INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Hepatitis caused by hepatitis C virus (HCV) has become a major emerging infectious disease problem, with an estimated 170 million people infected worldwide (8). In industrialized countries, HCV accounts for 20% of acute hepatitis cases, 70% of chronic hepatitis cases, 40% of end-stage cirrhosis cases, and 60% of hepatocellular carcinoma cases (8). It has become one of the most common reasons for liver transplants (16).

HCV is a positive-sense single-stranded RNA virus belonging to the family Flaviviridae. Clinical strains exhibit a great degree of genetic heterogeneity, and it is thought that such strain differences influence the clinical outcome after infection as well as transmission patterns among susceptible hosts (1, 32, 37, 40). Hence, genotypic characterization of HCV has become an important part of clinical management and public health control of hepatitis C. At the 2nd International Conference of HCV and Related Viruses, a consensus classification scheme was proposed for HCV. According to this system, HCV is classified into six major genotypes, which are further divided into subtypes (29). Genotypic differentiation of HCV has served as an important epidemiologic tool for study of the geographic distribution of HCV genotypes, their routes of transmission, and their association with certain risk groups. Genotypes 1, 2, and 3 are found throughout the world and constitute the major genotypes in Japan, Western and Eastern Europe, and North America (20, 38, 40). Genotypes 4, 5, and 6 appear to be more restricted in their distribution: genotype 4 has been found mostly in Central and Northern Africa and in the Middle East, type 5 has been reported from South Africa, and type 6 has been identified in Southeast Asia and Hong Kong (18, 22, 28, 40).

Following infection with HCV, patients exhibit differences in the likelihood of developing chronic infection, responding to interferon therapy, progressing to cirrhosis, and developing hepatocellular carcinoma. Although factors such as the duration of infection, gender, and alcohol use have been associated with severe outcomes of hepatitis C (40), whether these differences in clinical outcomes can be also attributed to HCV genotypic differences is not clear. Several studies have shown that genotype 1, especially subtype 1b, is associated with a less favorable outcome after interferon treatment (7, 10, 15, 21, 22, 26, 35, 38, 40). There is also evidence that subtype 1b is associated with an increased risk for the development of severe sequelae such as liver cirrhosis and hepatocellular carcinoma (19, 37). However, other studies do not show this association (2, 13, 32). The uncertainty about these associations may be resolved by studies that specifically examine the role of HCV subtype 1b in clinical outcomes in additional sites around the world.

Several methods for the genotyping of HCV exist. The reference method is direct sequencing of products amplified from clinical material. Several PCR-based methods for the differentiation of strains by comparison of differences in the sizes of the products amplified by type-specific primers have been reported (14, 23, 24, 27, 36). Another PCR-based method relies on differences in the agarose gel electrophoretic band patterns of a PCR-amplified product digested with restriction endonucleases (6). A commercial DNA hybridization method called the line probe assay relies on genotype-specific probes based on 5' noncoding (NC) sequences to hybridize to a product amplified from a clinical sample (34). Most of these methods are labor-intensive, costly, and confined to research or reference laboratories, thus limiting the studies needed to address the types of questions related to the clinical significance of HCV genotypes posed above. Simpler and more accessible tests are needed. This is especially true in countries such as the Czech Republic, where HCV poses a significant public health problem and there is a limited ability to diagnose infection at the genotypic level.

