ABSTRACT
The genotype of the infecting hepatitis C virus (HCV) helps determine the patient's prognosis and the duration of treatment. Heteroduplex mobility analysis (HMA) is a rapid, inexpensive method for genotyping of HCV that does not require sequencing. We developed an HMA that uses temperature gradient capillary electrophoresis (TGCE) to differentiate HCV genotypes. A 56-bp region of the HCV 5′ untranslated region (UTR) that was conserved within a genotype yet whose sequence differed between genotypes was amplified for HMA-TGCE analysis. HCV amplicons of types 1, 2a, 2b, 3a, 4, and 6a were hybridized in pairs and analyzed by TGCE. Amplicons hybridized to the same subtype yielded one homoduplex peak, while hybridization of different subtypes resulted in heteroduplexes and generated multiple TGCE peaks. Heteroduplexes contain thermodynamically unstable nucleotide mismatches that reduced their TGCE mobilities compared to those of homoduplexes. Three HCV subtypes (subtypes 1a, 3a, and 4) generated unique peak patterns when they were combined with each genotype analyzed and were chosen as the reference genotypes. A blinded study with 200 HCV-infected samples was 97% accurate compared to genotyping by 5′ UTR sequence analysis. The majority of discordant results were unexpected sequence variants; however, five of nine sequence variants were correctly genotyped. The assay also detected and correctly genotyped mixed HCV infections. Compared to conventional HMA, TGCE improves the resolution, with better separation of heteroduplexes and homoduplexes. All common HCV genotypes can be detected and differentiated by this HMA-TGCE assay.
Hepatitis C virus (HCV) is an enveloped single-stranded RNA virus in the family Flaviviridae. HCV infects the liver, resulting in acute hepatitis or chronic liver disease, and it is the leading causative agent of adult liver transplants in the United States. Worldwide, 170 million people are infected with HCV, mainly from the sharing of tainted drug needles or by other means of exposure to infected blood (2, 10, 19, 20, 21). HCV has 6 major genotypes and greater than 50 subtypes. Genotyping of HCV is important for determination of the appropriate treatment protocol, prediction of a sustained virological response, and epidemiological studies (9, 12, 13, 21). The most common HCV type causing infections in the United States is type 1, which has a poor response to treatment (10, 12, 13, 19, 20).
Interferon and ribavirin are used in combination to treat HCV infections. Patients infected with HCV type 1 require treatment for 48 weeks, while patients infected with types 2 and 3 demonstrate no additional benefit from treatment after 24 weeks (9, 10, 12, 13). It is important to keep the length of treatment as short as possible to lower drug expense, reduce the duration of side effects, and increase the rate of adherence to the treatment protocol. Even with the longer treatment protocol and combination therapy, half as many patients infected with type 1 as patients infected with types 2 and 3 have a sustained response (2, 9, 10, 12, 13). Patients infected with the other major genotypes, genotypes 4, 5, and 6, are also poor responders (12).
The 5′ untranslated region (UTR) of the HCV genome is commonly used for HCV detection and genotyping assays because it is highly conserved yet contains specific polymorphisms that distinguish the six major genotypes and some subtypes (17, 18, 20). The methods used to determine the HCV genotype include line probe assays, sequencing, and heteroduplex mobility analysis (HMA) (3, 5, 12, 13, 15, 18, 20). Sequencing is the reference standard, but it is a time-consuming, complex procedure and may not detect mixed infections, which can require an additional DNA cloning step before sequencing (3, 5, 17). HMA is simple to perform and less expensive than other HCV genotyping methods (18). For HMA, two nonidentical but related DNA molecules are paired, denatured, and reannealed to create a mixture of homoduplexes and newly created heteroduplexes. The heteroduplexes are more structurally unstable than the homoduplexes due to mismatches and unpaired nucleotides. These structural differences cause heteroduplexes to have reduced mobilities compared to those of the homoduplexes. The resulting reduction in band mobility and peak mobility can be detected by polyacrylamide gel electrophoresis and capillary electrophoresis, respectively (14, 16). Mixed infections and quasispecies (sequence variants) can be detected by this procedure without DNA cloning and sequencing (16, 17).
