table 2). None of the 122 human immunodeficiency virus–infected patients in group V had anti-ARFP (43–141) antibodies. In contrast, one (1.1%) of the 93 hepatitis B surface antigen–positive patients in group VI was positive and seven (7.5%) fell in the gray zone (Table 2), suggesting cross-reactivity between the two viruses, as reported with another ELISA assay using a similar peptide.13
Table 2. Specificity of Anti-ARFP (43–141) Antibody Detection via ELISA in Patients with No Markers of Viral Infection (Group IV) and in Patients Infected with Human Immunodeficiency Virus (Group V) or HBV (Group VI)
Anti-ARF (43–141) Negative Response, n (%)
Gray Zone, n (%)
Anti-ARF (43–141) Positive Response, n (%)
Group IV (no viral infection)
Group V (human immunodeficiency virus infection)
Group VI (HBV infection)
The intra- and inter-assay coefficients of variation were determined by testing 15 sera five times each in five different runs. They ranged from 2.5% to 8.1%, suggesting that our ELISA had excellent precision and reproducibility.
Seroconversion to ARFP (43–141) in the 17 Group I Patients with Acute HCV Infection.
Seroconversion to ARFP (43–141) was studied in the 17 group I patients who were infected by one of two distinct HCV genotype 1b strains during an outbreak in a hemodialysis unit.16 The seroconversion profiles were compared for ARFP (43–141) and for structural and nonstructural HCV proteins in 82 serum samples from the 17 patients, who were sampled roughly every month during the 6-month follow-up period.
Figure 1 shows the dynamics of anti-ARFP (43–141) antibody titers in the 17 patients. Three patients seroconverted to ARFP (43–141) during the study period (patients 1, 6, and 14), and anti-ARFP (43–141) antibodies were transiently detected in another patient (patient 3). Patients 1 and 6 were both infected with HCV strain A, whereas the strain infecting patient 14 could not be identified. Patient 3 was also infected by HCV strain A. Figure 2 shows anti-ARFP (43–141) antibody seroconversion dynamics relative to ALT and HCV RNA dynamics in these three patients. Seroconversion occurred 4 and 8 weeks after the ALT peak in patients 1 and 6, respectively, and at the time of the ALT peak in patient 14.
Figure 3 shows the profiles of seroconversion to ARFP (43–141) and to eight structural or nonstructural HCV epitopes of the core, E1, E2, NS3, NS4, and NS5 proteins. Fifteen patients seroconverted to the core and nonstructural proteins, while patients 13 and 16 had incomplete seroconversion profiles. Thus, the failure to detect anti-ARFP antibodies in 14 patients was not related to an inability to mount HCV-specific humoral responses.
Analysis of the Nucleotide Sequence of the Core Coding Region in the 17 Group I Patients with Acute HCV Infection.
In order to assess whether seroconversion to ARFP (43–141) was associated with specific nucleotide sequence motifs in the core-coding region that could explain the occurrence of frameshift events, we characterized the core-coding region quasispecies distribution in the 17 patients with acute HCV infection. Twenty-four clones per sample were generated, and their sequence was determined at baseline (earliest available sample) in all patients except two in whom PCR amplification of the core protein-coding region was not possible (one patient infected with strain B who did not seroconvert to ARFP, and one patient infected by an unidentified strain who seroconverted to ARFP). Because two patients infected with HCV strain A infection seroconverted to ARFP (43–141) during follow-up, quasispecies sequence analysis was repeated on the latest available sample in all the patients infected by HCV strain A (except for one patient who did not seroconvert to ARFP and in whom PCR amplification of the target region was not possible). A total of 504 full-length core protein-coding region sequences were analyzed.
The G97A peptide used to detect anti-ARFP (43–141) antibodies in our ELISA showed a high percentage of homology with the putative ARFP sequences deduced from the core protein-coding sequence in the patients, ranging from 88.5% to 91.0% (99 amino acids). In patients infected by HCV strain A, only patient 1's major quasispecies variant harbored a single nucleotide variation (C to U at position 93) relative to the strain A consensus sequence. The major variant of all strain B–infected patients had the same sequence as the consensus strain B sequence.
Several mechanisms have been forwarded to explain the production of ARFP proteins, including +1 ribosomal frameshifts at various positions of the ORF (codons 8–11, 42, and 144), and internal initiations of translation at codons 26 or 86–88 in the +1 reading frame. All of these positions were carefully examined in the viral sequences of the patients with and without anti-ARFP (43–141) antibody seroconversion (Fig. 4). First, no AGA to AAA substitution at codon 9 and no ACC to AAC substitution at codon 11 that would have created a polyA stretch5, 21 was observed in any variant from any patient, regardless of ARFP (43–141) seroconversion (Fig. 4A). Second, the regions surrounding codons 42 and 144, which have been reported to harbor a double frameshift event,8 were highly conserved and did not differ between the patients who seronconverted to ARFP (43–141) and those who did not (Figs. 4B,C). Third, codon 26 in the +1 reading frame, which has been described as a potential site for internal initiation of translation of ARFP proteins,9 had GTG and GCG sequences in 99.2% and 0.8% of analyzed sequences, respectively (Fig. 4D). The sequence did not therefore differ between the patients with and without ARFP (43–141) seroconversion, nor did the sequence spanning codons 28–35, which has been suggested to contain a signal that enhances the selection of codon 26 for initiation of translation (Fig. 4D). Finally, strong conservation was also observed in the region spanning codons 81–88, a suggested region of translation initiation in the +1 reading frame,10 and particularly the ATG sequence of codons 86 and 88. A few variations were observed in minor quasispecies variants that did not differ according to ARFP (43–141) seroconversion status.
