Detection and identification of highly variable viral sequences is important for tracking infectious outbreaks and determining treatment regimens using targeted drug therapy. This report describes a single tube assay that is able to distinguish extensive sequence variation in hepatitis C virus (HCV) by using mismatch tolerant probes to analyze single-stranded amplicons generated with reverse transcription linear-after-the-exponential PCR (RT-LATE-PCR). Detection and identification of sequences from the 5′ non-coding region (NCR) of 31 different HCV strains was first evaluated via hybridization of two fluorescently labeled, mismatch-tolerant probes to synthetic DNA strands. The resulting data were used to calculate the ratio of fluorescent signals for the two probes over a wide temperature range as well as the melting temperature (Tm) of each probe with the targets. Although the Tm measurements alone distinguished only 5 sequences from the others, fluorescent signal ratio analysis provided a unique set of values for 27 of the 31 strains. RT-LATE-PCR was then used to amplify Armored RNA (AR) containing the 5′ NCR of five different strains of HCV. Melting analysis of the resulting single-stranded DNA with the two probes distinguished all five AR sequences. This assay can be expanded to include additional gene segments, and it points the way to construction of highly informative single-tube assays for HCV and other RNA viruses.
Diagnostic assays based on the identification of nucleic acids are now recognized as a more accurate and sensitive means of identifying viral infections, as compared with the previous gold standard assays based on cell culture or antibodies (1). In the case of HCV, viral RNA in blood is the first reliable marker for acute infection, making it useful for monitoring responses to therapy and identifying chronic infections (2). Sensitive, quantitative commercial assays for HCV based on real-time RT-PCR have been described (3, 4). The increased automation of these assays decreases the chances for operator error and laboratory contamination, since sample handling following amplification is avoided.
Single-stranded DNA is generated from RNA targets using RT-PCR with unequal concentrations of gene-specific primers and is hybridized to two mismatch-tolerant probes. The fluorescence signal ratios of the two probes are able to distinguish extensive sequence variation.
But HCV, like many other RNA viruses, has a high degree of genetic variability as a consequence of an error-prone RNA polymerase. Strains of HCV are classified into 11 genotypes and 70 subtypes (3). Genotype must be determined before therapy is initiated, because that information determines the dose and duration of treatment, as well as the probability of a favorable outcome (2). Since the FDA-approved detection assays do not identify genotype or subtype, a second assay must be done. Current commercially available tests include the TRUGENE HCV Genotyping Assay (Siemens Healthcare, Erlangen, Germany), which requires sequencing of an amplified region from the 5′ noncoding region (NCR) of the viral genome, and Line probe assays (INNO-LiPA, Innogenetics, Ghent, Belgium), which hybridize short PCR fragments to a series of short immobilized oligonucleotide probes under stringent conditions. Unlike the detection assays, these analytic tests require handling of PCR products and take much longer to complete. The Abbott RealTime HCV Genotype II assay (Abbott Molecular, Abbott Park, Il) includes primers and probes for both the 5′ NCR and the more variable nonstructural 5b (NS5b) region for improved subtype identification. The shorter assay time, reduced handling, and subtyping capability are improvements compared with those earlier methods (5), but the 91% subtyping accuracy for genotype 1 falls short of the 99% accuracy achieved by a second generation Line probe assay (6). The requirement of three reaction tubes per sample in the Abbott assay underscores the inability of a TaqMan probe to identify multiple sequence variations.
