Detection of Low-Level K65R Variants in Nucleoside Reverse Transcriptase Inhibitor–Naive Chronic and Acute HIV-1 Subtype C Infections

Abstract

To substantiate reports of greater emergence of the K65R nucleoside reverse transcriptase inhibitor (NRTI) mutation in human immunodeficiency virus type 1 (HIV-1) subtype C, we examined natural low-level K65R expression in subtype C relative to subtypes B and AE. We used allele-specific polymerase chain reaction to screen HIV-1 amplified by reverse-transcription high-fidelity polymerase chain reaction from subtype C–infected South African women and infants and CRF01(subtype AE) from Thailand; all subjects were NRTI naive. We found low-level K65R of unknown clinical significance in NRTI-naive subtype C–infected women and infants at frequencies above the natural occurrence in subtypes B and AE. The frequent appearance of subtype C frameshift deletions at codon 65 supports a propensity for transcription error in this region.

A global scale-up of antiretroviral therapy (ART) and increasing interest in antiretroviral drugs for use in prophylaxis to prevent human immunodeficiency virus (HIV) transmission has raised concern over whether subtype variation may differentially select for drug resistance. A few reports have claimed that subtype C, the predominant HIV-1 clade worldwide, may select resistance to nonnucleoside reverse transcriptase (RT) inhibitors [1] and nucleoside/nucleotide RT inhibitors (NRTIs) (reviewed in [2]) more easily than other HIV subtypes. The drug resistance mutation K65R has received particular attention, because it confers resistance to tenofovir and moderate resistance to lamivudine/emtricitabine, NRTIs much favored for their high potency and tolerability.

In the absence of drug pressure, for a region of the HIV genome to yield a frequency of error higher than what would be expected from random transcriptional inaccuracy, an element of the virus genome may itself compel introduction of the mutations. Possible genetic features that might explain this include nucleotide sequences or folding that causes the polymerase to slip and pause on the template [3]. If indeed particular HIV-1 subtype sequences can more frequently generate RT missteps at sites associated with drug resistance, then drug resistance selection might be greater in ART-experienced areas where those subtypes are prevalent.

We previously evaluated subtype B clinical specimens from the era before the introduction of ART and found that the K65R mutation was not detectable at frequencies >.3% [4]. To obtain evidence of natural K65R quasispecies expression in HIV-1 subtype C at frequencies >.3%, we used sensitive real-time polymerase chain reaction (PCR) assays to screen subtype C–infected NRTI-naive women and their infants who became infected. Mother-infant pairs were further assessed for evidence of K65R vertical transmission. To compare subtype C with an additional subtype, we also screened subtype AE specimens for K65R above the frequency seen with subtype B. Our findings shed light on the differential natural expression of K65R in subtype C HIV-1.

METHODS

Study Populations

To assess the natural frequency of the K65R mutation, we evaluated masked plasma RNA samples from (1) 99 South African ART-naive, chronically subtype C–infected pregnant women before they received intrapartum single-dose nevirapine and (2) 20 8-week-old infected infants born to transmitting mothers and only given single-dose nevirapine at birth [5, 6]. The women were participants in mother-to-child prevention studies in 2001–2005. In 5 infants, a plasma sample was also collected 10 days after birth (n = 25 infant samples). The subtype C results were compared with those from 112 randomized subtype AE–infected, ART-naive participants enrolled in a Thai open-label trial in 1999–2000 (EPV40001; baseline viral loads, 10⁴–10⁶ copies/mL) and with our historical data on 138 pre–ART era subtype B clinical samples collected in 1982–1985 [4]. The Thai specimens were previously available samples kindly provided by ViiV/GlaxoSmithKline for drug resistance assay evaluation. The South African and Thai studies were approved by local institutional review boards, and the subjects provided informed consent for storage of blood for future HIV testing. The institutional review board of the Centers for Disease Control and Prevention approved this study after determining that the recoded, archived specimens were not linked to personally identifiable subjects.

Reverse-Transcription PCR

HIV genomic RNA was extracted from patient blood plasma using the Qiagen UltraSens Virus Kit or Qiagen MagAttract Viral RNA M48 kit (Qiagen M48 BioRobot). The templates for real-time PCR testing were generated by reverse-transcription PCR with murine leukemia virus RT and GeneAmp High Fidelity PCR polymerase (Applied Biosystems), using primers that spanned codons 51–236 of the HIV-B/AE RT [4] and codons 20–272 of the HIV-C RT, as detailed elsewhere [5].

