|Home | About | Journals | Submit | Contact Us | Français|
We report a real-time PCR assay capable of detecting drug-resistant human immunodeficiency virus type 1 reverse transcriptase K65R mutant virus at a level of 0.5% in polymorphic patient plasma specimens. Fifty-three treatment-naïve and 20 treatment-experienced specimens were successfully genotyped with the new method. Results were in agreement with population sequencing and the labor-intensive single-genome sequencing method.
Human immunodeficiency virus type 1 (HIV-1) exhibits tremendous genetic variation, allowing virus to escape host immune response and develop drug resistance to antiretroviral therapeutics (1). To manage HIV-1 infection, genotypic and phenotypic methods are widely used to monitor drug-resistant viruses within infected individuals. Standard methods determine the majority genotype or phenotype present in the plasma virus population but have limited ability to detect minor subpopulations (<25%) of drug-resistant virus (9, 12, 18). Minor subpopulations of resistant virus can arise during antiretroviral therapy, from transmission of drug-resistant virus, and from discontinuation of a drug treatment (2, 15, 16). The clinical relevance of minor resistant subpopulations on treatment outcome has not been well established due, in part, to a limited ability to detect and quantify such subpopulations. Single-genome sequencing is an ultrasensitive method for identifying the presence of resistant subpopulations, but it is highly labor intensive (17). More recently, a series of less-demanding yet sensitive approaches have been reported, including a line-probe assay based on hybridization to probes, real-time PCR-based assays, and phenotypic detection utilizing a hybrid element of retrotransposon TY1 and HIV-1 reverse transcriptase (RT) (6, 8, 11, 14, 15, 20, 22). Although these assays demonstrated the ability to detect several drug resistance mutations, none were reported to detect the K65R mutation of HIV-1 RT from clinical patient specimens.
The K65R substitution in HIV-1 RT is the primary mutation known to be selected by tenofovir disoproxil fumarate, the oral prodrug of tenofovir, resulting in a two- to fourfold reduction in susceptibility to this drug (10). K65R is also selected by abacavir, didanosine, stavudine, and zalcitabine (4, 5, 7, 21, 23, 24). Despite broad usage of didanosine, abacavir, and tenofovir disoproxil fumarate, detection of the K65R mutation with standard methods for treatment-experienced patients remains infrequent (3, 19). To allow the study of minority K65R subpopulations we developed a highly sensitive assay for detection of K65R RT mutant HIV-1 viruses from clinical plasma samples.
Recently a real-time PCR technology called MultiCode-RTx (EraGen Biosciences) was reported that quantitated M184V and K65R HIV-1 RT mutants down to a level of 0.01% in plasmid DNA mixtures (13). However, likely due to interfering polymorphisms, we found this method unable to amplify 6 of 12 isolates from a multidrug-resistant plasmid panel (AIDS Research and Reagent Program; data not shown). We now report a modification to the MultiCode protocol allowing rapid detection of minority K65R mutant populations in clinical samples. The new protocol for detection of K65R can utilize viral RNA or amplified RT-PCR products (Fig. (Fig.11 and Table Table1).1). The key modification was inclusion of a third “curative” forward primer with a high annealing temperature terminating one base upstream of the K65R mutation site. Several initial PCR cycles at an annealing temperature above that of the allele-specific forward primer pair can amplify a cured target with identity to the allele-specific primers. The curative primer tolerates polymorphic mismatches while still allowing for target PCR amplification. Subsequent allele-specific real-time PCR cycles with a two-stage annealing temperature are then used (Fig. (Fig.2B)2B) as previously described (13). Incorporation was monitored using a LightCycler 1.2 instrument (Roche) with two channels, one for each allele.
To construct a standard curve for quantitation, K65K and K65R viral stocks of NL4-3 were generated from MT-2 cells. HIV RNA content of the viral stocks was determined by the Amplicor HIV-1 Monitor assay, version 1.5 (Roche). Viral stocks at 106 copies/ml were mixed to obtain standard curve mixtures ranging from 0.1% to 50% K65R. RNA isolated from the standards was used directly in the MultiCode-RTx assay (Fig. (Fig.1C).1C). Assay background and detection limits were established by multiple measurements of standard curve samples. For one-step RT-PCR starting directly from RNA, the detection limit was 0.1% for K65R (Fig. (Fig.2A)2A) and was limited by the paucity of target molecules (~15 copies). To determine the sensitivity of detection of K65R in samples with lower viral loads, viral standard curve mixtures were diluted to 105 copies/ml and then RT-PCR amplified using the primers and PCR conditions listed in Table Table1.1. When amplified RT-PCR products were used as the target for the MultiCode-RTx assay, the limit of detection for K65R was 0.5% (Fig. (Fig.2B)2B) due to increased assay background in the 0% K65R control. Of note, the assay conditions described here are applicable to subtype B RT sequences. We found that presence of an AAG polymorphism at codon 65 that is typical for subtype C can result in higher background and requires redesign of allele-specific primers (data not shown).
Fifty-three plasma samples from treatment-naïve HIV-1-infected individuals with no known resistance mutations by population sequencing were tested for the presence of low-level K65R. Fourteen samples with HIV RNA values ranging from 0.5 × 106 to 1.4 × 106 copies/ml were directly subjected to the analysis (Table (Table22 and Fig. Fig.2A).2A). Additionally, amplified RT-PCR products from 39 treatment-naïve patient samples with a large range of viral loads were analyzed for the presence of K65R by the MultiCode-RTx assay. Among these 53 treatment-naïve patient samples, no K65R was detected above the background amplification cutoff of 0.5% (Table (Table22 and Fig. Fig.2B2B).
Twenty clinical samples from 13 treatment-experienced patients with viral loads between 112 and 26,841 copies/ml were also tested for K65R (Table (Table2).2). Seven of 20 samples contained a polymorphic mutation at the binding site for the allele-specific primer; three of those changes resulted in an A62V substitution commonly associated with K65R. Detection of a full K65R genotype by standard sequencing was reproduced in 11 of 11 samples. One sample containing a mixture of K65R and K65K according to the results of standard sequencing was confirmed by the MultiCode assay to contain 60% K65R. In addition, we tested seven samples that may have contained subpopulations of K65R due to treatment history but were shown to be wild type by standard population sequencing. Results indicated subpopulations of K65R in four of those samples ranging from 1.4% to 25% (Table (Table22 and Fig. Fig.2B).2B). One of these patients (patient 3) developed a full K65R mutation at a subsequent time point confirmed by both the MultiCode-RTx assay and population sequencing. To confirm detection of K65R subpopulations, clonal analysis was performed for three patients. K65R was present in 2 of 77 (3%) and 4 of 63 (6%) clones analyzed by single-genome sequencing for sample 2 and 3a, respectively, which agreed with estimates of 1.4% and 11.1%, respectively (Table (Table2).2). Clonal sequencing also confirmed the presence of K65R mutant viruses in patient 4, with 47 of 93 clones (50%) harboring the K65R mutation and with the assay indicating 25% K65R. Overall, the clonal sequencing and the methodology yielded estimates of subpopulation frequencies that were in agreement within a factor of approximately twofold. In conclusion, the improved MultiCode-RTx real-time PCR methodology described here successfully detects minor subpopulations of K65R in clinical specimens with low viral load and polymorphic sites from treatment-experienced subtype B patients.
We thank Derrick Goodman, Joshua Waters, Jeanette Harris, Brandi Chappell, Florence Myrick, Nicolas Margot, and Damian Mccoll for technical support and intellectual input.
Published ahead of print on 27 September 2006.