The reduced mobility of heteroduplexes due to nucleotide mismatches can be increased by temperature gradient capillary electrophoresis (TGCE) (7, 16). Heteroduplexes and homoduplexes can be differentiated by TGCE by heating the amplicons so they partially denature during migration through capillaries. Heteroduplexes denature at a lower temperature and migrate more slowly than homoduplexes during capillary electrophoresis (7). A homoduplex generates only one TGCE peak, while heteroduplexed samples can yield up to four peaks (7).
We developed an HMA-TGCE assay to differentiate HCV genotypes 1a/1b, 2a, 2b, 3a, 4, and 6a. HMA-TGCE identified sequence variants by their unique peak patterns, and these variants were verified by sequencing. Mixed infections were identified by HMA-TGCE, and the viruses in the mixed infections were genotyped as well. The accuracy of the newly developed HMA-TGCE HCV genotyping assay was validated with 200 blinded patient samples.
MATERIALS AND METHODS
HCV-infected clinical samples.The HCV-infected clinical samples used for this study were previously quantified and genotyped by ARUP Laboratories. Plasma samples in EDTA anticoagulant were submitted to ARUP Laboratories and were then centrifuged and frozen at −20°C. The viral load was obtained by the AMPLICOR HCV MONITOR (version 2.0) assay (Roche Molecular, Indianapolis, Ind.). The HCV genotype was determined by sequence analysis of the 244-bp HCV 5′ UTR product amplified by the AMPLICOR assay. Use of this region for HCV genotyping by sequence analysis is well established (5, 6), and this is the same region used for the line probe assay (15). The bidirectional sequence obtained by use of Big Dye terminator chemistry (Applied Biosystems, Foster City, Calif.) on an ABI PRISM 377 DNA sequencer was edited with Sequencher (version 4.1) software, generating a consensus HCV sequence for each sample. The viruses in the clinical samples were assigned an HCV genotype by using MatchTools (version 1.0) software to match its consensus sequence to a library of known HCV genotypes. Clinical samples were deidentified with institutional review board (IRB) approval (IRB no. 7275). Thirty-seven plasma samples infected with various HCV genotypes and subtypes were used to develop the HMA-TGCE HCV genotyping assay.
HCV RNA extraction.HCV RNA was extracted from plasma samples with a QIAamp viral RNA extraction kit (Qiagen, Valencia, Calif.), according to the instructions of the manufacturer. In brief, virions were lysed and the RNA genomes were captured onto the spin columns provided with the kit. The HCV RNA was washed and then eluted by incubating the spin column for 1 min in 60 μl of sterile molecular biology-grade water, followed by a 1-min centrifugation at 6,000 × g at room temperature. The HCV RNA was collected and placed into a clean 1.5-ml Eppendorf tube and stored at −20°C until it was used.
RT-PCR.The 5′ UTR of the HCV RNA genome was amplified by using a OneStep reverse transcription (RT)-PCR kit (Qiagen), as stated in the Qiagen protocol. A sample volume of 2 μl of purified HCV RNA (approximately 100 to 1,000 IU of virions were extracted per reaction mixture) was added to a final RT-PCR mixture volume of 50 μl. The RT-PCR mixture contained the OneStep 1× enzyme, OneStep 1× buffer, 0.32 mmol of each deoxynucleoside triphosphate (dATP, dCTP, and dGTP) per liter, 0.64 mmol of dUTP (GeneAmp; Applied Biosystems, Foster City, Calif.) per liter, 1 mmol of MgCl2 per liter, 0.5 μmol of each primer (CTGCGGAACCGGTGAGTACACC and ATCCAAGAAAGGACCC) per liter, 0.01 U of uracil-DNA glycosylase (Roche Molecular) per μl, and 0.3 U of RNase inhibitor (Roche Molecular) per μl. The primers generated a 56-bp genotyping amplicon (from positions −139 to −194 of the HCV 5′ UTR; GenBank accession NC_004102 ). The genotyping amplicon was identified by using an in-house sequence analysis program based on consensus in the primer site sequence between genotypes, internal sequence divergence between genotypes, and internal sequence conservation within genotypes. This area of the HCV genome has been used in other HCV genotyping assays (1, 6, 8, 15).