K65R-Specific Real-Time PCR Screening

All samples were screened using the Bio-Rad iCycler plus iQ4 optical unit and real-time PCR–based drug resistance assays [4] for the K65R mutation. The assay discriminates the second position of RT codon 65 for a guanosine transition (A→G), which is indicative of the arginine mutation. The oligonucleotide sequences for the subtype AE K65R assay were as follows: forward primer, HIV-AE_K65R.1F (5′-ATAYAATACTCCARTATTTGCTATAAACAG); reverse primer, HIV-B comREV [4]; FAM-labeled probes, AE Com 3.1P (5′-FAM-TCAGTAACAG“T”ACTAGATGTGGGAGATGCATAT, 80%) and AE Com 3.2P (5′- FAM-TCAGTAACAG“T”ACTGGATGTGGGGGATGCATAT, 20%). The K65R assay oligonucleotides for subtype C were as follows: forward primers, 65R.6F (5′-CAATACTCCAGTATT TGTCATACCAAG, 80%) and HIV-C 65R.5.1F (5′-AACACTCCARTATTTGCYATACCAAG, 20%); reverse primer, HIV-C 65.1REV (5′-TYTTTAACCCTGMTG GGTGTGGTAT); FAM-labeled probes, HIV-C 65.1P (5-FAM-TCAGGGARC“T”C AATAAAAGAACTCAAGACTTYTGGGA, 80%) and HIV-C 65.2P (5-FAM-TCAGG GAAC“T”YAAYAAAAGAACTCAAGACTTYTGGGA, 20%). Quotation marks indicate the nucleotide to which the Black Hole Quencher is attached. K65R-specific and total copy amplification cycles that cross the fluorescence thresholds were compared (ΔCT) to extrapolate the mutant frequency, as previously described. K65R in subtypes AE and C was screened for above the .3% threshold we had previously established as the quasispecies background level for subtype B [4]. Delta CT values of <9 and <11 PCR cycles for the subtype C and subtype AE assays, respectively, corresponded to >.3% K65R. Real-time PCR amplicons from reactions that were positive for K65R were sequenced to confirm the presence of intact mutant sequences, as described elsewhere [4, 7]. In-house bulk genotyping was also performed to determine whether K65R was detectable by conventional testing.

Clonal Confirmation

Clonal sequencing was subsequently performed on specimens that screened positive for K65R by real-time PCR to estimate mutant frequencies; 90–135 clonal sequences of 190 base pairs (RT codons 56–119) per positive specimen were compared with the sample bulk genotype and the K65R-positive real-time amplicon sequences to elucidate mutant subpopulations.

RESULTS

Subtype C K65R Assay Screening

K65R screening of the 99 South African antepartum samples yielded positive results for 6 women (6.0%) (Table 1). The K65R-positive real-time amplicon sequences revealed that for 2 women (SA1-m40 and SA2-m23) the K65R mutation was in frame, as evidenced by the primer 3'-terminus unambiguously recognizing a guanosine in the second position of codon 65 (Supplemental Figure S1). For 2 other women (SA1-m15 and SA1-m32), the mutation-specific amplicon sequence revealed that the primer was correctly binding a second-position guanosine at codon 65, but also a third-position guanosine in the second position, the latter a result of an adenosine missing in codon 65 (−1 frameshift product). For the remaining 2 women with positive screening results for K65R (SA2-m16 and SA2-m19), the amplicon sequence revealed that the assay detected only third-base guanosine frameshifts in codon 65, suggesting that no intact K65R was present.

Of the 99 women evaluated, 20 had HIV-1–infected infants who received only single-dose nevirapine at birth [7]. Screening of the infant plasma obtained at day 10 and week 8 after birth identified 3 infants (SA2-i6, SA2-i19, and SA2-i21) who were positive for K65R at preliminary screening (Table 1). K65R-specific amplicon sequencing predicted that all 3 infants had intact K65R variants as well as the −1 frameshift, suggested by mixed sequences at codon 65. The median ΔCTs for mothers and infants were 11.1 and 10.4 cycles, respectively.

Subtype AE K65R Assay Screening.

Sensitive screening of the 112 ART-naive Thai subjects identified 1 individual (.9%) who was positive for K65R, with a mutation frequency estimated at ∼.5% (ΔCT, 10.5 cycles). The mutation amplicon sequence demonstrated an intact mutation and showed no downstream resistance mutations linked to the K65R.

Bulk sequencing revealed no detectable mutations in the subtype AE and C samples that were positive using real-time PCR screening.

Clonal Screening

K65R clonal screening confirmed intact K65R mutants in 4 (4%) of 99 ART-naive, subtype C–infected South African women (SA1-m15, SA1-m32, SA1-m40, and SA2-m23) at frequencies of .7%–3.3% (Table 1). Clonal sequences surrounding codon 65 are shown in Figure 1. Two women who had intact K65R (SA1-m15 and SA1-m32) also had clones with an adenosine missing from codon 65. The other 2 women (SA2-m16 and m19) who were initially positive at screening had only codon 65 frameshift clones, which were consistent with their screening assay amplicon sequences.