LATE-PCR is an advanced form of non-symmetric PCR with expanded detection capability that offers the potential for rapid diagnosis and genotyping. LATE-PCR generates an abundance of single-stranded DNA that can be detected over a broad range of temperatures with mismatch-tolerant probes (7, 8, 9, 10). Here we report the design and initial testing of an RT-LATE-PCR assay for HCV. The capacity of the post-PCR closed-tube analysis to distinguish sequence variations was initially evaluated using synthetic DNA oligonucleotides representing 31 different HCV strains. The full RT-LATE-PCR protocol was then tested using Armored RNA (AR) (Asuragen, Austin, TX)containing the 5′ noncoding region (NCR) of five different strains of HCV. A new analytical method that uses ratios of the signals from the different fluorophores of two probes hybridizing to the single-stranded product improves the ability to distinguish variants. Future expansion of this technique to additional targets within HCV RNA, as well as other RNA viruses, is discussed. Materials and methods Synthetic DNA targets with HCV sequences
Thirty-one HCV strains with sequence variations in the 5′ NCR segment were identified using BLAST (http://blast.ncbi.nlm.nih.gov). GenBank accession numbers for those strains are listed in Supplementary Table S1. DNA targets 60 nucleotides in length were custom synthesized for each listed sequence by Eurofins MWG Operon (Huntsville, AL, USA). A mixture of DNA target, 400 nM of each probe, 1× Tfi Reaction Buffer, and 3 mM MgCl2 was tested using the melt portion of the protocol described below. Synthetic HCV RNA
AR containing the 410 nucleotide 5′ NCR for each of 5 HCV genotypes was purchased from Asuragen (Austin, TX, USA). The stock solutions were diluted with 1 mM MgCl2, 10 mM TRIS, pH 8.3, based on estimates of particle concentrations obtained from the manufacturer. The diluted solutions were mixed with primers and heated to denature the capsid proteins, as described below. LATE-PCR primers
Primers were designed using LATE-PCR criteria for efficient production of single-stranded DNA (8). Visual OMP (DNA Software, Ann Arbor, MI, USA) was used to design primers with the desired Tm and avoid extendible self-dimers and heterodimers. The limiting primer, 5′-AAGCACCCTATCAGGCAGTACCACA-3′, is an antisense sequence fully complementary to each of the 5 RNA targets and has a predicted Tm of 69.5°C with the corresponding DNA sequence. The excess primer sequence was 5′-TGACTGGGTCCTTTCTTGGA-3′ and has predicted Tm values of 63.0°C, 65.6°C, and 62.0°C with cDNA from genotypes 1, 2, and 3, respectively, and a predicted Tm of 67.1°C with the amplification product (i.e., the fully complementary sequence). Those Tm differences are due to mismatches at or near the 5′ end of the primer with the cDNAs. The predicted Tm values are based on the PCR buffers described below and take into account the contribution from 3 mM magnesium. Custom primers were purchased from Biosearch Technologies (Novato, CA, USA). LATE-PCR mismatch-tolerant probes
LATE-PCR probes were designed to have a Tm with any known target at least 10°C below the Tm of the limiting primer. This ensures high amplification efficiency during the exponential phase of amplification (double-strand DNA production) and the absence of probe hydrolysis during extension of the limiting primer. The probes have antisense sequences so that they will hybridize with the single-stranded product generated by the extension of the excess primer. Probe sequences and modifications were 5′-Cal Red 610- TCGGCTAGTAGTCTTGTGG-Black Hole Quencher 2 (BHQ2)-3′ and 5′-BHQ2- ATGTCCGGGTATTGAGTGGGTTGAT-Quasar 670–3′. In this orientation, the fluorophores are maximally separated when hybridized to the target in order to minimize fluorescence resonance energy transfer (FRET) or quenching. Estimates of the Tm for each probe with strain-specific sequences were made using Visual OMP software. The nucleotides at some positions of the probes are not complementary to the most common target sequence but were chosen to avoid highly destabilizing mismatches. Custom probes were purchased from Biosearch Technologies. Reverse transcription, amplification, and probe-target melt analysis
Optimized RT-PCR of HCV AR was carried out as previously described (11) with slight modification. Diluted AR was mixed with limiting and excess primers at concentrations of 250 nM and 5 mM, respectively, in 0.4 mM MgCl2, 10 mM TRIS, pH 8.3, and heated to 75°C for 3 min to denature capsid proteins, then incubated at ambient temperature for 10 min. The mixture was diluted 5 fold with a PCR reagent mixture to obtain final concentrations of 0.4 mM of each dNTP, 50 nM limiting primer, 1 µM excess primer, 400 nM of each probe, 100 nM Primesafe II (10), 0.08 U/µl Platinum Tfi exo(-) DNA polymerase (Life Technologies, Grand Island, NY, USA), 4 U/µl SuperScript III (Life Technologies), 1× Tfi Reaction Buffer, and 3 mM MgCl2 in a final volume of 25 µL. Thermal cycling and melting programs were done using a Stragene Mx3005P (Agilent Technologies, Santa Clara, CA). A 5 min RT incubation at 60°C was followed by a 2 min denaturation step at 95°C, then 50 cycles of 95°C for 10 s, 60°C for 12 s, and 68°C for 30 s. Two end point fluorescence detections were done during each 60°C step. Immediately following PCR, temperature was decreased from 75°C to 25°C in 0.5°C steps of 30 s, held for 2.5 min at 25°C, then increased from 25°C to 75°C in 0.5°C steps, holding each step for 30 s and taking 3 end point detections. Melting temperature values were obtained using the Mx3005P software. For ratio analysis, fluorescence data were exported to Excel and normalized using well factors based on detection values at 65°C (at which point there was no detectable hybridization between probes and target) in order to adjust for well-to-well variation of the thermal cycler. Results and discussion