Thermocycling was preformed on a GeneAmp 9700 instrument (Applied Biosystems) with the following parameters: the uracil-DNA glycosylase step was 50°C for 10 min, the RT step was 60°C for 30 min, and activation of the PCR was at 95°C for 15 min. Forty-five PCR cycles consisted of three steps: denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 60 s. The final extension was at 72°C for 4 min. The HCV PCR amplicons were not purified before use in the HMA-TGCE assay.
Heteroduplex formation.PCR amplicons were diluted to approximately 10 to 20 ng/μl with 1× PCR buffer II (Applied Biosystems). Six microliters of each genotyping amplicon alone or a 1:1 mixture of a test amplicon with a reference amplicon (type 1a, 3a, or 4) was overlaid with 10 μl of mineral oil, denatured at 95°C for 5 min, annealed at 50°C for 30 min, and cooled to 4°C for storage on a thermal cycling instrument (9700; Applied Biosystems) (18).
TGCE.TGCE was performed on a commercial instrument (Reveal mutation discovery system, reagents, and Revelation software; all from SpectruMedix LLC, State College, Pa.) (7). DNA samples were injected electrokinetically at 3 kV for 20 s, resulting in peak heights of approximately 10,000 intensity units with ethidium bromide staining. Optimal results were obtained when the temperature was ramped from 45 to 55°C over 18 min, and data were acquired over 35 min. Sequential camera images were converted to plots of image frame number (time) versus intensity units (DNA concentrations). Although the exact frame location for each sample was variable, the number of peaks and the spacing of the peaks allowed genotyping. The running time for a 24-sample tray was 1.5 h, with an additional 0.5 h required to wash the capillaries for the next run. After about 100 runs of a capillary cassette, a loss of resolution was observed, particularly with a type 1a-type 4 mixture. Three reference genotypes (types 1a, 3a, and 4) and the type 1a-type 4 mixture were run on each tray as controls.
To select the reference genotypes, all available HCV genotypes (genotypes 1a, 1a variant, 1b, 2a, 3a, 4, 4 variant, 6a, 2b, and 2b variant) were paired with each other, heteroduplexed, and analyzed by TGCE (data not shown). The three reference genotypes selected had unique peak patterns compared with those of all the other genotypes tested, and as the smallest reference panel, they were able to genotype the other genotypes. The amplicon regions of the three reference genotypes used for genotyping have the consensus sequence of their respective genotypes.
RESULTS
The HCV 5′ UTR genotyping amplicon.The 56-bp segment of the HCV 5′ UTR chosen for analysis (the genotyping amplicon) contains an 18-bp sequence that is conserved within a genotype but that differs between the common HCV genotypes (Table 1). To amplify all possible HCV genotypes, the primers for the genotyping amplicon annealed to highly conserved regions of the 5′ UTR flanking the 18-bp variable region. This was confirmed by analysis of 200 random HCV sequences that demonstrated complete identity at the 3′ ends of both primers for at least 10 bases. The results for the HCV isolates in 2,015 clinical samples sequenced and genotyped at ARUP were used to determine the HCV genotype frequency (Table 1). The most common genotypes were 1a, 1b, 2a, 2b, and 4, which together represented 98.8% of the HCV-infected patient samples. The rare HCV genotypes 1c, 2c, 3c, 3d, 5, 6b, and 7 to 11 were not identified in any of the samples submitted to ARUP. HCV subtypes 1c, 2c, 3b, and 3c/d had genotyping amplicon sequences that were identical to those of the HCV genotypes included in the study: 1, 2a, 4, and 3a, respectively (3, 15, 18). Of these subtypes, only the rare 3b subtype would be misclassified as a different genotype (genotype 4). The genotyping amplicons of genotypes 6b, 7, 8, 9, and 11 were identical to that of genotype 1 or 6a. Although genotypes 5 and 10 had a unique 18-bp sequence, they were not available for study (3, 15, 18).
Comparison of HCV subtype sequences
Patient samples with HCV subtypes 1a, 1b, 2a, 2b, a common 2b variant, 3a, and 4 were used to develop the HMA-TGCE HCV genotyping assay. The 56-bp genotyping amplicons of genotypes 1a and 1b have the same sequence, so they cannot be subtyped by this assay and are referred to as type 1. Type 2b had a common sequence variant present in 25% of the type 2b-infected samples analyzed. Rare, single clinical samples with type 2, type 6a, and sequence variants of types 2a, 1, and 4 were analyzed by HMA-TGCE (the sequences are listed in Table 1).
