ABSTRACT
Cell entry by HIV-1 is mediated by its principal receptor, CD4, and a coreceptor, either CCR5 or CXCR4, with viral envelope glycoprotein gp120. Generally, CCR5-using HIV-1 variants, called R5, predominate over most of the course of infection, while CXCR4-using HIV-1 variants (variants that utilize both CCR5 and CXCR4 [R5X4, or dual] or CXCR4 alone [X4]) emerge at late-stage infection in half of HIV-1-infected individuals and are associated with disease progression. Although X4 variants also appear during acute-phase infection in some cases, these variants apparently fall to undetectable levels thereafter. In this study, replication-competent X4 variants were isolated from plasma of drug treatment-naive individuals infected with HIV-1 strain CRF01_AE, which dominantly carries viral RNA (vRNA) of R5 variants. Next-generation sequencing (NGS) confirmed that sequences of X4 variants were indeed present in plasma vRNA from these individuals as a minor population. On the other hand, in one individual with a mixed infection in which X4 variants were dominant, only R5 replication-competent variants were isolated from plasma. These results indicate the existence of replication-competent variants with different coreceptor usage as minor populations.
IMPORTANCE The coreceptor switch of HIV-1 from R5 to CXCR4-using variants (R5X4 or X4) has been observed in about half of HIV-1-infected individuals at late-stage infection with loss of CD4 cell count and disease progression. However, the mechanisms that underlie the emergence of CXCR4-using variants at this stage are unclear. In the present study, CXCR4-using X4 variants were isolated from plasma samples of HIV-1-infected individuals that dominantly carried vRNA of R5 variants. The sequences of the X4 variants were detected as a minor population using next-generation sequencing. Taken together, CXCR4-using variants at late-stage infection are likely to emerge when replication-competent CXCR4-using variants are maintained as a minor population during the course of infection. The present study may support the hypothesis that R5-to-X4 switching is mediated by the expansion of preexisting X4 variants in some cases.
INTRODUCTION
HIV-1 entry is initiated by sequential interactions of its envelope glycoprotein, gp120, with its principal receptor, CD4, and a coreceptor, either CCR5 or CXCR4. HIV-1 variants that utilize CCR5 or CXCR4 are called R5 and X4, respectively, while those that utilize both CCR5 and CXCR4 are called R5X4, or dual. In general, R5 variants are predominant throughout the course of infection, while CXCR4-using variants, X4 or R5X4, emerge at later stages of infection in approximately half of HIV-1-infected individuals and are thought to be correlated with decreased CD4 count and disease progression (1–5). However, how CXCR4-using variants emerge at late-stage HIV-1 infection is unclear. The shift from CCR5- to CXCR4-using variants has been suggested to be mediated by a stepwise transition via intermediate R5X4 (6, 7). In contrast, CXCR4-using variants reportedly emerge during acute-phase infection in some individuals (8–13), which suggests that CXCR4-using variants may have been transmitted, but their replication was restricted, probably due to their higher susceptibility to humoral immune responses like neutralizing antibodies (14–17). Given that these restricted CXCR4-using minor variants maintain their infectious potency, they may reemerge at late-stage infection as the immune system deteriorates.
In clinical settings, evaluation of coreceptor usage of HIV-1 in infected individuals is also essential before starting a CCR5 inhibitor, maraviroc, as the presence of preexisting CXCR4-using variants can be selected after the treatment (18). However, maraviroc treatment sometimes did not select CXCR4-using variants, even in individuals predicted to have mixed infection with CCR5- and CXCR4-using variants (19). Of note, the first case of cured HIV-1, the “Berlin patient,” did not show a rebound of CXCR4 variants after stem cell transplantation from a homozygous CCR5-Δ32 donor, although next-generation sequencing (NGS) detected CXCR4-using variants in the patient (20). However, rapid rebound of CXCR4-using variants was observed in a patient who received allogeneic human stem cell transplantation from a homozygous CCR5-Δ32 donor (“Essen patient”), although whether CXCR4-using variants were present before the transplantation was not reported (21, 22). These findings suggest that CXCR4-using variants detected by genotypic assays are not always replication-competent ones. Therefore, detecting replication-competent CXCR4-using minor variants is important for these therapies.