In this report, we describe a simple PCR-based test that rapidly detects HCV in clinical specimens and that simultaneously distinguishes HCV subtype 1b from the other genotypes. This strain-typing assay is based on a method called restriction site-specific PCR (RSS-PCR). This method takes advantage of nucleotide sequence polymorphisms that occur at restriction endonuclease recognition sites. The RSS-PCR technique is unique in that it generates a "fingerprint" pattern without the use of any endonucleases. The technique was first applied to the typing of dengue viruses (9, 17) and was later used to differentiate enteric Escherichia coli strains (12). For HCV, the technique was modified as a nested RSS-PCR because the organism cannot be cultured in vitro and levels of viremia vary greatly among patients. While this approach can be designed to differentiate one HCV genotype from the others, in this report we describe as one example a method that specifically differentiates HCV subtype 1b from the other genotypes. This technique should facilitate studies that examine the significance of infection with HCV subtype 1b on clinical outcome in any laboratory capable of performing PCR assays.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Serum samples. Serum samples from patients attending liver disease clinics in Prague, Czech Republic, were prospectively collected from October 1998 to January 2000. The entire collection consisted of serum samples obtained from 256 patients with viral hepatitis or hepatitis of unknown origin associated with abnormal liver function tests. All of the patients were citizens of the Czech Republic living in Prague. None of the patients was on antiviral treatment (interferon, ribavirin, amantadine, etc.) at the time of serum sample collection. All of the samples were screened by enzyme-linked immunosorbent assay (ELISA) for hepatitis A (hepatitis A virus immunoglobulin M assay [HAV Total Assay; Bio-Rad, SA, Paris, France]), hepatitis B (MONOLISA Ag HBs PLUS; Bio-Rad, SA), and hepatitis C (MONOLISA anti-HCV PLUS, version 2; Bio-Rad, SA). In the present study, we analyzed the samples that were positive by both the ELISA for HCV and reverse transcription (RT)-PCR for HCV (Amplicor PCR; Roche Molecular Systems, Inc., Pleasanton, Calif). In addition, we analyzed 15 serum samples from patients with a known diagnosis of hepatitis C from San Francisco, Calif. All of these U.S. samples had previously been tested by PCR (Amplicor PCR), and the viral titers had been determined. The HCV genotypes in these samples had been identified by a line probe assay (INNO-LiPA HCV II; Innogenetics, Inc., Alpharetta, Ga.) by a hepatitis laboratory. These U.S. serum samples served as positive controls. As negative controls, we tested clinical serum samples from Prague patients with liver disease which were negative by the HCV ELISA and RT-PCR tests. All of the samples were stored at 70°C until analyzed and were processed under similar conditions within a period of 6 weeks.

Primer design. The nested RSS-PCR strain-typing method described in this report is based on a procedure previously reported for dengue viruses (9, 17). The first step in the modified procedure involves the use of two external primers (primers Bukh-E1 and RSS-E2, Table 1), which were used to amplify a 661-base segment spanning nucleotide positions 285 in the 5' NC region and 376 in the core (C) region, based on the nucleotide numbering system used for prototype HCV genotype 1a strain reported by Choo et al. (5). Primer Bukh-E1 is identical to that previously reported by Bukh et al. (4). Five additional primers were used for the nested PCR to generate multiple band patterns with the 661-bp amplicon as a template. Two of these nested primers (called Bukh-1 and Bukh-2), identical to those used by Bukh et al. (4), were designed to specifically amplify a conserved segment in the 5' NC region to generate an amplicon of identical size (256 bp) in every reaction. The other three primers (primers RSS-I1, RSS-I2, and RSS-I4) were designed to specifically distinguish HCV genotype 1b from the other HCV genotypes. Their primer sequences were based on the sequence of prototype genotype 1b strain HCV-J, reported by Kato et al. (11). The nucleotide sequences indicated in Table 1 are those of the prototype genotype 1b strain (HCV-J) that correspond to the position numbers based on the prototype genotype 1a strain (5). In each case, we designed the primers to introduce a single nucleotide mismatch at the 3' end between the corresponding genotype 1a and 1b sequences. Two of these primers (primers RSS-I1 and RSS-I4) were based on nucleotide sequences within the 5' NC and C regions that contained restriction endonuclease recognition sites; RSS-I1 contains a BstUI site and RSS-I4 contains a MaeII site.

PCR amplification. Viral RNA was extracted from 100-µl aliquots of serum by the silica particle purification method previously described by Boom et al. (3). RNA pellets were resuspended in 60 µl of DNase- and RNase-free water (Gibco; BRL Products, Gaithersburg, Md.) containing 1 µl of RNase inhibitor (Recombinant Ribonuclease Inhibitor; Promega Corp., Madison, Wis.). The extracted RNA was immediately used for RT-PCR. The RT-PCR assay was performed with 5 µl of the extracted RNA added to 20 µl of an RT-PCR mixture consisting of 50 mM potassium chloride, 10 mM Tris-HCl (pH 8.5), 3.0 mM magnesium chloride, 0.01% gelatin, 200 µM concentrations of each of the four deoxynucleoside triphosphates, 10 mM dithiothreitol, 0.25 µM concentrations each of the primers Bukh-E1 and RSS-E2, 0.63 U of avian myeloblastosis virus reverse transcriptase (Promega), 10 U of RNAsin (Promega), and 0.63 U of Taq DNA polymerase (AmpliTaq; Perkin-Elmer Corp., Foster City, Calif.). RT was conducted at 42°C for 60 min, followed by 30 amplification cycles consisting of 94°C for 1 min, 55°C for 1 min, and 72°C for 2 min, with a final extension at 72°C for 5 min.