HCV genotyping by HMA-TGCE.The amplicons of known sequence and genotype were mixed in pairs, denatured, and hybridized as previously described for an HMA (18). The homoduplexes and heteroduplexes present in the mixture were separated by TGCE. If the paired amplicons had the same sequence or were not denatured and hybridized, they ran as one peak on TGCE, as only homoduplexes were present. Pairs of HCV genotyping amplicons with divergent sequences formed heteroduplexes with nucleotide mismatches that decreased their thermodynamic stabilities compared to those of the homoduplexes. This resulted in heteroduplexes with reduced mobilities, and they resolved from the homoduplexes as additional peaks by TGCE.
The HMA-TGCE peak patterns generated by combining the various HCV genotypes with each of three reference genotypes (genotypes 1a, 3a, and 4) demonstrated reproducible peak patterns unique to each genotype (Fig. 1A). HCV type 1 generated one peak (homoduplex only), three peaks, and a shoulder peak with reference genotypes 1a, 3a, and 4, respectively. Type 2a had two peaks with each of the three reference genotypes. Types 2b and 3a had similar banding patterns with types 1a and 4, but each had a unique peak pattern with reference genotype 3a. Type 2b heteroduplexed with reference type 3a had two widely separated TGCE peaks (30 to 35 frames apart), while test sample type 3a paired with reference type 3a contained only a homoduplex and ran as one peak. HCV type 4 was distinguished from the other genotypes by a shoulder peak with reference type 1a and one homoduplex peak with reference type 4. An extensive reference panel was not needed to genotype the unknown isolates in HCV-infected patient samples; only three reference amplicons were required: those for types 1a, 3a, and 4.
Typical HMA-TGCE peak patterns obtained with the three reference genotypes. (A) Electropherograms from the HMA-TGCE HCV genotyping assay, with frames on the x axis and intensity units on the y axis. Representative traces for genotypes 1, 2a, 2b, 2b variant, 3a, and 4 are shown either alone (homoduplex only) or paired with each of the three reference genotypes, genotypes 1a, 3a, and 4. The number of peaks is provided under each graph. When the same genotypes were paired, one homoduplex peak was generated, as indicated by an asterisk. Graphs labeled a, b, and c are for comparison to the graphs in panel B. (B) Panel a, HMA-TGCE peak pattern for type 6a and reference type 4 (compare this pattern with that for type 1 and reference type 4 in panel A; see text); panel b, peak pattern for type 2a variant and reference type 1a (compare this pattern with that for type 2a and reference type 1a in panel A); panel c, peak pattern for type 4 variant and reference type 4 (compare this pattern with that for type 4 and reference type 4 in panel A). ref, reference.
A common variant of type 2b was detected by altered spacing of the HMA-TGCE peaks when it was mixed with reference type 1a or 3a (Fig. 1A). One of the heteroduplex peaks, formed by the type 2b variant mixed with reference type 1a, was less delayed than the type 2b heteroduplex peak obtained with reference type 1a. The only type 6a-infected sample tested had peak patterns similar to those of type 1, except when it was hybridized to reference type 4, with which it yielded two peaks (Fig. 1B, panel a).
Rare sequence variants of types 2a, 1, and 4 were identified by their unique HMA-TGCE peak patterns when they were mixed with the reference genotypes. The divergence of the sequence from the consensus sequence may shift the relative position of the heteroduplex peaks (e.g., type 2b variant) or change the number of peaks observed from the typical HMA-TGCE genotype peak pattern. For example, the type 2a sequence variant had three peaks (one of which was a shoulder peak) with reference type 1a, while genotype 2a had two peaks with reference type 1a (Fig. 1A and B, panels b). A type 4 variant paired with reference type 4 had two peaks, not the one homoduplex peak expected for a typical type 4 isolate (Fig. 1A and B, panels c).