LATE-PCR makes it possible to generate high concentrations of single-stranded DNA that can be analyzed at the end point using probes which hybridize over a wide temperature range. A variable sequence in 31 HCV strains was identified using GenBank, and a synthetic single-stranded oligonucleotide for each was tested to establish whether 2 mismatch-tolerant probes labeled with different fluorophores were sufficient to distinguish all variants. Each of the synthetic targets was tested in triplicate at 100 nM, 200 nM, and 400 nM (9 total samples for each strain sequence); concentrations of the single-stranded DNA target were comparable to those expected from RT-LATE-PCR amplifications. The nucleotide variations and mean Tm of each target with each of the two probes is shown in Supplementary Table S1. The targets in Supplementary Table S1 are organized into three groups based on the Tm with the Cal Red probe. The melt derivative plots for the Cal Red and Quasar probes with strain sequences in the middle rows (strain identifications boxed; Tm with the Cal Red probe from 56.0°C to 57.1°C) are shown in Figures 1A and 1B, respectively. Only samples with target concentrations of 200 nM are shown for clarity. Although peak heights varied with target concentration, no significant change in Tm was observed for a given target sequence. To determine if the combination of Tm with the 2 probes could unambiguously identify each of the targets, a range of 3 standard deviations above and below the mean Tm for each target set of 9 samples was calculated, as that would represent over 99% of samples of a normal distribution. Each target sequence in this group generated Tm ranges that overlapped with those from one or more other targets, indicating that distinguishing sequences based solely on Tm with the probes would not be reliable. Overall, only 5 of the 31 tested strain sequences could be distinguished from all others based on the probes having a unique combination of Tm ranges.
Tm measurements provide only a limited description of the hybridization process between each probe/target pair. Accordingly, the same melt curve data were used to calculate ratios of the Quasar/Cal Red fluorescent signals over a range of 40°C to 60°C for each of the strain sequences. Figures 1C and 1D show this ratio analysis for the same strains as in Figures 1A and 1B, but in this case, samples at all target concentrations (100 nM, 200 nM, 400 nM) were included because the ratio values are independent of target concentration. The resulting ratio values provide a more detailed analysis of probe hybridization, because the ratios reflect the relative affinities of the two probes to the variants over a wide temperature range. To determine if this analysis could reliably identify each of the targets, the range of 3 standard deviations above and below the mean ratio value was calculated over the range from 40°C to 60°C in 5°C intervals for each strain sequence. In the group of strains shown in the figures, only the sequences of strains 3k.JK049 and 3k.JK055 were found to have overlapping ranges of ratios at all temperatures; the range of ratios was unique for each of the other sequences.
Similar analyses were done for the other sequences listed in Supplementary Table S1. Those having Tm with the Cal Red probe below 54°C were analyzed using a ratio of the Quasar signal at a given temperature divided by the Cal Red signal 10°C lower. Using different temperatures enabled analysis over the full melting range of each probe. Reproducible ratios were observed regardless of whether signals from the two probes were measured at the same or different temperatures.
Overall, 27 of the 31 target sequences exhibited a unique profile using multi-temperature ratio analysis. Two pairs of strain sequences overlapped with one another, but had ratio profiles distinct from all other sequences. In both cases, the overlaps were observed between sequences that differed only by mismatches near the 3′ end of the Quasar-labeled probe.
The results demonstrate that mismatch-tolerant hybridization probes generate high resolution information for hypervariable sequences. Ratios of fluorescent signals of two hybridization probes provide far more information than single probes, because ratios take advantage of any difference in hybridization over a wide temperature range, as compared to the Tm, which simply identifies the midpoint of denaturation. Unlike TaqMan probes, Molecular Beacons, or Line probes that only recognize a single sequence under stringent conditions, mismatch-tolerant probes can distinguish multiple variations, even when those variations show minor melting temperature differences with individual probes. The wide temperature range makes probe hybridization and detection possible, even with targets that have a large number of nucleotide differences from a reference sequence, making this method particularly well suited for distinguishing viral strains and other variable sequences.