Validation of the nested RSS-PCR method. Serum samples obtained from patients in San Francisco who had hepatitis C and whose HCV genotypes were known (on the basis of the results of the line probe assay) were used as positive controls to generate PCR band patterns. Band patterns generated from the clinical serum samples from Prague patients were compared to the patterns produced from these control samples. In addition, we sequenced the 5' NC and NS5b regions of the viral genome from the test (Prague) and control (U.S.) clinical samples to further validate the nested RSS-PCR test results. We amplified and sequenced the product generated with primers Bukh-1 and Bukh-2 (Table 1), which contains the 5' NC region spanning nucleotide positions 276 to 50 (5). This segment includes a 176-bp sequence between nucleotide positions 244 and 69 used by Simmonds et al. (31) and Smith et al. (33) to genotype HCV. In addition, we verified the genotypes of some of the samples by sequencing a 222-bp amplified fragment within the NS5b region of the HCV genome, corresponding to positions 7975 to 8196 in the prototype 1a virus (numbered as in reference 5), as reported previously (30). The amplified DNA fragments were sequenced in both directions by dideoxynucleotide chain termination reactions with the ABI Prism 310 Genetic Analyzer (Perkin-Elmer). The sequences were compared to prototype HCV genotype sequences deposited in GenBank with DNASIS software (Hitachi, South San Francisco, Calif). The accession numbers of the prototype genotype sequences used to compare the 5' NC sequences were as follows: 1a, M62321; 1b, D90208; 2a, D00944; 2b, D01221; 2c, D10075; 3a, D14307; 3b, D11443; 3c, D16612; 4a, M84848; 4b, M84845; 4c, M84862; 4d, M84832; 4e, M84828; 4f, M84829; 5a, M84860; and 6a, M84827. Sequences of NS5b region from the U.S. and Prague clinical samples were validated by comparison to other NS5b sequences deposited under accession numbers L23435 to L23475 (30).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Serum samples. Of the serum samples collected from 256 patients with liver disease in Prague from October 1998 to January 2000, 242 had been previously tested by the HCV ELISA; 126 (52%) of them had a positive result. Of these, 107 were also analyzed by the HCV RT-PCR test, and 78 (73%) of them were positive for HCV RNA. Seventy-five (96%) of the RNA-positive samples were analyzed by the newly developed nested RSS-PCR subtyping procedure. The remaining three samples were weakly positive by RT-PCR and hence could not be typed by the RSS-PCR method. As positive controls, we used 15 HCV RNA-positive serum samples, whose viral titers were known, obtained from patients in San Francisco whose infecting HCV genotype was known on the basis of the results of the line probe DNA hybridization assay and sequence analysis of the 5' NC and NS5b regions (Table 2).

RSS-PCR. The nested PCR was designed to generate six distinct amplicons of various sizes (606, 478, 451, 323, 256, and 101 bp) from the prototype genotype 1b strain (strain HCV-J), as shown schematically in Fig. 1. The number of bands actually observed in agarose gels varied, however, probably due to the difficulty in resolving the two amplicons of similar size (478 and 451 bp) and the small (101-bp) product. However, four predominant bands representing amplicons of 606, 478 or 451, 323, and 256 bp were reproducibly and consistently observed with HCV subtype 1b (see below). The 256-bp amplicon was produced from a segment in the 5' NC region that is highly conserved among all HCV genotypes, and its amplification indicates the presence of any HCV RNA in the test sample. Thus, this PCR product served as an internal positive control in every nested reaction.

View larger version (35K): [in this window] [in a new window]   FIG. 1.   RSS-PCR patterns generated from U.S. serum samples containing HCV of known genotype. Lane A diagrammatically represents the predicted electrophoretic band pattern and the sizes of the products amplified from a sample that contains HCV genotype 1b RNA. By the line probe assay, the samples were found to contain HCV genotypes 1a (lane 1), 1b (lane 2), 2a (lane 3), 2b (lane 4), 3a (lane 5), 4 (lane 6), 4 (lane 7), 4 (lane 8), 6 (lane 9), 6a (lane 10), and 3a (lane 11). Lane MW, 100-bp molecular size marker (the numbers on the left are in base pairs). Although the sample in lane 9 was shown by the line probe assay to contain genotype 6 HCV, analysis of the 5' NC sequence of the sample found that it differed from the prototype genotype 6 5' NC sequence by 7 nucleotide positions. However, the NS5b sequence was more than 93% identical to the HCV genotype 1b NS5b sequences deposited in the GenBank database.

