|Home | About | Journals | Submit | Contact Us | Français|
In order to survive prolonged treatment with antiretroviral nucleoside analogs, the human immunodeficiency virus type 1 (HIV-1) is selectively forced to acquire mutations in the reverse transcriptase (RT) gene. Some of these mutations are more common than others and have become markers for antiretroviral resistance. For the early detection of these markers, a novel MultiCode-RTx one-step testing system to rapidly and simultaneously characterize mixtures of HIV-1 targets was designed. For cDNA, nucleotide polymorphisms for codon M184V (ATG to GTG) and K65R (AAA to AGA) could be differentiated and quantified even when the population mixture varied as much as 1 to 10,000. Standard mixed-population curves using 1 to 100% of the mutant or wild type generated over 4 logs of total viral particle input did not affect the overall curves, making the method robust. The system was also applied to a small set of samples extracted from infected individuals on nucleoside reverse transcriptase inhibitor therapy. Of 13 samples tested, all were positive for HIV and 10 of the 13 genotypes determined were concordant with the line probe assay. MultiCode-RTx could be applied to other drug-selected mutations in the viral genome or for applications where single-base changes in DNA or RNA occur at frequencies reaching 0.01% to 1%, respectively.
Reverse transcriptase (RT) inhibitors such as the nucleoside analogs (−)-(1S,4R)-4-[2-amino-6-(cyclopropylamino)-9H-purin-9-yl]-2-cyclopentene-1-methanol (abacavir; ABC), 9-[2-(phosphonomethoxy)propyl]adenine disoproxil fumarate (tenofovir; TDF), (−)-β-2′,3′-dideoxy-5-fluoro-3′-thiacytidine [emtricitabine; (−)-FTC], 3′-azido-3′-deoxythymidine (zidovudine; AZT), 2′,3′-dideoxyinosine (didanosine; ddI), 2′,3′-dideoxycytidine (ddC; zalcitabine), (−)-β-2′,3′-dideoxy-3′-thiacytidine (lamivudine; 3TC), and 2′,3′-didehydro-3′-deoxythymidine (stavudine; d4T) are currently approved for the treatment of advanced human immunodeficiency virus type 1 (HIV-1) infections (8). All of these nucleosides act in a similar way, namely, as chain terminators of the RT reaction after phosphorylation by intracellular kinases (3, 17). Treatment of HIV infection with these nucleoside inhibitors invariably leads to selection and accumulation of drug-resistant mutations that arise due to the intrinsically low fidelity of the polymerase (11). Perhaps the most well known of a very large number of mutations observed in HIV RT is HIV M184V (19, 23). A detailed overview of genotypic resistance profiles and their corresponding phenotypic resistance consequences is available (18). Determining how fast resistant mutants replicate in the presence of a drug or drug combination (replication capacity), identifying combinations of drugs that suppress HIV-1 replication in a way that minimizes emergence of resistance, tracking HIV-1 population makeup in various patient fluids, and determining and later predicting drug response through mutant analysis and mutant population percentages are all important issues to HIV-1 inhibitor development. Additionally, early detection of mutants in subjects infected with HIV-1 may reduce the likelihood of treatment failure by helping to change treatment before the problem is magnified for individuals undergoing salvage regimens. Despite this, there is still no good method for the early detection of these emerging drug-resistant mutants.
To this end, a recently developed technology system called MultiCode-RTx was used to quantify drug-resistant mutations within the HIV-1 pol gene in mixed populations (21). This new system exploits the use of an additional base pair made from 2′-deoxy-5-methyl-isocytidine (iC) and 2′-deoxy-isoguanosine (iG) to site specifically incorporate a quencher in close proximity to a fluorescent molecule (Fig. (Fig.1).1). Prior to running the RTx systems for mixed-population analysis, two target-specific forward PCR primers carrying single iC bases near distinct 5′-fluorescent reporters and a single reverse primer able to prime both targets are constructed. Using a commercially available reaction mixture containing iGTP-dabcyl, iC directs specific enzymatic incorporation of the iGTP-dabcyl in close proximity to each fluorophore. This incorporation reduces the fluorescence of reporter attached to the extended primers and is monitored using standard real-time PCR instrumentation. As the reaction proceeds, the instrument collects data in two channels (each target is analyzed using a distinct fluorophore and data are collected in distinct channels). As more and more of the labeled primers are used up, the fluorescence signal specific for that primer goes down. The PCR cycle at which the fluorescent passes below a determined threshold correlates to the number of initial target molecules present. The cycle threshold (Ct), in which fluorescence passes the threshold within each channel, is dependent on the copy number for each specified target. Standard curves constructed from Ct data from known concentrations of each target are used to determine concentrations within unknown samples. Additionally, the reaction can be analyzed for correct product formation after cycling is complete by melting the amplicons and determining their melting temperatures (Tm). This melting analysis can be used to verify that the anticipated amplicon was created.