In general, the coreceptor usage of HIV-1 variants in infected individuals is determined by phenotypic or genotypic assays (23). Various phenotypic assays have been established by producing pseudotyped virus, including the enhanced sensitivity Trofile assay (ESTA; Monogram Biosciences, South San Francisco, CA, USA) and others (24–26). However, these assays are time consuming, labor intensive, and not suitable for high-throughput testing. Therefore, genotypic assays are preferred for prediction of coreceptor usage. The genotypic assays are based on the amino acid sequences of the third hypervariable region of gp120 (V3), as it principally determines coreceptor specificity. Using bioinformatics tools and algorithms, including the 11/25 rule (based on the association of positively charged amino acids at position 11 or 25 of the V3 loop with CXCR4 coreceptors) (27), net charge (28–32), and potential N-glycan site (33, 34) of the V3 region, many prediction programs are now available. Among them, Geno2Pheno[coreceptor] is the most widely used; it gives a false-positive rate (FPR) quantitative value for the probability of falsely classifying an R5 variant as an X4 variant (35). However, some full-length env or V3 regions cloned from viral RNA (vRNA) obtained from plasma and used for genotypic and phenotypic assays could be derived from defective viral genomes rather than infectious virus. Therefore, whether vRNA obtained from plasma represents circulating replication-competent HIV-1 in infected individuals is not known.
To clarify whether vRNA cloned from HIV-1-infected plasma is derived from representative infectious virus, we sought to isolate fully replication-competent HIV-1 variants from plasma of HIV-1-infected individuals and to compare their sequences. Surprisingly, some isolated variants had different V3 sequences and coreceptor usage than those cloned from plasma. These results imply that representative plasma vRNA sequences may not be from replication-competent virus in infected individuals or that minor variants that use different coreceptors may have infectious potential but are restricted in vivo.
RESULTS
Coreceptor usage of cloned V3 region from plasma vRNA of CRF01_AE-infected individuals.HIV-1 that uses CXCR4 as a coreceptor must be detected in HIV-1-infected individuals prior to starting the CCR5 inhibitor, maraviroc. Although considerable data are obtainable on the frequency of CXCR4-using HIV-1 in subtype B infection, fewer data are available on CRF01_AE, which is prevalent in Southeast Asia, especially from phenotypic assays. Therefore, we evaluated HIV-1 coreceptor usage in a cohort of CRF01_AE-infected Vietnamese individuals in northern Vietnam (36, 37), using both genotypic and phenotypic assays. As the V3 region of gp120 principally determines coreceptor usage, we tried to amplify the V3 region of gp120 from plasma vRNA of these patients, using nested reverse transcriptase PCR (RT-PCR). We successfully cloned and directly sequenced the V3 regions in samples from 30 individuals; all sequences were suspected to be CRF01_AE subtype (Table 1). The false-positive rate (FPR) was calculated from the amino acid sequence of the V3 region by using the Geno2Pheno[coreceptor] algorithm (FPR indicates the probability of falsely classifying an R5 variant as a CXCR4-using variant). In plasma samples from individual VI-340, we found that the vRNA had distinct V3 sequences with different amino acids, which suggested mixed infection. To confirm mixed infection, the PCR product was cloned into a PCR vector; 19 clones were sequenced (Table 2). We found that 15 sequences were 35 amino acid (aa) residues long (FPR = 0.1), and 4 sequences were 36 aa residues long (FPR = 23.6). In total, we cloned 31 different V3 sequences from 30 individuals, of which 14 (45.2%) were below the 5% FPR cutoff, suggesting a high frequency of CXCR4 usage in this area (Table 1). To confirm coreceptor usage using phenotypic assays, we produced luciferase reporter HIV-1 pseudotyped with Env that carried each V3 region in its strain JR-FL background (38, 39). We used these viruses to infect NP-2/CD4 cells that expressed either CCR5 or CXCR4 (Table 1). We found that 5 of the 31 cases definitely used CXCR4 (16.1% of the total), of which 3 (9.6% of total) used CXCR4 exclusively (X4 variants), whereas 2 (6.4% of total) utilized both CXCR4 and CCR5 (R5X4 variants). There was no correlation between CXCR4 usage and clinical parameters, such as CD4+ T cell count, plasma viral load (pVL), and others (Table 1). To check whether Geno2Pheno[coreceptor] is applicable in these individuals, the concordance rate was calculated. Within the 5% FPR cutoff, the concordance between a phenotypic assay using recombinant virus and Geno2Pheno[coreceptor] was only 35.7%, while the 11/25 rule was completely matched (100%). No significant correlation was observed between CXCR4 usage and the net charge of the V3 region (40) or other positions, such as the basic residue at position 24 (41), arginine at position 18, and others in the V3 region (42).
