|Home | About | Journals | Submit | Contact Us | Français|
The emergence of human immunodeficiency virus type 1 resistance to raltegravir, an integrase strand transfer inhibitor, follows distinct and independent genetic pathways, among which the N155H and Q148HKR pathways are the most frequently encountered in treated patients. After prolonged viral escape, mutants of the N155H pathway are replaced by mutants of the Q148HKR pathway. We have examined the mechanisms driving this evolutionary pattern using an approach that assesses the selective advantage of site-directed mutant viruses as a function of drug concentration. These selective-advantage curves revealed that among single mutants, N155H had the highest and the widest (1 to 500 nM) selective-advantage profile. Despite the higher 50% inhibitory concentration, Q148H displayed a lower and narrower (10 to 100 nM) selective-advantage profile. Among double mutants, the highest and widest selective-advantage profile was seen with G140S+Q148H. This finding likely explains why N155H can be selected early in the course of RAL resistance evolution in vivo but is later replaced by genotypes that include Q148HKR.
Integrase strand transfer inhibitors (INSTI) are a novel class of human immunodeficiency virus (HIV)-specific antiretroviral drugs with potent antiviral activity in vivo in HIV-infected patients (4, 9, 14, 15). This class of drugs has proven particularly useful in treating patients who have failed multiple previous regimens of antiretroviral treatment and who are infected with viral strains expressing wide resistance to multiple other antiretroviral drugs (8, 21). In such patients, however, the association of the INSTI with a background of drugs from other classes that do not retain sufficient antiviral potency to permit full viral suppression by the combination has led to the selection of INSTI-resistant viruses carrying characteristic mutations in the integrase coding region of the HIV-1 pol gene (2, 3). The development of such resistance has been best studied for raltegravir, which is currently the only INSTI that has been approved for clinical use. Virological studies conducted in HIV-1-infected patients developing resistance to raltegravir have shown that resistance to this drug can follow three main genetically distinct mutational pathways (6, 11, 17). The N155 pathway is based on emergence of mutation N155H, which can be followed by selection of other integrase mutations such as E92Q. The Q148 pathway starts by selection of mutations Q148H, Q148R, or Q148K, most often followed by addition of mutation G140S. The third pathway, the Y143 pathway, is characterized by selection of mutation Y143R, frequently followed by the addition of mutations T97A or E92Q. Longitudinal studies in patients developing high-level resistance to raltegravir have led to several important observations. (i) The N155 pathway is generally observed early in the evolution of resistance and, after several weeks of HIV-1 evolution under pharmacological pressure by raltegravir, is replaced in most instances by emergence of mutations belonging to the Q148 or to the Y143 pathway (6, 17). (ii) Clonal analyses of viral quasispecies present in patients failing raltegravir have shown that viral genomes carrying mutations from different resistance pathways can coexist and coevolve in parallel in the same patient (6, 7, 11). (iii) Finally, these studies have revealed that the three raltegravir resistance pathways are mutually exclusive, since they are always found to evolve on separate viral genomes (6, 11). The mechanisms explaining the dynamics of emergence of resistance to raltegravir along individual mutational pathways and the absence of overlap between these pathways are not fully understood. In the present study, we examined the individual and combined impact of raltegravir resistance mutations belonging to the N155 or to the Q148 pathways on resistance levels, on RC, and on the selective advantage expressed by mutants relative to wild-type virus according to the concentration of raltegravir in the culture.
Seven NL4-3-based single or double integrase mutants were constructed by site-directed mutagenesis. These mutants included the single mutants N155H, Q148H, G140S, and E92Q and the double mutants N155H+E92Q, Q148H+140S, and Q148H+N155H. The mutants were first tested for raltegravir resistance using a previously described single-cycle assay based on transfection of 293T cells by replication-competent infectious molecular clones, infection of raltegravir-treated P4 indicator target cells, and quantification of Tat-induced β-galactosidase activity in P4 cells by a chlorophenol red-β-d-galactopyranoside (CPRG; Roche) colorimetric assay (1, 16, 20). Resistance to raltegravir was expressed as the fold change in the 50% inhibitory concentration (IC50) relative to wild-type NL4-3. The experiments were repeated at least three times with viral supernatants from independent transfections. The IC50 values are reported on Table Table11 and the corresponding inhibition curves on Fig. Fig.1.1. Among single mutants, Q148H expressed the highest level of resistance with a mean 78-fold change in IC50 relative to NL4-3. The fold change in IC50 for mutant N155H was slightly lower with a mean of 31.5-fold. Neither G140S nor E92Q expressed significant increases in IC50, with fold changes of 0.7 and 3.7, respectively. Among double mutants, the fold change values were 492-fold for the N155H+E92Q mutant and 1,436-fold for the Q148H+140S mutant. Surprisingly, although the association of these two mutations are never encountered on the same viral genome in vivo, the combination of Q148H+N155H resulted in a virus that was infectious and displayed a level of resistance that could not be measured within the range of raltegravir concentrations used in our assay, which extended to 25,000 nM.