To demonstrate that the RTx system is specific enough to quantitate mixed populations of two targets that differ by a single base, two assays that target two important single-base mutations (M184V and K65R) within the HIV-1 gene encoding RT were developed. These mutations are primarily associated with L-nucleosides such as 3TC and TDF, respectively (14, 20). Testing standards which varied in percent wild-type to mutant and total concentration were constructed using known quantities of both the wild-type and mutant targets. Using target-specific forward primers that differ in fluorophore and 3′-nucleotide, we report that quantities of both nucleic acid targets can be determined in tandem even when either target is in vast excess.
Proviral clone pNL4-3 (contributed by M. Martin) was obtained from the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health. The K65R and M184V mutations were introduced into a subclone of a region of pNL4-3 by using the QuikChange kit (Stratagene, La Jolla, CA) following the manufacturer's instructions to make single-base mutations in the subclone of pNL4-3. The sequence-validated mutant subclones were transferred back into the full-length pNL4-3 using standard cloning methods. Wild-type, K65R, and M184V HIV-1 sequences corresponding to bp 2300 to 3285 of pNL4-3 (GenBank accession no. M19921) were amplified from pNL4-3 plasmid DNA and mutant derivatives under standard PCR conditions. PCR products of 985 bp were cloned into pGEM-T vector (Promega, Madison, WI) using the T-A cloning method. Constructs and orientation were verified by DNA sequencing. Plasmids were isolated using Plasmid Mini kits (QIAGEN, Valencia, CA). Plasmid concentration was determined by optical density at 260 nm (OD260) using an extinction coefficient of 50 μg/ml per OD unit and used as targets and as runoff templates. Plasmids were linearized immediately 3′ of the HIV-1 insert using SalI (New England Biolabs, Beverly, MA). Following phenol extraction and ethanol precipitation, 1 μg of each linearized plasmid was transcribed into RNA using the Ampliscribe T7 transcription kit (Epicentre, Madison, WI) following the manufacturer's instructions. Transcription reactions were treated with RNase-free DNase to degrade plasmid template and then phenol extracted. Free nucleotides and pyrophosphate were removed by size exclusion chromatography using NAP5 columns (Amersham) and following the manufacturer's instructions. Following lithium chloride ethanol precipitation, RNA was treated again with DNase and then completely desalted with a second round of chromatography. The RNA concentration was determined by OD260 and an extinction coefficient of 0.117 mM RNA nucleotide per OD (Maniatis). RNA and plasmid DNA targets were diluted in 10 mM MOPS (morpholineethanesulfonic acid, pH 7.4), 0.1 mM EDTA.
Plasma samples were taken from HIV-1-infected individuals and were stored at −70°C until use. Subjects were treated with AZT, ddI, ddC, d4T, 3TC, or several combinations of these drugs. HIV-1 viral RNA was extracted from plasma samples using the Mini Viral RNA kit (QIAGEN). Leucopacs were obtained from Advance Biotechnologies, Inc. (ABI; Columbia, MD) and peripheral blood mononuclear (PBM) cells were separated using Ficoll-Hypaque density gradient centrifugation. Cells were stimulated with phytohemagglutinin (PHA; 2 g/ml) in RPMI 1640, supplemented with 20% heat-inactivated fetal bovine serum, recombinant human interleukin-2 (IL-2; 26 IU/ml), 250 U/ml penicillin, 250 μg/ml streptomycin, and 2 mM glutamine for 48 to 72 h before study. PHA-stimulated PBM cells (5 × 106) were transfected with 5 to 10 μg of wild-type and mutant plasmid DNA by electroporation. Both pure wild-type and mutant plasmids were used for transfection. Mixtures of wild-type and resistant pNL4-3 derivatives were also employed. Successful infection was confirmed by use of HIV-1 antigen detection. RNA from cell culture supernatants was extracted using the Mini Viral RNA kit. Extracted RNA was transcribed into cDNA using standard methods. These RNA samples were quantified for each genotype using RTx genotyping systems.