Clinical characteristics of HIV-1-infected individuals in northern Vietnam and amino acid sequence analyses of V3 regions from plasma vRNA and their coreceptor usage
V3 sequence analyses of isolated viruses
Isolation of HIV-1 from the plasma of HIV-1-infected individuals.As we cloned only the V3 region from plasma vRNA to determine coreceptor usage, whether viruses carrying these V3 regions in the plasma had intact env and were replication competent was unknown. We therefore sought to isolate infectious viruses from the plasma of HIV-1-infected individuals. To this end, primary CD4+ T cells were isolated from peripheral blood mononuclear cells (PBMCs) from an HIV-1 seronegative individual (U-4) and then cultured with IL-2 after stimulation with anti-CD3 monoclonal antibody (MAb). Approximately 100% of these cultured CD4+ T cells expressed CXCR4, while 10% expressed CCR5. In PBMCs without stimulation, we revealed that 100% and 20% of CD4+ T cells expressed CXCR4 and CCR5, respectively (Fig. 1B), indicating that the expression pattern of CCR5/CXCR4 was sufficient to support the replication of both R5 and CXCR4-using variants (43, 44). These cultured CD4+ T cells were incubated with HIV-1-positive (HIV-1+) plasma and cultured for 5 to 7 days. Positive viral culture was determined by p24 antigen (Ag) production in culture supernatant. We successfully isolated primary HIV-1 variants from six individuals; the V3 regions of these isolates were then cloned and sequenced. In three individuals (VI-065, VI-072, and VI-152), almost identical V3 sequences were obtained from these isolates (Table 2), which confirmed that the cloned V3 sequences obtained from vRNAs were from representative HIV-1 viruses replicating in the plasma. In contrast, the amino acid sequences of the V3 region obtained from the isolates in two individuals (VI-157 and VI-158) were quite different from their representative sequences cloned from the plasma vRNAs (Table 2). The V3 region sequences of these two isolates had higher net charges, having charged amino acids at position 11 or 25, and had lower FPRs and lacked potential N-glycan sites at position 6 compared with plasma vRNA sequences (Table 2). In addition, in the case of individual VI-340, only the CCR5-using V3 region sequences were obtained from the isolates, although CXCR4-using sequences were major clones in the plasma vRNA (Table 2). To confirm the coreceptor usage of isolated viruses, NP-2/CD4 cells expressing either CCR5 or CXCR4 with the long terminal repeat (LTR)-driven luciferase gene were established. These cell lines contain intact 5′-LTR linked to the firefly luciferase gene, which is activated by Tat produced from infected cells. Indeed, the isolates from four individuals (VI-065, VI-072, VI-152, and VI-340) used CCR5, whereas the isolates from two individuals (VI-157 and VI-158) used CXCR4 alone (Fig. 2A). Furthermore, the entry of isolates obtained from individuals VI-065, VI-072, VI-152, and VI-340 was totally inhibited by a high concentration of a CCR5 inhibitor (maraviroc), but not by a CXCR4 inhibitor (AMD3100), whereas that of isolates from individuals VI-157 and VI-158 was completely inhibited by AMD3100 but not by maraviroc (Fig. 2B). These findings confirmed that the isolates from individuals VI-065, VI-072, VI-152, and VI-340 were R5 variants and those from individuals VI-157 and VI-158 were X4 variants that exclusively utilized CXCR4 (pure X4).
Expression of CXCR4 and CCR5 on primary CD4+ T cells. Cultured CD4+ T cells (A) and PBMCs (B) from an HIV-1 seronegative individual (U-4) were stained with anti-CD4, anti-CXCR4, and anti-CCR5 MAbs. CD4+ T cells were gated and analyzed for CXCR4 and CCR5 expression.
Coreceptor usage and sensitivity to coreceptor inhibitors of isolated viruses. (A) NP-2/CD4 cells (Parent) or NP-2/CD4 cells that express CCR5 or CXCR4 with the LTR-driven luciferase gene were infected with the same amounts of replication-competent viruses (10 ng p24 Ag). Luciferase activities were determined after 48 h of infection. Cultures without HIV-1 (No HIV) and laboratory strains JR-FL (R5), NL4-3 (X4), and 89.6 (R5X4) were included as negative and positive controls, respectively. RLU, relative light units. (B) TZM-bl cells were infected with the same amounts of replication-competent viruses in the absence (None) or presence of maraviroc (MVC) or AMD3100 at 1 μM. Luciferase activities were determined after 48 h of infection. Data represent the extent of replication relative to that in the absence of maraviroc or AMD3100. Laboratory strains JR-FL (R5), NL4-3 (X4), and 89.6 (R5X4) were included as controls. Data are the mean values ± standard deviations (SD) from triplicate experiments.
