The diversity of Human Immunodeficiency Virus Type 1 (HIV-1) results from quick, high level viral replication  and the misincorporation of host RNA polymerase Ⅱ and viral reverse transcriptase [13, 16, 19]. Some mutations might have no effect on the fitness of HIV-1, but would accumulate during viral replication. Before the patients accepted high active antiretroviral therapy (HAART), some important drug resistance mutations might have existed with very low frequency in drug naÏve patients and variants with these mutations would become drug resistance strains .
Protease plays an important role during HIV-1 replication because it cleaves the long protein chains into small pieces for replacement into new virions as new virus particles begin budding off the cell membrane [3, 18]. Most primary PIs resistance mutations appeared to alter the structure of the substrate binding activity of protease by decreasing the PIs' binding ability but still allowing the natural substrate process [3, 10, 18].
Currently, the Chinese government freely provides drugs for HIV/AIDS treatment, including Lamivudine (3TC), Stavudine (D4T), Didanosine (ddI), Zidovudine (AZT), Efavirenz (EFV), Nevirapine (NVP) and Indinavir (IDV). First line drugs are D4T or AZT, ddI and NVP. IDV is used as an alternative drug, but with the increased resistance to first line drugs, IDV will be used more widely in China in the near future.
The generalized genotype drug resistance tests consist of obtaining bulk RT-PCR (reverse trans-cription-polymerase chain reaction) products from multiple viral molecules, sequencing, identification of mutations by submitting the sequences to the HIV-1 drug resistance database and interpretation of drug resistance mutations . Although these methods provide a mixture of viral quasispecies sequences (consensus sequences), they only detect the major variant in viral population and cannot provide the proportion of the variants in vivo. Palmer and colleagues developed a new drug resistance analysis method based on serial dilution, single-genome amplification and sequencing (SGA/S) . SGA/S reduced the quasispecies sequences to single molecules by more detailed approach and 20 to 40 single genome sequences could represent the quasispecies in each sample. Jesus and colleagues analyzed the changes of env gene during HIV-1 transmission by SGA . In this paper we report the evolution of IDV resistance, and a rare mutation pattern M46I/ G73S/L90M revealed by SGA.
The treatment regimens, viral load and CD4+ T cell counts were shown in Fig. 1. The results of viral load indicated that the therapy at the beginning was effective, but the occurrence of IDV resistance variants resulted in viral rebound at XLF3. The numbers of protease genes obtained from each time point were 33, 11, 15, 28, 10, and 52 respectively. The numbers of reverse transcriptase genes obtained from each time point were 46, 16, 16, 29, 10, and 54 respectively. All the sequences were submitted to GenBank and the accession numbers range from GU328709 to GU328846 and GU345711 to GU345744.
Before the IDV treatment (XLF1), there were no PI-resistance mutations. But there were some represented polymorphisms, such as A71V, and unrepresented polymorphisms in the protease gene, such as N37D, I62V, L63P, H69Y, I72E, V77I, and I93L (Fig. 2). At XLF2, NVP was discontinued, and replaced by IDV. The proportion of A71V declined from 93.9% to 72.7%, and the proportion of wild type variants rose from 6.1% to 27.3%. Other unre-presentative mutations changed as well as A71V.
The PI-resistance mutations G73S, M46I and L90M appeared after 3 months (at XLF3). The proportion of variants with the G73S mutation was 93.3% and higher than the other two mutations (Fig. 3). G73S appeared as two patterns. First, G73S occurred alone (Fig. 4), and it was sensitive to IDV. Secondly, it occurred together with mutations M46I, L90M to form the G73S/M46I/L90M pattern. This pattern caused intermediate resistance to IDV. However variants with the G73S/M46I/L90M pattern were only present in 13.3% of the viral population.
Figure 2. Detailed information of mutations in protease quasispecies sequences. The WT sequence was obtained from SHDB and the sequences of XLF were obtained from six serial time points. Some sequences which did not contain entire protease sequence were excluded from this figure.
Figure 3. The changes of PI-resistance mutations in this patient. Each line with different shape represents different PI-resistance mutations.
Figure 4. The evolution of PI-resistance patterns in this patient. Each block with different shading represents one mutation pattern for IDV. The x-axis represents these serial samples collected from each time point in this follow-up study. The y-axis represents the percentage of variants with each mutation pattern in the viral populations.
At the next time point (XLF4), the proportion of variants with G73S/M46I/L90M pattern rose from 13.3% to 85.7%. And variants with G73S alone declined from 80.0% to 7.14%. Half a year later (XLF5), there was only one drug resistance pattern, G73S/M46I/L90M. The G73S variants declined and were less than 10%. At the last time point (XLF6), IDV-resistance mutation patterns included G73S, G73S/M46I, D30G/G73S/M46I/L90M and G73S/ M46I/L90M. The former three patterns were rare and the proportions were 3.85%, 3.85%, 1.92% respectively. The latter was 90.4%.
Before the patient accepted IDV, the viral population did not have any drug resistance mutations for IDV, but had one representative polymorphism (A71V) in the protease gene. In untreated patients, A71V is a polymorphism, but will be turned to a non-polymorphism with multiple-resistance mutations. After receiving IDV, the percentage of variants with A71V declined, and the variants with transitional drug resistance mutation G73S became the dominant variants (Fig. 3). At the same time, the mutation pattern G73S/M46I/L90M occurred (ratio was 13.3%). Along with the therapy, G73S turned to G73S/M46I/ L90M, and variants with this pattern became the dominant variants. When the dominant resistance variants had been established, some new drug resistance mutations like D30G in protease occurred based on the G73S/M46I/L90M mutation pattern (Fig. 4).
