We investigated the emergence of the Q151M MDR complex in one of the two patients in the CHAP2 cohort study who had developed resistance via the Q151M pathway [19
]. The patient, designated P66, was infected with HIV-1 subtype C virus.
Dynamics of emergence and genetic linkage of Q151M MDR complex mutations
Patients enrolled in the CHAP2 cohort study had CD4 counts done approximately every 6 months and plasma was stored for retrospective viral load and genotypic testing. For patient P66, six samples were collected at 0, 4, 10, 17, 28, and 37 months after initiation of therapy; four of which were available for viral load testing and SGS analysis. The viral load and CD4% counts for patient P66 are shown in figure . We initially determined the development of Q151M MDR complex using SGS of full-length RT gene in the four sequential samples collected from patient P66 at 4, 17, 28 and 37 months. More than 30 single-genome sequences were generated per time point except for the 4- and 28-month time points when we obtained 6 and 0 sequences respectively. Genetic linkage analysis of the single genomes at 4, 17 and 37 months showed that the patient acquired the Q151M MDR mutations in the order: A62V, V75I and finally Q151M (Table ). The emergence of Q151M after the secondary mutations A62V and V75I is rare. In addition, the analysis showed that drug resistance mutation T69N was genetically linked to Q151M MDR mutations and was acquired prior to Q151M.
Clinical profile of patient P66. Longitudinal viral load, CD4% and ART regimen data for patient P66 during a 3-year follow up period starting from initiation of cART.
The sequential acquisition of Q151M MDR mutations and the frequency of other RT mutations linked to MDR mutations, in patient P66.
Accessory mutations in the DNA pol domain of RT have previously been demonstrated in the route to acquisition of Q151M MDR complex in subtype B viruses [12
]. We, therefore, determined whether accessory mutations developed in this subtype C HIV-1 virus and whether the C-terminal region of RT played a role in the emergence of the Q151M MDR complex. The emergence and presence of mutations in DNA pol domain, connection subdomain and RNase H domain were assessed by SGS, and their genetic linkage to Q151M MDR mutations was determined. A pre-treatment sample was not available for analysis from patient P66; therefore a codon change was scored as a mutation if it met one of the following criteria: (i) if it was a known drug resistance mutation as determined by International AIDS Society-USA (IAS-USA) [29
], (ii) if it was not present in sequences from a previous time point or underwent a significant change in frequency between time points. This analysis showed a cumulative increase in mutations in all RT domains (Table ). Mutations were identified at 12 codon positions in DNA pol domain, namely, 31, 33, 48, 68, 102, 123, 135, 174, 197, 202, 203 and 314; seven in connection subdomain, 357, 371, 386, 399, 403, 458 and 471; and one in RNase H domain, 517. The correlation between the progressive increments in the frequency of these mutations and the sequential acquisition of the Q151M MDR mutations suggested that they could be facilitating the emergence of the Q151M MDR complex. This notion is further supported by the observation that 18 out of the 20 mutations were present in a majority of the single genomes by 37 months and nearly half of them were present in all the single genomes (Table ).
The Q151M MDR mutations were also genetically linked to NRTI mutations M184IV and L210F, and NNRTI mutations E138A, Y181I and H221Y (Table ). Of note, the N348I mutation was identified in the connection subdomain of all single genomes at 4 months. However, the mutation was present in only one out of 33 single genomes at 17 months but none of the 31 single genomes at 37 months when the Q151M mutation emerged (Table ).
Intrapatient viral genetic diversity in the route to acquisition of Q151M MDR complex
The evolution and viral population dynamics within patient P66 were examined further by phylogenetic analyses. Maximum likelihood (ML) trees of the PR-RT single-genome sequences generated from the sequential samples of the patient are shown in Figure . In general, the ML-inferred genealogy clustered all single genomes from each time point within a monophyletic clade with corresponding progressive increases in genetic distances. Intriguingly, the analyses also showed a serial replacement effect with sequences from successive time points arising from a single branch of a cluster of sequences from a preceding time point. This suggests a serial founder effect in the development of Q151M MDR. Furthermore, ML-inferred genealogy of the sequences with drug resistance codons removed showed that the serial founder effect and monophyletic clustering of the sequences from each time point was maintained (Figure ). This indicates that the identified accessory mutations could be playing an important role in the evolution and development of the Q151M MDR.
Figure 2 ML phylogenetic analysis of single genome sequences. Branch lengths were estimated using the GTR model of substitution and are drawn in scale with the bar at the bottom representing 0.008 nucleotide substitutions per site. The colour of each tip branch (more ...)
High prevalence of some of the identified accessory mutations in subtype B and C infected patients
Next, we determined if the 20 accessory mutations that we identified in patient P66 were present in other patients who had developed resistance via the Q151M pathway. We compared mutation frequencies in subtype B or C samples from RTI-treatment naïve patients and Q151M-containing patient samples on the Stanford University HIV drug resistance database. A significant number of sequences (15 to 12,361) were available for analysis in each subgroup, except for connection subdomain and RNase H domain of Q151M-containing subtype C sequences, in which there was only one sample sequenced beyond the DNA pol domain. Therefore, the analysis for subtype C sequences could only be carried out for the DNA pol domain. This showed that eight out of the 12 codon positions identified in the DNA pol domain of patient P66 were significantly associated with the sequences containing the Q151M mutation compared to RTI-treatment naïve sequences. These codon positions were 31, 33, 48, 68, 123, 174, 202 and 203 (P ≤ 0.042; Table ). In contrast, two of these codon positions, namely 48 and 174, were not associated with the acquisition of Q151M in subtype B infected patients, but an additional two others were, namely 102 and 197 (P ≤ 0.029). Interestingly, codon positions 386 and 403 in connection subdomain were also significantly associated with the acquisition of Q151M in subtype B infected individuals (P ≤ 0.018). These data indicate that some of the accessory mutations identified in the DNA pol domain and connection subdomain of patient P66 are highly prevalent in patients who develop resistance through the Q151M pathway and that they could be playing an important role in the acquisition of the Q151M MDR.