Analysis of full-length envs from plasma and isolated virus.As the V3 region amino acid sequences of isolated viruses were substantially different from those cloned from the plasma vRNAs in individuals VI-157 and VI-158, these infected individuals might have had HIV-1 variants with mixed genotypes. To genetically analyze the env sequences of cloned vRNA and isolated virus in the same individual, we next tried to clone full-length envs of the plasma and the isolated virus and compare both amino acid sequences (Table 3). We successfully cloned both full-length envs from individuals VI-152, VI-157, VI-158, and VI-340 and compared them. The sequences of full-length envs from the plasma and the isolated virus from individual VI-152 were almost identical, as we expected. In individual VI-340, only X4-type sequences were cloned from the plasma vRNA, probably due to the few templates in plasma vRNA and the PCR conditions. For individuals VI-157 and VI-158, the sequences of full-length envs from the isolated virus were considerably different from those of plasma vRNAs, especially in the V1/V2 and V3 regions (Table 3). Phylogenetic analyses revealed that envs from each HIV-1-infected individual mapped separately, and those isolated and cloned from the plasma from the same subject were closely related but mapped with distinct clusters (Fig. 3).
Sequence analyses of full-length Envs of plasma vRNA and isolated HIV-1
Comparison of full-length Envs between clones from plasma vRNA and isolated viruses. (A) Full-length env regions were cloned from plasma vRNA and isolated HIV-1, respectively. Amino acid sequences of cloned full-length Envs from plasma vRNA and isolated HIV-1 were aligned using CLC Sequence Viewer. Conservation level is shown at the bottom of each alignment. The approximate locations of V1-V2 and V3 regions are shown at the bottom of the alignment. (B) Phylogenetic tree was constructed using the neighbor-joining method. Laboratory strains HXB2D, BaL, and JR-FL were also included. Bar represents genetic distance.
Coreceptor usage of full-length Envs isolated and cloned from plasma.As we cloned only the V3 regions to determine coreceptor usage from plasma vRNAs, we next confirmed whether the cloned full-length Envs were functional and had the same coreceptor usages as those of the cloned V3 regions. Env expression vectors were constructed as previously described (45), and luciferase reporter HIV-1 viruses with these Envs were produced. NP-2/CD4 cells that expressed CCR5 or CXCR4 were used to determine the coreceptor usage of these pseudotyped viruses (Fig. 4). We found that Envs in isolates from individuals VI-157 and VI-158 exclusively used CXCR4, whereas Envs cloned from the plasma from these individuals used CCR5. Similarly, Env from isolated HIV-1 from individual VI-340 used CCR5, but full-length Env cloned from the plasma exclusively used CXCR4. These results confirmed that isolated Envs and Envs cloned from plasma were both functional.
Coreceptor usage of cloned full-length Envs from plasma vRNA and isolated viruses. NP-2/CD4 cells (Parent) or NP-2/CD4 cells that expressed CCR5 or CXCR4 were infected with the same amounts (10 ng of p24 Ag) of luciferase reporter HIV-1 pseudotyped with full-length Env cloned from plasma vRNAs or isolated viruses. Luciferase activities were determined 48 h after infection. Data are the geometric mean values ± SD from triplicate experiments.
Higher entry efficiency of Envs of X4 variants isolated from VI-157 and VI-158 in primary CD4+ T cells.Since X4 variants were isolated from individuals VI-157 and VI-158 though X4 Envs were not cloned from the plasma, it is speculated that viruses carrying these X4 Envs possessed higher replication efficiency in the primary CD4+ T cells that we used for the isolation of HIV-1 in vitro than did those carrying representative R5 Envs cloned from the plasma. To investigate this, CD4+ T cells from the same donor (U-4) were infected with luciferase reporter HIV-1 pseudotyped with full-length Envs cloned from the plasma vRNA or isolated from individuals VI-157 and VI-158. We found that X4 Envs isolated from the plasma had significantly higher entry efficiency than R5 Envs cloned from the plasma vRNAs of individuals VI-157 and VI-158 (Fig. 5). These results indicated that our in vitro culture conditions were likely to select minor X4 variants in individuals VI-157 and VI-158 due to higher efficiency of replication in the primary CD4+ T cells.
Entry efficiency of cloned full-length Envs from plasma vRNA and isolated viruses in primary CD4+ T cells. The primary CD4+ T cells were infected with the same amounts of luciferase reporter HIV-1 pseudotyped with various full-length Envs from the plasma vRNA and isolated viruses using spinoculation. Luciferase activities were determined 48 h after infection. Envs of strains JR-FL (R5) and NL4-3 (X4) were included as a control. Data are the geometric mean values ± SD from triplicate experiments (**, P < 0.01; ***, P < 0.001).