Forward and reverse primers were designed against plasmid pNL4-3 sequence from GenBank accession no. M19921 using Visual OMP software (DNA Software, MI) and set to have target specific Tms of 50°C and 60°C, respectively. Single-base changes in the forward primers were introduced when testing patient samples to better reflect B-clade consensus. The forward primers contain 5′-end target-independent tails which increase the Tm by ~10°C to allow touchdown PCR cycling (2). Forward primers also contain single 5′ iC and separable fluorophores 6-carboxyfluorescein (FAM) and hexachlorofluorescein (HEX). Primer sequences are given in Table Table11.
For each assay, PCR primers were used at the following concentrations: forward specific primers at 200 nM for M184M, 150 nM for M184V, 100 nM for K65K, and 100 nM for K65R; and reverse specific primers all at 400 nM. The PCR conditions involved 25-μl reaction mixtures in 1× ISOlution buffer (EraGen, Madison, WI) and Titanium Taq DNA polymerase (Clontech, Palo Alto, CA) at the manufacturer's recommended concentration. For one-step MultiCode-RTx reverse transcriptase PCR assays, the conditions included the following: 0.5 U/μl Maloney murine reverse transcriptase and 5 mM dithiothreitol, with an additional 5 min at 50°C added prior to the denaturation step.
Three PCR cycling parameters with ramp rates of 20°C/s unless otherwise specified on the Roche LightCycler 1 (Roche, Indianapolis) were used as follows. Parameter 1 involved 2 min at 95°C; 1 cycle of 1 s at 95°C, 1 s at 45°C, and a 1°C/s ramp to 20 s at 72°C; 50 cycles of 5 s at 95°C, 5 s at 55°C, and a 1°C/s ramp to 20 s at 72°C (single read); and melt at 60 to 95°C with a 0.4°C/s ramp (step read). Parameter 2 involved 2 min at 95°C; 1 cycle of 1 s at 95°C, 1 s at 45°C, and 20 s at 72°C; 50 cycles of 5 s at 95°C, 5 s at 55°C, and a 1°C/s ramp to 20 s at 72°C (single read); and melt at 60 to 95°C with a 0.4°C/s ramp (step read). Finally, parameter 3 involved 2 min at 95°C; 1 cycle of 1 s at 95°C, 1 s at 45°C, and 20 s at 72°C; 100 cycles of 1 s at 95°C and 1 s at 55°C to 20 s at 72°C (single read); and melt at 60 to 95°C with a 0.4°C/s ramp (step read).
The PCR cycling parameters on the ABI Prism 7900 (ABI, Foster City, Calif.) real-time thermal cycler were 2 min of denaturing at 95°C and 1 cycle of 5 s at 95°C, 5 s at 45°C, and 20 s at 72°C, followed by 45 cycles of 5 s at 95°C, 5 s at 60°C, and 20 s at 72°C with optical read. A thermal melt at a 7% ramp rate with optical read from 60 to 95°C was performed directly following the last 72°C step of thermal cycling.
To obtain quantitation curves and to curve fit unknowns, two channel fluorescence data were exported as text files from the Roche LightCycler-1 analysis software (version 5.32) or from the ABI 7900 Prism SDS software (version 2.1) and then analyzed with MultiCode-RTx analysis software (EraGen Biosciences, Inc., Madison, WI). The RTx software imports data from most commercially available real-time instruments and performs cycle threshold and melt curve analyses to determine signal decrease during the amplification and signal change during the melt. An Excel (Microsoft, Redmond, WA) spreadsheet was used to analyze and produce ΔCts.
Two RTx assays targeting two key HIV-1 pol gene mutations (M184V and K65R) were designed and tested. To build specificity into the assays, the competitive forward primer sets (labeled with different fluorophores and 3′-end base compositions) contained additional target-independent regions to allow the annealing temperatures to be raised by 10 to 15°C after the first round of amplification. Standard reverse primers were constructed to minimize amplification artifacts as well as to compensate for other known mutations that lie within the priming region. In total, seven M184 forward, six K65 forward, and two reverse designs were constructed. The best combinations as determined by sensitivity and specificity criteria were chosen. The sizes of the amplicons generated were 121 bases for K65 and 103 bases for M184. The designed assays were then tested using mixtures of cloned DNA sequences and transcribed RNA templates.