Analysis of the frequency of accessory mutations in RTI-treatment naïve and Q151M-containing sequences on Stanford University HIV database.
C-terminal mutations are not associated with decreased susceptibility of Q151M-containing viruses to NRTIs in patient P66
Consequently, we investigated whether the C-terminal mutations we observed affected susceptibility to NRTIs. Unique restriction sites were introduced in RT and IN genes without changing the amino acid coding, in both the packaging vector and cloned patient fragments in order to facilitate RT domain-swapping (Figure ). The patient-derived RTs remained d4T-susceptible until the development of the Q151M mutation at 37 months, when there was a significant increase (~16-fold) in IC50
values compared to wild-type RT (Figure ; P
< 0.002). At most we observed a 1.3-fold change in susceptibility to d4T at 4 or 17 months leading us to conclude that Q151M is the main contributor to d4T resistance in the Q151M MDR complex. The patient-derived RT exhibited a 23-fold increase in 3TC IC50
values at 4 months which did not increase at 17 and 37 months despite the acquisition of the Q151M MDR mutations (Table ). The effect on susceptibility to 3TC was probably due to M184I/V mutations which were seen by 4 months. The 23-fold reduction in susceptibility is relatively lower than observed in other studies [30
]. This could be because our assay uses full-length RT fragments derived from clinical isolates. It has recently been shown that the use of a co-evolved or subtype-specific C-terminal region of RT can influence the magnitude of drug resistance observed in a phenotypic drug susceptibility assay [32
Figure 3 NRTI susceptibilities and replicative capacity associated with RT domains of patient P66. (A) Schematic representation of full-length and chimeras of subtype C wild-type and patient-derived RT gag-pol expressing vectors used for drug susceptibility and (more ...)
3TC, AZT and FTC susceptibilities associated with RT domains of patient P66.
Analysis of susceptibilities of patient-derived RTs to the CHAP2 second-line NRTIs ddI and ABC showed a cumulative decrease in susceptibility in the order; 1.2- and 1.7-fold at 4 months, 4- and 6-fold at 17 months, and finally 9.9- and 10.8-fold at 37 months, respectively (Figure ). Thus, unlike d4T the cumulative acquisition of mutations on the route to Q151M MDR complex results in a parallel cumulative decrease in susceptibilities to ABC and ddI. In addition, the recombinant viruses expressing patient-derived RTs exhibited decreased susceptibilities to NRTIs FTC of >79-fold at 4 months and AZT of >15-fold at 37 months (Table ) but remained susceptible to TDF even after the acquisition of the Q151M mutation at 37 months (Figure ) with no significant increases in IC50
> 0.18). The susceptibility to TDF could probably be influenced by the presence of M184V which has been shown to increase HIV-1 sensitivity to TDF [33
The expression of the patient-derived DNA pol domain at 37 months plus wild-type C-terminal region or coevolved connection subdomain showed no significant differences in IC50 values to d4T (P > 0.05) suggesting that none of the identified C-terminal mutations in patient P66 at 37 months contributed to the reduction in susceptibility to d4T (Figure ). Similarly, the coevolved C-terminal region did not contribute to 3TC resistance, including the previously identified N348I mutation at 4 months, neither did they contribute to the decreases in susceptibility to ABC, ddI or FTC (Figure and and Table ). However, we observed an effect of the C-terminal mutations at 37 months to AZT, with the co-evolved C-terminal region contributing a 2.5-fold increase in AZT resistance (Table ).
Finally, we determined the effect of the mutations on susceptibility to NVP, the NNRTI used for first-line therapy in the CHAP2 cohort study. The recombinant viruses expressing the patient-derived C-terminal region at 4 months, but not at 17 or 37 months, exhibited a 5-fold increase in the NVP IC50 value relative to wild-type (P < 0.002; Table ). The decrease in NVP susceptibility associated with the C-terminal domain at 4 months is likely due to the presence of the N348I mutation in the connection subdomain which disappears at later time points.
NVP susceptibilities associated with RT domains of patient P66.
Connection subdomain mutations in patient P66 partially restore replicative fitness of Q151M MDR-containing viruses
Since we did not observe any association of C-terminal mutations at 37 months with a decrease in susceptibilities to first-line drugs, we evaluated their effect on virus replicative capacity by infecting HEK293T cells with equivalent amounts of virus. The patient's sample before initiation of therapy was not available, thus the replicative capacity of the viruses measured by relative luciferase light units was compared to that of the virus expressing full-length patient-derived RT at 4 months. The patient-derived RT at 4 months had already developed the M184I mutation which is known to affect viral replicative fitness [35
]. The virus expressing the full-length patient-derived RT containing the Q151M mutation at 37 months demonstrated ~42% replicative capacity of full-length patient-derived RT at 4 months (P
< 0.0001; Figure ). This was further significantly decreased to ~22% (P
< 0.0001) when the patient-derived DNA pol domain at 37 months was expressed in combination with wild-type connection subdomain and RNase H domain. This decrease in replicative capacity was fully compensated (to ~55% replicative capacity) by the coexpression of the coevolved connection subdomain at 37 months. In contrast, replicative capacity of the full-length patient-derived RT at 17 months was comparable to that at 4 months. This suggests that the Q151M mutation, as well as being the main determinant of drug resistance in the Q151M MDR complex, also has a more significant effect on virus replication fitness that is partially restored by mutations in the connection subdomain.