NGS to detect the circulating minor HIV-1 variant populations with different coreceptor usage in HIV-1-infected plasma.Isolates from two individuals, VI-157 and VI-158, had different amino acids in the V3 region and different coreceptor usage from the representative clones in the plasma vRNAs; these sequences were not detected by conventional Sanger sequencing. Therefore, these infectious viruses may have been derived from the minor population in the plasma. In addition, only R5 variants were isolated from the plasma, whereas major V3 sequences in the plasma vRNA were suggested to be X4 variants in the case of individual VI-340. To confirm whether these minor sequences were indeed present in the plasma vRNA, NGS was performed. Illumina libraries were obtained from plasma vRNA of individuals VI-157, VI-158, and VI-340 using nested PCR. To fully cover all clones from the plasma vRNA, sequences conserved between plasma vRNA and isolated viruses in these cases were redesigned for the nested PCR primers. Approximately 10,000 copies of the plasma vRNA per reaction were chosen for nested PCR, which were estimated from the plasma viral load of each case. The number of total reads of Illumina libraries was more than 3 × 106 in all three individuals, demonstrating that these Illumina libraries are likely to fully cover all sequences cloned from the plasma vRNA. As the V3 regions of R5 and X4 variants in each case had quite different nucleotide sequences between the vRNA and isolated HIV-1, we sought to detect R5- and X4-type sequences from Illumina libraries, which had different nucleotide sequences at both the 5′ and 3′ site of the V3 region in each case (Fig. 6A). First, we could detect both R5 and X4 clones at the 5′ site of the V3 region in each case. The numbers of sequences in major clones (R5 for individuals VI-157 and VI-158 and X4 for individual VI-340) were around 2 × 105 in all three cases, whereas those in minor clones (X4 for individuals VI-157 and VI-158 and R5 for individual VI-340) varied by individual (20,000, 6,000, and 70,000 in individuals VI-157, VI-158, and VI-340, respectively) (Fig. 6B). We next analyzed the nucleotide sequences at the 3′ site in the V3 region, to rule out any artificial template switch by PCR. The numbers of second nucleotide sequences that corresponded to the same coreceptor usage in the first nucleotide sequences were counted. Indeed, almost the same numbers of sequences were found in both major and minor clones in each individual, confirming the presence of both clones in the plasma vRNA. Overall, the number of minor variants was dependent on the individual, but these minor variants with different coreceptor usages having replication capacity were indeed present in the plasma of these three individuals.
NGS of plasma vRNA. (A) NGS libraries were made by nested PCR of cDNA synthesized from the plasma vRNA of HIV-1-infected individuals. The first and second nucleotide sequences in the V3 region that correspond to usage of each coreceptor are shown. (B) The numbers of reads identical to the first sequences are shown by the black bars. The numbers of reads identical to the second sequences are shown by the gray bars. Data from the NGS libraries of VI-157, VI-158, and VI-340 are shown. Data from the NGS library of VI-152 are not shown, since the V3 sequences were totally identical.
DISCUSSION
Before initiating use of the CCR5 inhibitor maraviroc in HIV-1-infected individuals, the lack of CXCR4-using variants must be confirmed. Although large amounts of data have been accumulated regarding coreceptor usage of subtype B HIV-1, coreceptor usage of the CRF01_AE subtype has been less documented. A high frequency of CXCR4-using variants in CRF01_AE has recently been shown, using genotypic assays such as Geno2Pheno[coreceptor] (46–48). However, the frequency of CXCR4-using variants in CRF01_AE was overestimated compared with the results of phenotypic assays, such as ESTA, as previously described (8, 49). In the present study, we also found that 45.2% of the samples examined were below the 5% FPR cutoff by Geno2Pheno[coreceptor], which suggests a high frequency of HIV-1 variants with CXCR4 usage in northern Vietnam, where CRF01_AE is predominant. However, in our phenotypic assay, the frequency of CXCR4-using variants was 16.1%. Thus, the concordance rate was 35.7% when we applied a 5% FPR cutoff. However, a recent phenotypic study of coreceptor usage in CRF01_AE in Hong Kong showed that the estimated frequency of CXCR4-using variants was 39.1% (50). These results suggest that the frequency of CXCR4-using variants is dependent on the area where CRF01_AE-subtype infections are predominant. Notably, however, the frequency of pure X4 variants was higher than that of R5X4 variants (9.6% versus 6.4%) in this area. In general, most CXCR4-using variants were R5X4, but not X4, in subtype B HIV-1 infections (31, 51–53), which suggests that the switch to X4 from R5 variants is mediated by stepwise transition via intermediates such as R5X4 (7, 14, 16, 17). Therefore, the higher frequency of X4 variants suggests that the emergence of X4 variants is likely to be mediated by direct evolution from R5 variants or due to direct transmission of X4 variants that are maintained throughout the course of infection.