For all 3 HIV-1–infected infants who were screening assay positive (SA2-i6, SA2-i19, and SA2-i21), clonal analysis confirmed intact K65R mutants at frequencies of 2.2%–6.7% (3/20 infants overall; 15%). One infant, SA2-i21, had intact K65R at both the 10-day and 8-week time points. Furthermore, the 3 K65R-positive infants all had codon 65 frameshift mutants. One clone from SA2-i6 had 2 nucleotides missing from codon 65 and 2 missing from codon 66 (−4 frameshift). In no matched pair was intact K65R identified in both mother and infant. The difference in prevalence of K65R between infants (15%) and mothers (4%) did not achieve statistical significance (P = .057; Fishers exact test).

Two of the 3 infants with intact K65R, SA2-i6 and SA2-i21, also had the nonnucleoside RT inhibitor mutations, Y181C and K103N + Y181C, respectively, from the single-dose nevirapine exposure, a finding not known to be connected to the appearance of K65R, as indicated in the Stanford HIV Drug Resistance Database RT Treatment Profile for NVP (http://hivdb. stanford.edu/cgi-bin/RTMutSummary.cgi).

Clonal analysis of the subtype AE–infected subject who had a positive screening result for K65R identified the mutation intact in 1 of 90 clones (1.1%), with no other mutations linked in this K65R clone. We further screened this individual with sensitive testing for other thymidine analog mutations (M41L, K70R, T215 F/Y), and randomly sequenced 45 additional clones from this subject, and could not find evidence of other mutations to indicate transmitted drug resistance or any frameshifts at codon 65.

Phylogenetic analysis of K65R variants to exclude low-level cross-contamination revealed that intact K65R clones from infants formed discrete branches with their source bulk sequences and were distinct from the variants identified in mothers (MEGA 4.1 software; data not shown).

DISCUSSION

We describe the natural appearance of low-level K65R variants in South African NRTI-naive adults and infants with subtype C infections at frequencies higher than those detected for subtype B and AE quasispecies. The absence of evidence for mother-infant K65R transmission supports a spontaneous emergence of K65R that was nearly 4-fold more prevalent in acutely infected infants than in chronically infected women. However, incomplete infant viral load data prevented us from ascertaining whether greater viremia, typically associated with rampant, acute infection, could have contributed to increased K65R expression.

The most striking finding with subtype C viruses was the frequent appearance of frameshift mutations at codon 65, which suggested that the region is a mutagenesis “hot spot.” This transcription error has recently been demonstrated in vitro for subtype C and was suggested to be the possible result of sequence-induced RT slippage and pausing, leading to nucleotide dislocation [8, 9]. An earlier evaluation of the same mothers using an M184V-specific assay found that the M184V quasispecies frequency was very similar to that of subtype B (data not shown), further indicating that the extent of mutations observed around codon 65 was not a result of indiscriminate mutagenesis in subtype C. The only K65R variant identified in the subtype AE population could not be explained by the same dislocation mechanism reported for subtype C [8, 9], because there was no adjacent guanosine in codon 65, and, furthermore, the absence of codon 65 frameshifts also implied a less-volatile genomic region.

Although additional screening did not identify intact K65R clones in mothers of infants with K65R, we did not screen the number of clones that could have compensated for random selection error. Therefore, it is possible that intact variants were present above the target frequency in those mothers but were missed in the random clone analyses. Furthermore, the years sampled for each subtype varied, and unrecognized thymidine analog resistance may have been transmitted in the contemporary subtype C and AE populations. However, the absence of other mutations provided no indication that our results may have been biased by cases of transmitted drug resistance. Although recent reports have indicated that PCR generates spurious low-level K65R with subtype C [10], we found intact variants in infants at frequencies above the reported artifact frequencies. Moreover, if the observed codon 65 errors are indeed template driven, they would be expected to occur with lower-fidelity HIV RT in vivo.

The data suggest that subtype C generates a higher quasispecies frequency of codon 65 mutations. This finding may help explain what has been reported as greater selection of K65R in subtype C–infected subjects receiving ART, mostly from the use of stavudine [2, 11–13]. However, longitudinal studies of K65R at baseline and during therapy are needed to ascribe clinical significance to low-level quasispecies variants. The detection of subtype-specific spontaneous frameshift variants may confound resistance screening; thus, data from point mutation testing must be analyzed carefully.

Funding

This work was supported entirely by intramural funds from the Centers for Disease Control and Prevention.

References

Author notes