To test for assay robustness, the M184V- and K65R-specific assays were performed using four cycling conditions on two instruments as described in Materials and Methods. Mixtures of cloned DNA targets were tested to obtain Ct curves. We used 10-fold increases and decreases of both mutant and wild-type DNA targets from 103 to 107 copies per reaction with a total of 107 copies per reaction. This setup provided mixtures that varied in wild-type and mutant targets from 0.01 to 100%. We then determined linear regression analyses for individual channels of log[copy number] versus Ct. Each condition for M184 afforded tight r2 values which varied from 0.991 to 0.998 for the wild-type M184M channel and 0.990 to 0.997 for the mutant M184V channel (Table (Table2).2). Linear regressions for the K65 system had r2 values, which varied from 0.949 to 0.995 for the wild-type K65K channel and 0.977 to 0.994 for the mutant K65R channel. By measuring the difference between the Ct channels of wild-type and mutant (defined as ΔCt), ΔCt standard curves were established for each condition (Fig. (Fig.2).2). These could be used to determine the makeup of unknown sample mixtures.
RNA target molecules of runoff transcripts were mixed in ratios from 1 part M184V in 100 (1% M184V) M184M to 1 part M184M in 100 M184V (1% M184M) at a total concentration of 107. The RNA mixtures were analyzed using a one-step MultiCode-RTx setup and performed on the ABI 7900 and monitored simultaneously for FAM and HEX. After 45 rounds of PCR we were able to differentiate either target when in mixed ratios of 1.0 to 100% (Fig. 3A, B). The specificity of the method is exemplified by the ability to differentiate between pure samples and the samples containing 1% of the differing genotype. For the mutant-specific M184V assay, the difference in Ct values between 0 and 1% was 3 cycles. This specificity is at least 1 or 2 orders poorer when compared to the assays above where DNA was directly targeted. The Ct difference could not be determined for the wild-type assay as the drop in fluorescence did not cross the threshold for the pure M184V sample.
To demonstrate that the RTx method could successfully genotype full-length viral RNA, PBM cells were transfected with pNL4-3 wild-type and pNL4-3 M184V proviral DNA. Extracted RNA mixtures of wild-type and M184V RNA were prepared and analyzed in the same manner as stated above for runoff transcripts. The results were comparable to those obtained with the T7 RNA transcript targets (Fig. (Fig.4).4). Quantitation curves demonstrate that even when mixed populations of RNA isolated from cultured HIV-1 are used for targets, the assay continues to be quantitative, with r2 values approaching 1.0. Amplification was observed in one of the no-target control (NTC) reactions but was dropped from consideration due to a spurious melt profile.
The method can discriminate mixed populations even through a large range in overall concentration. To demonstrate this, total RNA concentrations of M184 from 103 copies to 107 copies were varied using mixtures of targets from 1% to 99% M184V and RTx analysis was again performed. We constructed the ΔCt curves for each of the series and overlaid the results. The results indicate that the RTx system allows for percentages to be determined even as the overall concentration changes over 4 orders of magnitude and suggests that quantitation of the mixed viral populations may not need to be determined prior to mixed-population analysis (Fig. (Fig.55).
To demonstrate the ability of the MultiCode-RTx method to determine the drug-resistant genotype of viral populations from human plasma, viral RNA extracted from 13 plasma samples containing HIV-1 and previously genotyped by the line probe assay (LiPA) were used (22). These samples were from a cohort of subjects undergoing NRTI therapy between the years 1994 and 1997 which displayed either M184M, M184V, or a mixed population of M and V by LiPA. The extracted RNA was tested in duplicate using the M184V system as described above. RTx was able to amplify and genotype all samples (Table (Table3).3). RTx results from 10 of the samples were completely concordant with LiPA. Of the three that were not 100% concordant, two samples displayed a mixture of M and V where LiPA only detected V. For the third sample, the opposite was true: RTx showed 100% M while LiPA displayed a mixture.
We have introduced an assay that uses MultiCode-RTx genotyping technology for the simultaneous analysis of mixed targets that differ by a single nucleotide. The approach combines a type of competitive allele-specific PCR and an additional base pair. The additional base pair allows for site-specific quencher incorporation opposite a variety of fluorophores in order to simultaneously observe fluorescent change in multiple channels over the time course of the assay. The number of PCR cycles required to observe signal change below a threshold is proportional to target concentration. Thus, PCR cycle threshold data can be used to develop quantitation curves and determine the levels of targets in unknown samples.
Using this method, we demonstrated the ability to quantitate mixed populations of nucleic acid targets specific to HIV-1 and drug-resistant variants of HIV-1 that differ by a single nucleotide change, even when the minority species is present at 1 part in 10,000. The method was tested on two of the most important drug selected mutations, M184V and K65R. The ability to quantitate the K65R mutation even when it is a minor species is of particular importance since most of the nucleoside and nucleotide inhibitors select for this mutation. The method was also shown to be extremely robust. Data using four different cycling parameters for both assays demonstrated high levels of specificity. In addition when the total concentration of targets fluctuated from 103 to 107, the difference in Ct values for any given mixed population stayed constant. Combined, the results indicate that the method may have uses in a wide variety of applications.