In general, determining the coreceptor usage of HIV-1 by phenotypic assay is based on the use of pseudotyped virus with cloned full-length env or recombinant full-length env that carries the cloned V3 region but does not use the full-length HIV-1 genome. Therefore, whether the HIV-1 viruses carrying the cloned env possess the ability to replicate is unknown. In the present study, we sought to isolate HIV-1 variants from plasma to confirm whether the virus carrying the cloned V3 region was replication competent. We successfully isolated six primary HIV-1 variants from plasma of HIV-1-infected individuals. Among them, we found that three of six isolates had V3 region sequences similar to those from the plasma vRNAs, indicating that the viruses carrying the cloned V3 sequences were replication competent in these HIV-1-infected individuals. However, we found that two isolates, from individuals VI-157 and VI-158, had V3 sequences that differed from those in plasma vRNAs. Furthermore, these isolates were found to use CXCR4 exclusively (pure X4), whereas the cloned V3 sequences in the plasma had CCR5-using phenotypes. Similarly, in the case of individual VI-340, whose cloned sequences showed mixed V3 region sequences with the X4 variant dominant in the plasma, only R5 variants were isolated from the plasma. These results indicated that these isolates existed as the minor population in vivo, although they were fully replication competent in vitro. Indeed, using NGS, we were able to identify the sequences that corresponded to the isolated X4 sequences in the plasma vRNA from individuals VI-157 and VI-158. Rozera et al. showed that the proviruses of CXCR4-using variants in monocytes are in the minor population (54) and then suggested that they may be replication competent by showing that the viral genomes in monocytes spread to T lymphocytes at a different time point (55). In the present study, we did confirm that minor variants existing in plasma were fully replication competent. These findings suggest that the use of maraviroc, a CCR5 inhibitor, is inadvisable in this area, even though conventional genotypic and phenotypic assays do not detect the CXCR4-using variants.
Notably, R5 and X4 variants coexisted without intermediate R5X4 variants in individuals VI-157, VI-158, and VI-340. Thus, these findings also support the idea that X4 variants may emerge from R5 variants without stepwise transition via R5X4 in certain HIV-1-infected individuals, as discussed above. Alternatively, X4 variants may be cotransmitted with R5 variants or superinfected at later times in these infected individuals. It is unknown how X4 variants were transmitted in these individuals, since human-to-human transmission is generally mediated by R5 variants. Although it is possible that CXCR4-using variants were transmitted in individuals homozygous for CCR5-Δ32, as previously described (56), the frequency of the CCR5-Δ32 mutation was quite low in this area (57). However, transmission of CXCR4-using variants has been reported recently (10–12, 46, 58–60). Furthermore, it has also been shown that transmission through intravenous drug use was significantly associated with the presence of CXCR4-using variants (61). Since the frequency of transmission through injecting drug use in HIV-1-infected individuals is quite high in Vietnam (around 30%) (62), it is possible that CXCR4-using variants would also be directly transmitted in this area. Indeed, individual VI-157, who had an X4 variant as the minor population, was an injecting drug user.
Interestingly, isolated X4 variants had higher replicative capacity in vitro, although their replication was restricted in vivo. How these variants are restricted in vivo remains to be determined. Generally, the coexistence of multiple replication-competent HIV-1 variants in one individual possibly needs different compartmentalization of each variant. Since R5 and CXCR4-using variants preferentially target CD4+ memory T cell and naive T cell subsets, respectively (43, 63), it is possible that R5 variants infect the memory CD4+ T cell subset, while X4 variants infect the naive CD4+ T cell subset. Due to the lower turnover of naive cells (64), the replication kinetics of X4 variants is expected to be relatively low in vivo. In addition, the CXCR4-using variants have also been reported to target other compartments, such as monocytes (54, 55) or hematopoietic stem cells (65). Lower cell turnover or a quiescent status of these infected cells may also be associated with the restricted replication of X4 variants in vivo.