The same mutants were next tested for their replicative capacity (RC) in the absence of drug, using a previously described P4-based single-cycle assay (12, 13, 22). P4 cells were infected with serial dilutions of transfection supernatant from each clone after normalization for HIV-1 p24 content. The expression of β-galactosidase in P4 cells, measured using the CPRG colorimetric assay, was then plotted against p24 concentration in the inoculum and slopes were calculated by linear regression (18). RC was expressed as the ratio of the slope calculated for each mutant virus to that of reference NL4-3. Again, these experiments were repeated independently at least three times. The mean RC values for each mutant are shown on Table Table1.1. Among single mutants, three viruses displayed limited loss of RC relative to wild-type NL4-3: N155H (mean RC = 59.9%), E92Q (mean RC = 70.1%), and G140S (mean RC = 61.3%), whereas Q148H displayed the most significant reduction in RC with a mean of 28.0%. Analyses of the RC of double mutants revealed that addition of G140S to Q148H restored RC to values comparable to those for the wild type (mean RC for Q148H+140S = 90.5%). In contrast, the addition of E92Q to N155H reduced RC to a mean value of 33.6%, which is lower than that of each of the two single mutants involved. Although the combination of Q148H+N155H resulted in an infectious and highly resistant virus, the RC of this mutant was markedly reduced at 14.4% of the wild type.
While the IC50 of a mutant virus only measures the level of inhibition exerted by a given antiviral drug relative to the same virus in the absence of drug, regardless of the extent that this virus is able to replicate and compete with other variants in the same conditions, the virus RC value only assesses its ability to replicate in the absence of drug. None of these two parameters is fully able to reflect the selective advantage expressed by a mutant virus relative to the wild type according to the concentration of drug. Therefore, to better understand the selective forces driving the dynamics of raltegravir resistance pathways in treated patients, we elected to evaluate the selective-advantage profiles of each of the integrase mutants described above by using the method previously described by Mammano and coworkers for protease inhibitors (13, 19). In these experiments, P4 cells were treated with a wide range of raltegravir concentrations and inoculated with virus suspensions that have been carefully normalized for HIV-1 p24 content. Each mutant was tested in parallel with reference NL4-3 on the same microwell culture plate, ensuring that viral infectivity in the presence of different concentrations of the drug were measured by CPRG colorimetry in the same conditions for the mutant and the wild type. We then calculated a “selective index,” representing the ratio of the optical density for the mutant relative to that measured for the wild type at each concentration of raltegravir. Mean selective index values from at least three independent experiments were then plotted against raltegravir concentrations to generate the curves presented on Fig. Fig.2.2. Any selective index value above 1 is an indication that at the corresponding concentration of raltegravir, the studied mutant has a positive replicative advantage compared to wild-type. As shown on Fig. Fig.2A,2A, on which selective index curves for all four single mutants are plotted, the only single mutant with a clear selective advantage relative to wild-type NL4-3 was N155H, which consistently displayed positive selective index values across concentrations of raltegravir ranging from 1 to 500 nM. Surprisingly, in spite of its higher resistance potential, as measured by IC50, mutant Q148H showed only very limited selective advantage, over a narrow range (10 to 100 nM) of raltegravir concentrations. A very different picture emerged from analysis of the selective index of double mutants, revealing a higher and wider selective-advantage profile for the Q148 pathway. Indeed, mutant Q148H+140S exhibited the strongest selective advantage, across the widest range of raltegravir concentrations (0.1 to 10,000 nM). Mutant N155H+E92Q showed a lower and narrower (1-500 nM) selective index curve, which, in spite of its higher IC50, did not seem to differ from that measured with mutant N155H alone. Finally, although the Q148H+N155H expressed a nearly complete lack of susceptibility to raltegravir, the selective-advantage curve for this mutant did not reach positive values at any point across the range of tested raltegravir concentrations.