Resolution of heteroduplex detection.The HMA-TGCE assay detected heteroduplexes in two of four paired HCV subtypes that differed by only a single nucleotide. Two paired genotypes that differed by 1 bp, type 4 variant-type 1 and type 2b-type 2b variant, had detectable heteroduplexes that ran as shoulder peaks by HMA-TGCE as compared to the type 2b homoduplex (Fig. 2A to C). The pairs type 6a-type 1 and type 2b variant-type 2a also differed by a single base pair, but they ran as one peak by HMA-TGCE (the results for type 6a-type 1 are shown in Fig. 2D). All other samples analyzed had 2 to 7 nucleotide differences in the 56-bp genotyping amplicon and generated detectable heteroduplexes (Fig. 1 and Table 1). The resolution of heteroduplex detection by TGCE is 1 to 2 nucleotides in the 56-bp genotyping amplicon. In general, as the hybridized amplicons increased in sequence divergence, the heteroduplex peak was slower in mobility. This is a common occurrence in HMA assays, in which the degree of sequence variation can be correlated with heteroduplex mobility (4, 11, 16, 18).
Resolution of HMA-TGCE HCV genotyping assay. (A) HMA-TGCE electropherogram for a homoduplex sample (genotype 2b only); (B to D) HMA-TGCE peak patterns for paired genotyping amplicons with 1 nucleotide divergent.
Detection and genotyping of mixed HCV infections.When clinical samples generate multiple HMA-TGCE peaks without added reference types, a mixed infection is indicated. HMA-TGCE also allowed genotyping of mixed infections without additional cloning and/or sequencing. By comparing the peak pattern of the clinical sample to the HMA-TGCE pattern of paired control genotypes, the genotypes of the mixed infection can often be determined. In Fig. 3A, without an added reference, sample 1 had three HMA-TGCE peaks. The peak pattern for sample 1 is identical to the unique HMA-TGCE peak pattern for the type 2b variant paired with type 1. Sample 2 had three peaks when it was analyzed alone, a peak pattern consistent with two possible mixtures: genotype 2b-genotype 1 or genotype 3a-genotype 1 (Fig. 3B). The addition of reference type 3a yielded a low-intensity heteroduplex peak 35 frames away from the homoduplex peak. This is typical of mixtures of types 3a and 2b and thus indicates that sample 2 contains type 2b and type 1. Further HMA-TGCE data for combinations of samples 1 or 2 with each of the three reference types confirm the deduced genotypes (data not shown). Both mixed HCV-infected samples were sequenced, and the results concurred with the HMA-TGCE data.
HMA-TGCE peak patterns for isolates from patients with mixed HCV infections. Patient samples 1 and 2 each contain two HCV genotypes. (A) HMA-TGCE data for sample 1 alone displayed next to the control HMA-TGCE peak pattern for the type 2b variant paired with reference type 1a for comparison; (B) HMA-TGCE data for sample 2 alone and mixed with reference type 3a. The control HMA-TGCE peak patterns are presented for comparison. The paired genotypes used for HMA-TGCE are listed below each control graph. ref, reference.
Detection of minor species.Various ratios of paired genotyping amplicons (types 1b and 2b) were analyzed by HMA-TGCE for detection of heteroduplex peaks and identifiable genotype patterns (Fig. 4). Heteroduplex peaks could be detected when the minor amplicon species was present at 5 to 9%, but the minor species had to be present at 20% to determine the genotype. This experiment was repeated with another set of HCV-infected samples, genotypes 3a and 2a, which yielded the same results (data not shown).
Detection of minor species by HMA-TGCE. Genotype 2b was mixed with genotype 1b and analyzed by HMA-TGCE. Genotype 2b was present at 100, 50, 33, 20, 9, and 5% of the total amplicon species, as indicated. Similar results were found when type 1b was present as the minor species.
Blinded genotyping study with HCV-infected samples.Of 200 clinical samples tested in a blinded manner, the HCV isolates from 195 were successfully amplified and genotyped according to their HMA-TGCE peak patterns. The five samples from which virus was not amplified had the lowest titers among the blinded samples and demonstrated one homoduplex peak with each of the three reference types. This is the expected HMA-TGCE pattern for no viral amplification. Compared to the results of HCV genotyping by sequence analysis, the isolates in 97% (190 of 195) of the samples were correctly genotyped (Table 2). An incorrect genotype was identified for only 3 of 134 HCV type 1-infected samples (98% correlation). The isolate in one of these samples was correctly genotyped as type 1 by repeating the assay. The other two type 1-infected samples for which the genotypes were incorrectly identified appeared to contain genotype 6a, and one sample seemed to have a mixture of types 1 and 6a. Although sequencing of the 5′ UTR clearly identified the isolates in these samples as type 1, a singe base change was present in the genotyping amplicon that would result in a type 6a HMA-TGCE peak pattern.