Taken together, the existence of fully replication-competent minor CXCR4-using variants in plasma may account for the coreceptor switch from R5 to CXCR4-using variants in some individuals, though it is not known whether these findings are unique to the CRF01_AE subtype or to this area. Further studies to define viral and host factors that relieve the restriction of CXCR4-using variants are necessary to fully understand the coreceptor switching mechanism.
MATERIALS AND METHODS
Ethics statement.The study protocol was approved by the institutional ethical review boards of the National Hospital of Tropical Diseases and the Ethics Committee for Epidemiology and General Study in the Faculty of Life Sciences in Kumamoto University. Written informed consent was obtained from all individuals studied according to the Declaration of Helsinki.
Reagents and cells.The human embryonic kidney 293T cell line was obtained from American Tissue Culture Collection (Manassas, VA, USA) and maintained in Dulbecco’s modified Eagle’s medium (DMEM; MilliporeSigma, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS; Thermo Fisher Scientific, Waltham, MA, USA). The TZM-bl cell line was supplied by the AIDS Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases (Bethesda, MD, USA), and maintained in DMEM supplemented with 10% FBS. PM1/CCR5 cells were maintained in RPMI 1640 (Millipore Sigma) supplemented with 10% FBS and G418 (0.1 mg/ml) as previously described (25, 66). The CD4-expressing glioma cell line NP-2/CD4 (67, 68) was provided by H. Hoshino (Gunma University), and its derivative cell lines that expressed either CCR5 or CXCR4 (67, 68) were maintained in Eagle’s minimum essential medium (MEM; Millipore Sigma) supplemented with 10% FBS and appropriate antibiotics. To determine the coreceptor usage of primary isolates, cells of NP-2/CD4 and its derivative cell lines were transduced with a lentiviral vector that carried an HIV LTR-driven luciferase gene and then selected by puromycin. The CD4+ T cells were prepared from peripheral blood mononuclear cells (PBMCs) from an HIV-1 seronegative individual by using magnetic beads conjugated to anti-human CD4 monoclonal antibody (MAb) (Miltenyi Biotec, Bergisch Gladbach, Germany). Isolated CD4+ T cells were cultured in wells coated with anti-human CD3 MAb (clone OKT3; MBL, Aichi, Japan) in RPMI 1640 medium with 10% FBS, 20 ng/ml of recombinant human interleukin 2 (rhIL-2), and 2.5 ng/ml of rhIL-4 for 7 days. The cultured cells were stained with anti-CD4-fluorescein isothiocyanate (FITC; BD Biosciences), anti-CCR5-allophycocyanin (APC; BioLegend), and anti-CXCR4-phycoerythrin (PE; BioLegend) antibodies, and then the surface expression of CCR5 and CXCR4 on these cells was analyzed by using a FACSCant II instrument (BD Bioscience). The CCR5 inhibitor maraviroc (69) and the CXCR4 inhibitor AMD3100 (70, 71) were supplied by the AIDS Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases (Bethesda, MD, USA).
Sequencing of V3 region from plasma vRNA and prediction of its coreceptor usage.Viral RNA was extracted from plasma of HIV-1-infected individuals using the QIAamp blood minikit (Qiagen, Hilden, Germany). cDNA was synthesized from the vRNA using the SuperScript III first-strand synthesis system for RT-PCR (Thermo Fisher Scientific) with random hexamers as previously described (36). The V3-spanning region was amplified using forward primer 5′-AATGTCAGCACAGTACAATGTACAC-3′ or 5′-GAGCCAATTCCCATACATTATTGT-3′ and reverse primer 5′-GCCCATAGTGCTTCCTGCTGCT-3′ or 5′-GCCCATAGTGCTTCCTGCTGCTCCCAAGAACC-3′ for the first PCR and forward primer 5′-TCAACTCAACTGCAGTTAAAT-3′ or 5′-TGTGCCCCAGCTGGTTTTGCGAT-3′ and reverse primer 5′-AGAAAAATTCCCCTCTACAATTAA-3′ or 5′-TATAATTCACTTCTCCAATTGTCC-3′ for the second PCR (72–74). The amplified product was directly sequenced or cloned into the TOPO vector (Thermo Fisher Scientific) for sequencing using an ABI Prism 3771 automated sequencer (Thermo Fisher Scientific). The V3 amino acid sequences were submitted to the Geno2Pheno[coreceptor] algorithm (https://coreceptor.geno2pheno.org) to predict the coreceptor usage of the V3 loop. The results are given with a quantitative value, the false-positive rate (FPR), that defines the probability of falsely classifying an R5 variant as an X4 variant (35). The 11/25 rule, which is based on the association of positively charged amino acids at position 11 or 25 of the V3 loop with CXCR4 coreceptors, was also applied to predict HIV-1 coreceptor usage (27).