Overall, no differences in the core or +1 reading frame sequences were found that could explain ARFP (43–141) seroconversion in two of the patients infected with HCV strain A, and no such difference was found between strains A and B. The studied regions were highly conserved, and no sequence known to be potentially associated with a +1 frameshift was found in any variant from any patient.
Predictive Analysis of Conserved Secondary Structures in the Core Protein Coding Sequence.
Because differences in the structure of RNA elements could play a role in frameshift events and may not be readily identified via sequence analysis of the core protein coding region, we conducted a predictive analysis of RNA folding by using RNAfold software. The regions spanning nucleotides 43–169 and 438–518 were used to predict the structures of stem loops V and VI and of the terminal stem loops, respectively. Only nonredundant sequences from each patient were included in the analysis, and sequences harboring one nucleotide substitution at a fully conserved position among all the other sequences were removed in order to exclude potential PCR errors. A final set of 76 sequences spanning stem loops V and VI (53 and 23 sequences from patients infected by strains A and B, respectively) and 69 sequences spanning the terminal loops (51 and 18 sequences from patients infected by strains A and B, respectively) was analyzed in comparison with two reference HCV genotype 1b strains.
We observed strong conservation of predicted structural elements in the core coding sequence, with only a minority of sequences generating different RNA folding predictions (5% for stem loops V and VI, 2.2% for terminal stem loops). Figure 5 shows the predicted secondary structures of stem loops V and VI. Four folding predictions were generated, with the strain A consensus sequence, patient 1's major quasispecies sequence, the strain B consensus sequence, and the reference sequence M58335. Stem loop V successively includes a long 10-bp stem followed by a conserved bulge, a 4-bp stem and the 5-nucleotide loop. This structure was conserved in the HCV strain A consensus sequence, whereas an additional bulge at the beginning of the long stem was predicted in the HCV strain B consensus sequence (Fig. 5). Stem loop VI successively includes a 7-bp stem followed by a conserved bulge, a 7-bp stem, two conserved excluded adenines, a variable 10- to 13-bp stem, another conserved bulge, a 4-bp stem, and the final loop (6 nucleotides) (Fig. 5).
The strain A consensus sequence structure differed from that of the reference strains and strain B by a variation in the first bulge of stem loop VI (Fig. 5). However, this difference was not found in the major quasispecies variant from patient 1, as a C to U substitution at position 93 restored the reference secondary structure (Fig. 5). Thus, patients 1 and 6, both of whom seroconverted to ARFP (43–141), had different predicted stem loops VI, whereas stem loop VI was identical in patient 6 and in the patients infected by strain A who did not seroconvert to ARFP (43–141). Only one substitution distinguished patients 1 and 6 from the other patients infected by strain A, namely an A to G substitution at position 59, which altered the conserved bulge in stem loop V and added one base pair to the stem (Fig. 5). Other point substitutions were found in patient 1 and patient 6 (Fig. 5).
The terminal stem loops were also highly conserved among the different patients as a result of strong conservatory constraints on the sequence of the core protein. No difference was found between the patients infected by strain A who did and did not seroconvert to ARFP (43–141). Rare substitutions sometimes affected the number of predicted stem loops (data not shown), but they bore no relationship to ARFP (43–141) seroconversion.
Prevalence of Anti-ARFP (43–141) Antibodies in Patients with Acute Hepatitis A or B (Group II).
None of the nine patients with acute hepatitis A had detectable anti-ARFP (43–141) antibodies. One blood donor with a hepatitis B surface antigen seroconversion (out of 15 patients with acute hepatitis B) was found to be positive on anti-ARFP (43–141) ELISA but was negative on repeat testing, confirming possible cross-reactivity between the two viruses.
Prevalence of Anti-ARFP (43–141) Antibodies in Patients with Chronic HCV Genotype 1 Infection (Group III).
Twenty-six (12.3%) of the 216 patients with chronic HCV genotype 1 infection (group II) had detectable anti-ARFP (43–141) antibodies in our ELISA.