HCV genotypes identified by HMA-TGCE in a blinded study of 195 HCV samples
HCV type 2a was correctly identified in six of seven samples. The isolate in one sample was genotyped as a type 2b variant by HMA-TGCE. Upon sequencing of this isolate, the genotyping amplicon sequence agreed with that for the type 2b variant genotype, but the rest of the 5′ UTR contained both type 2a- and type 2b-specific sequences. This particular isolate was too variable to be classified as a specific subtype, but it still could be genotyped as a type 2 isolate. The isolates in only two samples were type 4, and one was a type 4 sequence variant that was misidentified as a type 1 variant. The genotyping amplicon of the type 4 variant had 2 nucleotide changes compared with the typical type 4 sequence. The results for all HCV isolates genotyped by HMA-TGCE as type 2b and type 3a had a 100% correlation to the results of sequence analysis.
Three samples tested in a blinded manner generated HMA-TGCE peak patterns that indicated that they contained mixed infections. The isolates in these samples were sequenced, and each contained two possible nucleotides at one site within the genotyping amplicon (data not shown).
DISCUSSION
A larger amplicon of the HCV 5′ UTR (175 to 224 bp) is commonly used to genotype HCV by HMA or sequence analysis (3, 5, 15, 17, 18). HMA-TGCE analysis of large HCV amplicons yields peak patterns that are complex and difficult to interpret, because single-nucleotide variations present within the same genotype can be detected (7). For this reason, we studied a smaller, 56-bp 5′ UTR amplicon that had sequence conservation within a genotype, sequence divergence between genotypes, and highly conserved regions for PCR amplification.
All HCV genotypes analyzed (types 1, 2a, 2b, 2b variant, 3a, 4, and 6a) had unique sequence differences within the 56-bp genotyping amplicon that correlated to unique HMA-TGCE peak patterns when the clinical isolates were paired with three reference genotypes: genotypes 1a, 3a, and 4. HCV subtypes 1a and 1b could not be distinguished because they have identical sequences within the genotyping amplicon (3). As the paired HCV amplicon sequences became more divergent, the heteroduplex peaks increased in their TGCE separation from the homoduplex peak, in agreement with previous HMA assays (4, 14, 16).
Heteroduplexes generated from two sets of paired samples in which the isolates diverged by only 1 nucleotide were separated from homoduplexes. This result highlights the increase in resolution obtained by TGCE over that obtained by the standard means of detection by HMA and polyacrylamide gel electrophoresis (PAGE) (17, 18). The HMA-PAGE assay could not detect heteroduplexes with 4 or fewer nucleotide changes (17). In that assay, heteroduplexes of paired HCV isolates (type 4-type 1, type 2a-type 2b, and type 2a-type 4) did not separate from homoduplexes and ran as one band by PAGE (18). In contrast, these pairs had detectable heteroduplex peaks by TGCE (Fig. 1). The only genotype pairs that were not separated by TGCE (type 1-type 6a and type 2a-type 2b variant) had sequences that differed by 1 bp. However, even the isolates in these samples could be genotyped by HMA-TGCE when they were analyzed with the three reference types. All other tested genotype pairs that differed by 1 or more nucleotides yielded detectable heteroduplexes. In another study that used HMA and capillary electrophoresis, a resolution of 2 nucleotide differences in a 179-bp fragment of the HCV hypervariable region 1 was obtained (14).
Li et al. (7) showed that a minor amplicon species present at a level of 10% of the total sample tested by HMA could be reliably detected by TGCE. White et al. (17), using HMA and PAGE, detected minor HCV species present at levels of 4 to 10%. In our study, dilution studies using HMA-TGCE demonstrated that a minor species can be detected when it is present at levels of 5 to 9%, but the minor species needs to be present at a level of 20% to generate distinguishable peak patterns to allow genotyping.