Validation of nested RSS-PCR results by comparison to results for serum samples containing HCV RNAs of known genotypes. U.S. serum samples 2529 and 884, whose HCV titers were known, had been found by the line probe assay to contain HCV genotype 1b RNA (Table 2). The nested RSS-PCR electrophoretic band patterns generated for the strains from these samples were identical to each other and were consistent with the predicted pattern for genotype 1b (Fig. 1, lane A), while the strains from the remaining samples produced patterns which were distinct from that for genotype 1b (Fig. 1, lanes 1 and 3 to 11). The products of 478 or 451 and 323 bp were always seen together for HCV subtype 1b, while the products of 606, 478 or 451, and 101 bp were observed at various frequencies and intensities among subtype 1a and genotype 2, 3, 4, and 6 strains. Thus, subtype 1b generated an RSS-PCR pattern distinct from those for subtype 1a and other genotypes.

View larger version (92K): [in this window] [in a new window]   FIG. 2.   RSS-PCR patterns generated from clinical serum samples from patients with hepatitis C in Prague, Czech Republic. In each panel, the RSS-PCR patterns are compared to the subtype 1b RSS-PCR pattern. (A) Samples predicted to contain HCV subtypes 1a (lanes 1 to 4, 7, and 8) and 1b (lanes 5 and 6). Lanes 1 (strain 094) and 3 (strain 152) contain samples that contain HCV subtype 1a with a 5' NC sequence that was 100% identical to that of the prototype genotype 1a HCV, whereas the samples in lanes 2 (strain 096) and 4 (strain 145) contained HCV that differed from the prototype 1a strain by a single CA change at position 138. The genotype 1a pattern in lane 8 was generated from a sample (strain 048) that contained HCV with the same CA nucleotide change, in addition to four other changes upstream. (B) Clinical samples predicted to contain HCV of subtype 1b. Lanes 1 to 3, HCV RNA with a 5' NC sequence 100% identical to the prototype genotype 1b sequence; lane 4 (strain 177), a sample that contained HCV with an extra A at position 137. (C) Comparison of genotype 1b patterns (lanes 1 and 2) with a predicted genotype 3a RSS-PCR pattern (lane 3). Note that the clinical sample from Prague in lane 3 generated a pattern identical to one of the subtype 3a-containing samples from the United States shown in lane 11 in Fig. 1. The U.S. sample was confirmed by sequencing to contain the HCV genotype 3a.

Validation of nested RSS-PCR by sequence analysis of the clinical samples. We sequenced the 256-bp product from the 5' NC regions from all 75 Prague samples to confirm the results presented above. Of the 54 strains found by RSS-PCR to have the genotype 1b pattern, 46 (85%) had 5' NC region sequence that was 100% identical to the corresponding 5' NC region sequence of prototype strain HCV-J. Six others had sequences that differed by only 1 nucleotide within the 176-bp 5' NC segment from that for strain HCV-J. These nucleotide changes were not located at positions used by Simmonds et al. (31) and Smith et al. (33) to differentiate the HCV genotypes. All six had a G at position 99, just like the other 46 strains. Simmonds et al. (31) found that most strains of subtype 1a, as well as strains of other genotypes, have an A at this position (position 99). Four of these six strains had an identical single TC change at position 94, one had a CT change at position 138, and the last one had a CA change at 138. We found the same 5' NC nucleotide changes in other HCV genotype 1b sequences deposited in GenBank. We compared the NS5b sequences of the strains in these six samples and found that all were of subtype 1b. The additional two samples with the subtype 1b pattern as determined by RSS-PCR had an extra A at position 138. This change was not found among available GenBank subtype 1b 5' NC sequences, but the NS5b sequences of these strains were more than 93% identical to the GenBank subtype 1b NS5b sequences.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Genotypic analyses of HCV are often confined to reference or research laboratories because of the need for nucleic acid sequencing or the use of kits whose high costs limit their accessibility to most laboratories. Here, we report a modification of a simple new PCR-based strain-typing method that can be used to generate fingerprint patterns representing HCV genotypes directly from clinical material. This method relies on nucleotide sequence polymorphisms that occur at restriction endonuclease recognition sites without the use of any restriction endonucleases. The technique was first developed for the typing of dengue viruses (9, 17) and was more recently applied to the typing of E. coli strains associated with diarrheal diseases (12). The application of this method to the typing of HCV, however, required modification because this organism cannot be cultured in vitro. Thus, this is the first application of the RSS-PCR method in which strain typing is performed directly from clinical specimens. This approach may be useful for the typing of other pathogens that cannot be easily isolated or cultured. Furthermore, the test was designed in such a way that the generation of a pattern not only provides information about the genotype but also simultaneously indicates the presence of HCV RNA in a clinical sample. The observation of a 256-bp PCR product from each sample indicates the presence of HCV-specific RNA, which adds to the specificity of the test.