Because SGA can obtain quasispecies sequences, it could be used for the linkage analysis of special drug resistance mutations, whereas the general genotype drug resistance assay could not . Four protease inhibitor (PI) resistance mutations (major resistance mutations: M46I, L90M. minor resistance mutations: A71V, G73S) in 6 serial samples of XLF had been found. A71V occurred prior to the patient receiving IDV. At XLF3, G73S emerged and was presented as two patterns, A71V/G73S (80.0%) and A71V/M46I/ G73S/L90M (13.3%). M46I/L90M also appeared at XLF3, and all of M46I/L90M appeared with A71V/ G73S (Table 1). When the regimen 3 was sustained for 4 months, variants with the A71V/M46I/G73S/ L90M pattern became the major population with a proportion higher than 90.0%. Because A71V was a polymorphism and occurred with multiple PI-resistance mutations in this case, it was not included in this mutation pattern. The rare mutation pattern was defined as G73S/M46I/ L90M.
Table 1. G73S, M46I and L90M Linkaged in protease in 9 sequences of XLF5
During the occurrence and development of IDV resistance, the variants with RTIs resistance mutations in reverse transcriptase maintained a high proportion.
Because this patient had received nucleotide reverse transcriptase inhibitors (NRTIs) (AZT, ddI, 3TC, D4T) when changing from regimen 2 to regimen 3 (Fig. 1), the frequency of TAMs (Thymidine Analog Mutations) (M41L, D67N, K70R, L210W, T215Y, K219Q) was very high (more than 80.0%, Table 2). TAMs were very stable, and always maintained a high proportion. The proportion of TAMs at each time point were 91.3%, 81.3%, 93.8%, 96.6%, 80.0%, 92.6% respectively. M184I causes high-level resistance to 3TC and low-level resistance to ddI. M184I increases the susceptibility to AZT and d4T. M184I was very stable and was maintained as a high proportion (higher than 80.0%) at all 6 time points. N348I was a newly defined drug resistance mutation that occurred at a reverse transcriptase connection domain, and would cause low-level resistance to AZT and NVP . N348I was also stable, and the proportion at each time point were 91.3%, 81.3%, 87.5%, 100.0%, 80.0%, 90.7%. Combined with TAMs, E44D could cause low-level resistance to 3TC and almost all NRTIs. In this patient, E44D occurred with TAMs all the time and with a high percentage. L74I and V118I would cause low-level resistance to NRTIs when occurring with TAMs.
Table 2. The number (percentage) of NRTI-resistance mutations at each time point
When the therapy regimens changed from 2 to 3, NVP was discontinued. Mutation K101E in the reverse transcriptase could cause intermediate resistance to NVP and G190A could cause high-level resistance to NVP. The frequencies of these two mutations was very high, more than 80% at each time points. The percentage at each time point was 95.7%, 81.3%, 93.8%, 93.1%, 80.0%, and 92.6% respectively for the K101E mutation, and 93.5%, 81.3%, 93.8%, 100.0%, 80. 0%, and 92.6% for the G190A mutation. V179M was an unusual mutation and had a high percentage (higher than 80% at each time point). In this case, V179M occurred with K101E and G190A. M230I was also an unusual mutation (Table 3).
At the first four time points of this study, the PI-resistance mutations increased quickly, from 0.0% to 85.8%. But RTI-resistance mutations were always present in high proportions (more than 80.0%, Table 2 and 3). From XLF3 to XLF6, the PI-resistance mutations generally maintained a high percentage as well as RTI-resistance mutations (more than 80.0%, Fig. 2).
Table 3. The number (percentage) of NNRTI-resistance mutations at each time point
Some mutations, which were present less than 20% of the time but which could cause drug resistance during HAART, were described as low-level drug resistance mutations. The detection of low-level drug resistance mutations is very important for understanding the evolution of drug resistance. Drug resistance is not an all-or-no phenomenon and low-level drug resistance could increase and become a major population under the drug selection pressures and so drug resistance phenomenon should be described as minor or major, low-level or high-level. At XLF3, mutations M46I and L90M occurred, but the ratio was very low (13.3%, Table 4). The information about these two PIs mutations would be missed by a general genotypic drug resistance assay . As treatment was prolonged, variants with these two mutations became the major population and caused resistance to IDV. So the detection of these two mutations provided detailed, precise information for the evolution of IDV resistance.
Table 4. Detection of low-level resistance mutation
Some polymorphism in protease declined while the resistance mutations increased. Percentages of K43R, R41K and R57K decreased as the treatment continued (Table 5). K43R rebounded at XLF5. N37D, I62V, L63P, H69Y, I72E and V77I always maintained high proportions in the viral population. T31I, M36V, K43I, Y59C and E65K occurred only once in viral population during therapy. There were two interesting phenomenon (Fig. 2). K43R occurred when G73S disappeared. When G73S occurred and became dominant variants, K43R disappeared and only appeared when G73S was absent. R57K and I62 V had a negative correlation similar to as K43R and G73S.
Table 5. Proportions of polymorphism in protease gene.