Reverse transcription and amplification of the RNA sequences were done using each of 5 HCV ARs over a serial dilution range from 10,000 to 10 estimated particles. Each dilution was tested in triplicate. HCV 1a AR (Figure 2A) and HCV 2b AR generated signals from the Cal Red-labeled probe that were detected in real time, even though the probe melting temperature was below the annealing temperature and the signal was from partial hybridization of the probe. The point at which the fluorescent signal reached the detection threshold was delayed by ~3.5 cycles for each 10-fold dilution of the AR particles, demonstrating good amplification efficiency using the LATE-PCR primers. Real time signals were relatively low from HCV 2a/c AR and were undetectable from HCV 1b AR and HCV 3a AR due to the relatively low Tm of those sequences with the probe. Fluorescent signal was observed in real time or during melting analysis in all samples with an estimated 10,000 or 1,000 particles and in all samples with an estimated 100 particles except a single HCV 2a/c replicate, but only in 1 of 3 replicates with an estimated 10 particles for each AR type. Inability to detect RNA targets in some samples could be due to inaccurate estimates of target numbers, or to limits in the sensitivity of the RT-PCR protocol. Evaluating the true sensitivity of the assay will require testing clinical samples.
First derivatives of the fluorescent melt curves of the two hybridization probes targeting the amplified sequences are shown in Figures 2B and 2C. The peak values in these figures define the Tm for each probe-target hybrid. Each strain had a unique combination of Tm values with the two probes.
As with the synthetic targets, a more detailed evaluation was done using the ratio of the fluorescent signals from the two probes (Figure 2D). Each strain had its own unique pattern of ratios, but patterns among samples for a particular AR type were very similar, including those with different initial particle concentrations. The greatest variation was observed in a few samples that had generated very low fluorescent signals from the probes.
RT-LATE-PCR was also carried out using mixtures of two of the ARs in order to determine whether the presence of a small amount of a second sequence could be detected as a change in the fluorescence ratio pattern. The profiles generated by samples with 10,000 particles of HCV 1a AR, HCV 2b AR, or a mixture of the two are shown in Figure 3A. The profiles of the pure samples could be distinguished from those in samples containing either 5% or 10% of the minor component. The fluorescence ratio at 52°C of each mixed sample was compared with values from pure samples to determine if they were statistically different using the extreme studentized deviate (ESD) statistic for outliers. In the case of samples containing 5% or 10% HCV 1a, or 10% HCV 2b, each value was an outlier of the normal distribution generated by pure samples (P < 0.05). In contrast, only 2 of 4 samples with 5% HCV 1a had fluorescence ratios at 52°C that were outliers of pure HCV 2b sample values. However, close examination revealed that the slope of the profiles differed above 54°C (Figure 3B). Accordingly, the change in fluorescence ratio between 54°C and 56°C was compared and each mixed sample was found to have values that were outliers of the normal distribution from pure samples (P < 0.05).
The results presented here confirm that mismatch-tolerant probes can be used with RT-LATE-PCR to generate fluorescent signals even at temperatures well below those used during thermal cycling. Other recent studies show that it is possible to amplify gene segments that are several hundred nucleotides long and to hybridize those products with multiple probes having the same fluorescent signal (12, 13). In contrast, conventional real time PCR cannot take advantage of low temperature detection or mismatched probes because the amplification products become double stranded and prevent probes from hybridizing, which becomes more problematic for longer products (14). The RT-LATE-PCR protocol used here also takes advantage of previously published improvements of pre-incubating primers and targets, short RT incubation at high temperatures, and the use of Tfi polymerase for improved one-step results (11, 15).
This study represents a key step toward developing improved molecular detection for HCV and many other targets with variable sequences. A variable region of the 5′ NCR was chosen for analysis in order to demonstrate the ability of this method to distinguish nearly all sequence variations targeted by the probes. Two instances of nucleotide variations hybridizing near the end of one probe could not be distinguished, and modifications would be required if identification of those nucleotides were desired. However, many of the nucleotide variations in this region are not specific to HCV genotypes or subtypes, so cataloging all sequence variations for clinical identification may not be necessary. Instead, a clinical HCV assay could include primers and probes for nucleotide variations of the 5′NCR and NS5b that provide genotype and subtype information (16). Alternatively (or additionally), primers and probes could be included that identify specific genetic sequences for the viral enzymes targeted by new drugs, once the effects of those variations are better understood. In the future, it is likely that this could be accomplished in a single, closed-tube assay in less than two hours.