Our observation of seroconversion to ARFP (43–141) in three patients and transient detection in one patient with dated acute HCV infection proves that ARFPs are produced during HCV infection, that they are expressed early in the acute phase of infection, and that their production can trigger a specific humoral response. In two of the three patients who seroconverted, anti-ARFP (43–141) antibodies emerged several weeks after antibodies against the main ORF proteins, both structural and nonstructural. This may point to late ARFP production during infection or may simply reflect the slightly lower sensitivity of our anti-ARFP (43–141) antibody ELISA relative to commercial anti-HCV assays. In the third patient, anti-ARFP (43–141) antibodies were detected 1 week before anticore antibodies, and no other antibodies emerged during the short follow-up period of a few months. Our anti-ARFP (43–141) antibody assay was highly specific, virtually ruling out false-positive results. Importantly, anti-ARFP (43–141) antibodies were not found in acute viral hepatitis of other causes.
Our study population was exceptional, because the patients with acute hepatitis C were infected by one of only two HCV genotype 1b strains. This situation is ideal for studying the respective roles of viral and host factors. Indeed, two patients infected by strain A, but none of the patients infected by strain B, developed anti-ARFP (43–141) antibodies, suggesting that some HCV strains but not others express ARFP. However, the number of patients was relatively small, and a third patient with an untypeable strain also developed anti-ARFP (43–141) antibodies. More interesting is the fact that only two of the nine patients infected by the same HCV strain A experienced anti-ARFP (43–141) antibody seroconversion. In the patients who did not have detectable anti-ARFP (43–141) antibody seroconversion, it was not possible to know if (1) ARFP was not produced, (2) ARFP was produced but the patients failed to mount a detectable anti-ARFP (43–141) antibody response, or (3) the antibody response was not detectable with our assay. In this respect, it is of interest that 10 out of 17 patients had no evidence of anti-NS5B antibodies (Fig. 3), whereas it is obvious that they all produced the NS5B RNA-dependent RNA polymerase, the viral enzyme needed for replication. We were not able to study the possible role of host factors in this variability, or the underlying mechanisms. It also remains to be determined whether or not anti-ARFP (43–141) antibody seroconversion can occur later in patients who develop chronic HCV infection. This could not be studied, because our patients were treated when the acute infection did not resolve spontaneously after 6 months of follow-up.
The prevalence of anti-ARFP (43–141) antibodies was 12.6% in patients with chronic HCV genotype 1 infection, a rate lower than in a study by Komurian-Pradel et al.13 The difference could be due to a cohort effect (only patients infected with HCV genotype 1a were tested in the Komurian-Pradel study) or to the use of a longer antigen (160 amino acids instead of 99 in our assay) and/or a less stringent cutoff in the Komurian-Pradel study. Contrary to the acutely infected patients in our study, our chronically infected patients were unrelated and were infected by different HCV genotype 1 strains. Two factors may have played a role in the production of anti-ARFP (43–141) antibodies in these patients, namely the intrinsic capacity of the infecting viral strain to produce ARFP and host factors regulating ARFP production and/or the humoral response to ARFP.
In an attempt to identify viral characteristics potentially explaining why only some patients seroconverted to ARFP (43–141), we analyzed the sequence of the core protein-coding region quasispecies. The aim was to detect sequences known to be associated with ARFP production in major or minor quasispecies populations. We focused on sites previously implicated in ARFP production, such as the adenine cluster at codons 8–11, the frameshift codons 42 and 144, and the internal initiation codons 26 and 85–88.3, 5, 7–10 No sequences potentially involved in ARFP production were found, and no difference was seen between strains A and B or between the HCV strains infecting patients who seroconverted and those who did not. Likewise, predictive structural analysis of the core protein coding sequence showed no differences between the patients who seroconverted and those who did not. These results suggest that ARFPs are probably encoded by minority variants that cannot be detected by clonal sequence analysis of serum samples. Alternatively, the expression of ARFP-encoding variants may be transient, leading to brief ARFP production that is sufficient to trigger a humoral response in some patients but not in others. The lack of seroconversion could be due to a lack of ARFP production, excessively brief ARFP expression, an expression level too low to elicit specific antibodies, or an inability of our assay to detect low levels of these antibodies. The observed conservation of predicted ARFP sequences in our patients rules out a failure of our assay to detect antibodies directed at different antigenic sequences.
In conclusion, our observation of anti-ARFP (43–141) antibody production in some hemodialysis patients acutely infected with one of the two HCV genotype 1b strains during a nosocomial outbreak is the strongest argument to date that ARFPs are produced during HCV infection. Our findings also indicate that ARFP production occurs early during infection. ARFP could be generated by minority quasispecies variants or variants that occur transiently at the beginning of infection. Our findings shed no light on the possible function of ARFP. The patients who did and did not seroconvert did not differ in their clinical outcomes, severity of infection, or seroconversion profiles. The role of ARFP in the replicative cycle and pathogenicity of HCV is still unknown.
We are grateful to Erwin Sablon, Geert Maertens, and Filip de Kayser (Innogenetics) for preparing the INNO-LIA HCV IV prototype assay.