These preliminary results led us to test the RTx method on a set of 13 samples from HIV-1-infected individuals. The results were evaluated by comparison to those previously determined using LiPA. Ten of the RTx results showed 100% concordance with LiPA, while 3 showed some minor inconsistencies in population percentages. Reasons for these population differences are unclear, since the level of mixtures should be within the range of both methodologies. Further studies are under way to perform single-genome sequencing on the samples.
Other methods for biallelic polymorphism testing from pooled samples exist. The first was discussed by Kwok et al. (13). Since then, a large number of other techniques have been proposed to further simplify mixed-population studies—some more complicated then others (1, 4, 5, 7, 16, 22, 25, 28, 29). But easy-to-use and ultraspecific methodologies do not seem to be available for mixed-population quantitative testing. One method called “needle-in-a-haystack” exceeds the specificity described here with sensitivities down to 1 in 106 yet requires complicated multiple steps (27). Other real-time PCR methods that are easy to use such as SYBR green have been used for mixed-population analysis (4). However, SYBR green methodology does not allow for the simultaneous real-time detection of multiple targets which limits the specificity. For example, SNP allele frequency was measured be dividing the pools between two PCR each containing a primer pair specific to one or the other variant. Mismatch amplification under these conditions is typically delayed by 10 cycles. Mixtures of 19:1 could be identified using the multiwell approach. This level in mixture detection range may be due to well-to-well variation, lack of competition between primers for the target, or reaction conditions. There are other techniques that enable differential fluorescent labeling of the primers (15, 26). These should also allow for mixed-population genotyping, yet to our knowledge there are no published reports that demonstrate specificity near what is described for the RTx method. Other techniques employing amplicon-specific hybridization probes are well entrenched within the science community (10, 24). Ironically this is due to the perception that probes are needed to gain specificity. Yet no literature reports showing single-base specificity in mixed populations below the standard 50:50 SNP studies could be found. This is not to say that these probe-based systems could not be used in such studies, but the limits placed on hybridization methods may make this difficult (12). This being said, an approach which combines allele-specific PCR and the TaqMan real-time probe hybridization system was able to detect a single base change in a mutant DNA target in a sample with 1,000-fold-greater wild-type targets. Yet both targets (wild type and mutant) could not be detected simultaneously since the hybridization probe was not discriminatory and ratios of the two species were not determined (6). Previously, our group has used a method called LiPA to analyze mixed populations of HIV-1 (9, 22). LiPA, which requires post-PCR handling, has the benefit of analyzing multiple targets simultaneously. Yet we found that LiPA was unable to detect the K65R target and could not reproducibly detect subpopulations below 4%. The MultiCode-RTx system is a faster method and allows for a higher level of analytical specificity. In addition, we have shown that the RTx system can quantitate both wild-type and mutant populations. The LiPA system is a solid-phase post-PCR methodology and therefore is semiquantitative.
The MultiCode-RTx method presented is not a substitute for sequencing of course. Since only one site can be analyzed per assay, the RTx method is not appropriate for scanning entire genomes and the mutation tested must be previously known. The method also cannot determine the linkage of mutations present at low frequency on the same genome as heteroduplex tracking assays can. Yet RTx should be useful in a wide variety of other applications, such as early detection of emerging drug-resistant strains, determining allele frequency of biallelic polymorphisms in pooled samples, and early detection of cancer in patients. Our results show that RTx may have utility in drug development or clinical testing when it relates to HIV resistant strain emergence. It should be noted, however, that the primers and targets used to construct the results did not take into account the diversity of all the HIV targets that could be found in a clinical setting. Efforts are under way to address the diversity issue using two approaches. The first employs primers that bind noncontiguous regions of the HIV genome. This approach allows highly polymorphic regions to be bypassed. At lower temperatures where hybridization of noncontiguous regions to the target occurs, primer extension can take place. After duplex formation, priming sites that are identical to the primer are formed. In subsequent rounds of PCR, the annealing temperature is increased to the Tm of the primers and exponential amplification begins. The second and more standard approach is introduction of degenerate bases; particularly at the third position of the amino acid codon. The specificity of either approach still needs to be determined using larger sample numbers, but the data so far show promise for RTx HIV-1 mixed-population analysis.
We would like to thank the National Institutes of Health for financial support through the STTR grant AI058888, Emory University's NIH CFAR grant 5P30-AI50409, and the Department of Veterans Affairs.