Financial support was provided by Smiths Detection Diagnostics and Brandeis University.
The authors are co-inventors of patented LATE-PCR, which has been assigned to Brandeis University.
Address correspondence to Kenneth E. Pierce, Department of Biology, Brandeis University, Waltham, MA. E-mail: [email protected]
1.) Josko, D. 2010. Molecular virology in the clinical laboratory. Clin. Lab. Sci. 23:231-236. 2.) Chevaliez, S. 2011. Virological tools to diagnose and monitor hepatitis C virus infection. Clin. Microbiol. Infect. 17:116-121. 3.) Le Guillou-Guillemette, H., and F. Lunel-Fabiani. 2009. Detection and quantification of serum or plasma HCV RNA: mini review of commercially available assays. Methods Mol. Biol. 510:3-14. 4.) Maheshwari, A., S. Ray, and P.J. Thuluvath. 2008. Acute Hepatitis C. Lancet 372:321-332. 5.) Martró, E., V. González, A.J. Buckton, V. Saludes, G. Fernández, L. Matas, R. Planas, and V. Ausina. 2008. Evaluation of a new assay in comparison with reverse hybridization and sequencing methods for hepatitis C virus genotyping targeting both 5′ noncoding and nonstructural 5b genomic regions. J. Clin. Microbiol. 46:192-197. 6.) Chevaliez, S., M. Bouvier-Alias, R. Brillet, and J.M. Pawlotsky. 2009. Hepatitis C virus (HCV) genotype 1 subtype identification in new HCV drug development and future clinical practice. PLoS ONE 4:e8209. 7.) Sanchez, J.A., K.E. Pierce, J.E. Rice, and L.J. Wangh. 2004. Linear-After-The-Exponential (LATE)-PCR: An advanced method of asymmetric PCR and its uses in quantitative real-time analysis. Proc. Natl. Acad. Sci. USA 101:1933-1938. 8.) Pierce, K.E., J.A. Sanchez, J.E. Rice, and L.J. Wangh. 2005. Linear-After-The-Exponential (LATE)-PCR: optimizing primer design for high yields of specific single-st randed DNA and improved real-time detection. Proc. Natl. Acad. Sci. USA 102:8609-8614. 9.) Sanchez, J.A., J.D. Abramowitz, J.J. Salk, A.H. Reis, J.E. Rice, K.E. Pierce, and L.J. Wangh. 2006. Two-temperature LATE-PCR endpoint genotyping. BMC Biotechnol. 6:44. 10.) Rice, J.E., J.A. Sanchez, K.E. Pierce, A.H. Reis, A. Osborne, and L.J. Wangh. 2007. Monoplex/multiplex linear-after-the-exponential-PCR assays combined with PrimeSafe and Dilute-'N’-Go sequencing. Nat. Protoc. 2:2429-2438. 11.) Pierce, K.E., R. Mistry, S.M. Reid, S. Bharya, J.P. Dukes, C. Hartshorn, D.P. King, and L.J. Wangh. 2010. Design and optimization of a novel reverse transcription linear-after-the-exponential PCR for the detection of foot-and-mouth disease virus. J. Appl. Microbiol. 109:180-189. 12.) Pierce, K.E., H. Khan, R. Mistry, S. Goldenberg, G. French, and L.J. Wangh. 2012. Rapid detection of sequence variation in Clostridium difficile genes using LATE-PCR with multiple mismatch-tolerant probes. J. Microbiol. Methods 91:269-275. 13.) Rice, J.E., A.H. Reis, L.M. Rice, R.K. Carver-Brown, and L.J. Wangh Fluorescent Signatures for Variable DNA Sequences. Nucleic Acids Res. In press. 14.) Vet, J.A.M., and S.A.E. Marras. 2005. Design and optimization of molecular beacon real-time polymerase chain reaction assays. Methods Mol. Biol. 288:273-290. 15.) Reid, S.M., K.E. Pierce, R. Mistry, S. Bharya, J.P. Dukes, C. Volpe, L.J. Wangh, and D.P. King. 2010. Pan-serotypic detection of foot-and-mouth disease virus by RT linear-after-the-exponential PCR. Mol. Cell. Probes 24:250-255. 16.) Sandres-Sauné, K., P. Deny, C. Pasquier, V. Thibaut, G. Duverlie, and J. Izopet. 2003. Determining hepatitis C genotype by analyzing the sequence of the NS5b region. J. Virol. Methods 109:187-193.