Mixed HCV infections can be genotyped by HMA-TGCE without DNA cloning or sequencing. Three samples from the blinded study appeared to contain mixed infections by their HMA-TGCE peak patterns. The predicted mixed sequence was confirmed by sequencing of the isolates from these three samples. However, the sequence outside of the genotyping amplicon revealed only one genotype. Two of the mixed infections contained type 2b and the type 2b variant, so the subtype was correctly identified even with the sequence variation. The other mixed-infection sample contained type 6a and type 1. By sequencing, this sample was found to contain type 1 outside of the genotyping amplicon and an internal base change characteristic of type 6a. One additional blinded sample was predicted by HMA-TGCE to contain a type 6a isolate due to the same base change, but the sequence of the 5′ UTR indicated that it was a type 1 isolate. These rare type 1 isolates that have a type 6a sequence within the genotyping amplicon would require sequencing for differentiation.
Genotype sequence variants were identified by their unique HMA-TGCE peak patterns. Although the determination of sequence variations may not be important for the appropriate treatment of HCV-infected patients, the specific detection of these variants places an atypical HMA-TGCE peak pattern in the correct genotype. For example, HCV genotype 2b has a common sequence variant, and all the type 2b variants were correctly identified in the blinded study. Other sequence variants were not as common, and most of the misidentifications in the blinded study (four of five samples) were sequence variants. Two samples contained a type 1 sequence variant that genotyped as type 6a, and another sample contained a type 4 sequence variant that genotyped as a type 1 variant. The variant in the fourth sample was correctly genotyped, although the subtype was misidentified.
It is important to separate HCV genotypes 2 and 3 from all others, because patients infected with these genotypes have a better prognosis and require a shorter duration of treatment than infections with the other genotypes (9, 10, 12, 13). Therefore, misidentification of type 1 variants as type 6a and misidentification of type 4 variants as type 1 will not affect the treatment protocol or the prognosis. However, subtype 3b cannot be differentiated from type 4 by HMA-TGCE, even though subtype 3a can be distinguished from type 4. Type 3b has the same genotyping amplicon sequence as the type 4 consensus sequence (the type 4 variant has a different sequence and HMA-TGCE pattern). This could lead to a change in the standard treatment protocol for a patient infected with the rare HCV 3b subtype (which occurred at a frequency of <0.1% in our statistical analysis). Type 3b can be differentiated from type 4 by sequencing the 5′ UTR. The isolates in all samples found to be type 4 by HMA-TGCE should be sequenced to rule out the possibility of a misidentified type 3b. Approximately 2% of clinical samples would need this additional testing to determine the isolate genotype, because genotypes 3b and 4 are rare in the U.S. population. Sequencing may not be cost-effective in areas of the world with higher frequencies of types 3b and 4. Alternatively, an additional region of the 5′ UTR (to include nucleotides −238 and −235) that has divergent sequences between types 3b and 4 (15) could be used in an HMA-TGCE assay to differentiate these genotypes without sequencing. This region has 2 nucleotide changes between type 3b and type 4, and the HMA-TGCE assay can detect this level of divergence (Table 1 and Fig. 1). The HMA-TGCE HCV genotyping assay was designed and optimized with samples primarily obtained from individuals in the United States. In other areas of the world where the distribution of genotypes is different, modification of the target region and/or the use of reference samples may be required for the assay to adequately differentiate all genotypes, such as types 4 and 3b.
In conclusion, this report describes an HMA-TGCE assay for the genotyping of HCV. The assay relies on heteroduplex formation of an unknown sample to a panel of three HCV reference genotypes and detection by TGCE. The assay distinguishes all the common HCV genotypes and is 97% accurate to the subtype level. Detection and genotyping of mixed HCV infections are possible if the minor species is present at levels greater than 10 to 20%. This assay is simple to perform and has many advantages: only a small sample volume is required, the assay does not use radioactivity, the TGCE assay is automated, and there is no need for amplicon quantification or purification. Similar HMA-TGCE assays could also be used on other targets to quickly analyze regions of known sequence divergence within a strain or genotype that confer disease resistance or virulence.
ACKNOWLEDGMENTS
We thank Nora Arias for identification of HCV-infected patient samples, Lan-Szu (Bob) Chou for technical assistance, Melissa Seipp for administrative assistance, and Sam Page for sequencing support. We also thank SpectruMedix LLC for initial reagent and technical support.
FOOTNOTES
- Received 12 December 2003.
- Returned for modification 2 April 2004.
- Accepted 14 June 2004.
- Copyright © 2004 American Society for Microbiology