Recent studies on the phenotype of raltegravir-resistant viruses have shown that the individual impact of raltegravir resistance mutations on HIV-1 resistance to this drug in tissue culture-based susceptibility assays differs according to the mutation and to the number of mutations carried by the virus (6, 7, 17). Mutation N155H has been found to decrease HIV-1 susceptibility 10- to 50-fold relative to the wild type, a level of resistance that can be strongly increased by addition of mutation E92Q. Mutations Q148H or Q148R decrease susceptibility to extents that are comparable or slightly higher than those expressed by N155H, and addition of mutation G140S always results in a strong boost in resistance, together with a significant improvement in viral fitness. However, the fitness impact of single mutations at N155 and Q148 has been found in most recent studies to be relatively modest (6, 7) compared to the very strong fitness impairment observed by mutations selected in vitro by earlier generations of diketoacid INSTI compounds (5).
In view of their comparable impact on resistance and on fitness in vitro, it is therefore unclear why, in vivo, viral quasispecies having started evolution of resistance following the N155 pathway are almost always outcompeted by variants expressing the Q148 pathway. We hypothesize that this is due to the fact that resistance and fitness, as expressed as IC50 values and as RC in the absence of drug, respectively, do not fully express the selective advantage of viral mutants across a wide range of raltegravir concentrations, such as may be encountered by HIV-1 during its replication in vivo. Our results, examining the latter parameter, show that the single mutation that confers the strongest and widest selective advantage for replication in the presence of raltegravir relative to wild-type is N155H. Viral genomes bearing this mutation, however, fail to gain selective advantage by adding mutation E92Q: although this addition is indeed able to increase the IC50, it also penalizes viral RC and does not appear to enhance or widen the selective-advantage profile conferred by N155H alone. In patients developing this rare combination of mutations, it is possible that additional compensatory changes are required to increase fitness, a possibility that remains to be examined. Conversely, while the single mutant Q148H fails to express a significant selective advantage in spite of its higher IC50, the addition of G140S fully corrects RC, strongly increases IC50, and results in a selective-advantage curve that is both the highest and the widest among all viruses tested here.
These results are fully consistent with the dynamics of evolution of viral genotypes during HIV-1 raltegravir escape in patients failing combination therapy, including this drug. Recent results have shown that among viral clones analyzed during the early stages of escape, single N155H mutants are predominant, while at later stages, double mutants of the Q148 or Y143 pathway gradually dominate the viral population (6, 17). Among the complex population of pretherapy viruses, N155H is the first to emerge both in view of its higher RC, which determines its frequency in the population before treatment, and in view of its highest selective advantage in the presence of raltegravir. At that time, Q148H is only moderately favored over the wild type, as reflected by the fact that genomes carrying Q148H alone is rarely observed as a dominant population in patients failing raltegravir treatment. As pharmacological pressure by raltegravir continues to be exerted on these viral mixtures, the rare Q148H mutants will gradually acquire the G140S mutation, leading to a combination that has the best selective advantage over any of the other competitors, explaining its predominance after several weeks of viral evolution under raltegravir pressure. Of note, the more frequent occurrence of N155H, as opposed to Q148H, at early stages of raltegravir resistance evolution cannot be explained by nucleotide mutation biases: N155H requires an A→C mutation, and Q148H can result from either A→C or A→T mutations, which are equally likely to be generated during reverse transcription. Both substitutions result from single nucleotide changes, excluding the possibility that selection is biased by sequential nucleotide changes implying transient selection of an intermediate amino acid with a potential for altering viral fitness (10). Finally, our results showing that, in spite of a very high IC50, the low RC and the absence of any selective advantage by double mutant Q148H+N155H are fully consistent with the lack of overlap between the N155 and the Q148 pathways in vivo.
We thank Daria Hazuda (Merck) for the gift of raltegravir and Allan J. Hance for critically reading the manuscript.
Published ahead of print on 15 July 2009.