This RSS-PCR method for the typing of HCV was developed by using information obtained from previous studies that used the 5' NC region to genotype HCV strains. Davidson et al. (6) typed HCV by restriction fragment length polymorphism analysis of a sequence within the 5' NC region amplified by PCR and cleaved with restriction endonucleases. They were able to distinguish genotypes 1a, 1b, 2a, 2b, 3a, 3b, 4, 5, and 6 by this method. The nested RSS-PCR method was designed to target polymorphic sequences within the same 5' NC region used for restriction fragment length polymorphism analysis by Davidson et al. (6), in addition to the restriction site polymorphisms found in the C region of genotype 1b, using the sequence derived from strain HCV-J as a reference. Stuyver et al. (34) genotyped HCV by DNA hybridization using the line probe assay, in which sequence variability within the 5' NC region was also used as the target of a probe constructed from a DNA product amplified from clinical specimens. The line probe assay is commercially available as a kit, but its cost makes it inaccessible to most laboratories in regions of the world with increasing prevalence of HCV infections, such as Eastern and Central Europe, where specimens for this study were collected. One of the major advantages of the nested RSS-PCR method is that any laboratory capable of performing PCR can readily apply the procedure to the typing of HCV.

Differentiation of HCV strains into genotypes has already become an integral part of epidemiologic investigations of hepatitis C. Several studies have attempted to examine the role of HCV genotypes in clinical outcome following HCV infection, such as progression of the liver disease, response to interferon therapy, or development of chronic infection. Differences in clinical outcomes and responses to therapy in patients with HCV infections are well recognized. Amoroso et al. (1) found that the rate of progression to chronic infection after acute exposure to HCV was 92% among patients who were infected with genotype 1b, whereas that rate was 33 to 50% among those infected with strains of other genotypes. Zein et al. (37, 39) found that genotype 1b strains occurred significantly more frequently among patients with cirrhosis and those patients requiring liver transplantation. Patients infected with genotype 1b and possibly genotype 1a may have a more unfavorable response to treatment with interferon than those infected with genotypes 2 or 3 (38). On the basis of these observations, the European Association for the Study of the Liver in 1999 announced a consensus statement that for patients infected with genotype 1 and with viremia of >2 × 10⁶ copies/ml, 12 months of therapy with interferon and ribavirin should be given, whereas for patients infected with genotypes 2 or 3, 6 months of therapy with interferon and ribavirin should be given, regardless of the level of viremia (8).

The significance of the relationship between HCV subtype 1b infection and clinical outcome remains unclear, as others have not found such associations with this genotype (2, 7). Zein (40) suggested that HCV genotype 1b may have appeared in the human population before the other genotypes, and thus, patients exposed to HCV genotype 1b may have been infected for a longer time. Hence, the association of this genotype with severe disease may reflect differences in patients' durations of infection. The continued uncertainty about the relationship between genotype 1b and clinical outcomes of hepatitis C infection emphasizes the need to examine this question in depth with studies performed at more sites. The availability of a simple test to distinguish genotype 1b from other genotypes provides an opportunity to expand such studies.

It should be emphasized that the RSS-PCR method reported here should be regarded as a proof-of-concept approach to the typing of HCV, in that the primer designs can be modified to generate a fingerprint pattern specific to any genotype or subtype of interest. Primer designs may be modified to generate patterns based on the nonstructural genes (for example, the NS5b region) of HCV or to specifically target HCV subtype 1a. Hence, it offers great flexibility in the design of studies appropriate for local hepatitis C epidemiologic situations.

Other PCR-based methods for the typing of HCV exist (14, 23, 24, 25, 27, 36). However, the other reported methods require the use of restriction endonucleases to generate a fingerprint pattern or a larger number of primer sets designed to differentiate the HCV genotypes by displaying differences in the sizes of a single amplified product. The nested RSS-PCR method generates characteristic band patterns without endonucleases, which enhances specificity and minimizes false-positive signals due to contaminating and cross-reacting DNA. While the technique does not have the discriminatory power of direct sequencing, it was able to rapidly differentiate the common HCV subtypes circulating in Prague, which are similar to those circulating in Western Europe and North America. In fact, it correctly identified the virus in a U.S. sample to be genotype 1b when the virus was initially misclassified as genotype 6 by the line probe assay. The technique is considerably cheaper and simpler to perform than the line probe DNA hybridization assay and is certainly more feasible than sequencing in countries with limited abilities to do that level of testing on a routine basis. A simple method for the differentiation of genotype 1b from genotype 1a and other HCV subtypes should make it accessible to more geographic areas and laboratories to facilitate improved clinical outcomes and epidemiologic studies. Such studies may provide information that has major implications for clinical management of hepatitis C and possibly HCV vaccine development.