Virus isolation.Plasma samples from treatment-naive HIV-1+ individuals were collected at National Hospital of Tropical Diseases (NHTD) in Hanoi, Vietnam. To isolate primary HIV-1 from plasma samples, activated CD4+ T cells from an HIV-1-seronegative individual (as mentioned above) were infected with 2-fold-diluted plasma by spinoculation and cultured for 5 to 7 days in the presence of recombinant human IL-2 (PeproTech, NJ, USA). Viral replication was monitored by p24 Ag production in culture supernatant using a p24 Ag enzyme-linked immunosorbent assay (ELISA; Zeptometrix, Buffalo, NY, USA) according to the manufacturer’s protocol. The virus-positive culture was further expanded by infecting PM1/CCR5 cells as previously described (25, 66).
Cloning full-length env genes from plasma vRNA or isolated HIV-1 and phylogenetic analysis.To clone the full-length env genes of isolated viruses, DNA sequences were extracted from PM1/CCR5 cells infected with isolated viruses and used as templates for PCR. To clone the env gene from the extracted plasma vRNA, cDNA was first synthesized using the SuperScript III first-strand synthesis system for RT-PCR (Thermo Fisher Scientific) with random hexamers as previously described (36). The full-length env gene was then amplified using forward primer 5′-TAG AGC CCT GGA AKC ATC CRG GAA G-3′ or 5′-TAG ATC CTA ACC TAG AGC CCT GG-3′ and reverse primer 5′-TAG CCC ATC CAG TCC CCC CTT TTC-3′ or 5′-ACC TGY GGC YTG ACT GGA AAG CCT AC-3′ for the first PCR and forward primer 5′-CTT GGT ACC GAG CTC GTT AGG CAT CTC CYA TGG CAG GAA GAA G-3′ and reverse primer 5′-GGG AGG GAG AGG GGC GAT ATT RCT ACT TGT TAY TGC TCC AT-3′ for the second PCR. Full-length env nucleotide sequences were aligned using a CLC sequence viewer. The phylogenetic tree was constructed by the neighbor-joining method. The full-length Env expression vector was constructed as previously described (45).
Pseudotyped HIV-1 production and determination of coreceptor usage.For coreceptor usage by HIV-1 that carries a V3 region from plasma vRNA, an Env expression vector to carry the V3 loop from the plasma vRNA was constructed as previously described (38, 66, 75–77). The V3-spanning region that carries AflII and NheI sites was amplified from the first PCR product for the V3 sequencing as mentioned above, using forward primer 5′-GCACCTTAAGAAATCTGTAGAAATCAATTG-3′ (812.7AF1) and reverse primer 5′-GCTAGCTACCTGTTTTAAAGCTTTATACC-3′ (812.7NR1) (underlines denote the AflII and NheI site, respectively) (75). The amplified fragment carrying the AflII and NheI sites was cloned into a pCR-TOPO vector (Invitrogen) and sequenced. The AflII-NheI-carrying fragment of the V3 region was then introduced into the AflII-NheI cloning site of the pCXN-FLan vector as previously described (39, 75). A luciferase reporter HIV-1 pseudotyped with Env carrying the V3 region or full-length Env was prepared by transfecting 293T cells with pNL-LucΔBglII and each Env expression vector (25, 38, 75–77). The virus-containing culture supernatant was recovered and filtered through a 0.45-μm filter and then stored at −80°C until use.
Determination of coreceptor usage and inhibitor sensitivity of isolated viruses.Coreceptor usage of the isolated viruses was determined by infecting NP-2/CD4 cells that carried the HIV-1 LTR-driven luciferase gene, with or without each coreceptor. Briefly, the same amount of Gag (10 ng) was used to infect the NP-2/CD4 cell line expressing CCR5 or CXCR4, and luciferase activity was measured after 48 h of infection, using a luminometer, in triplicate experiments (Lumat LB 9501/16; EG&G Berthold).
Sensitivity to coreceptor inhibitors was determined using TZM-bl cells (44). Briefly, TZM-bl cells were first incubated with the CCR5 inhibitor maraviroc or the CXCR4 inhibitor AMD3100 at 1 μM; infectious viruses were then added and cultured for 48 h. The luciferase activity of infected cells was measured using a luminometer (Lumat LB 9501/16; Bertold). Sensitivity to coreceptor inhibitors was expressed as the percentage of infection in the absence of coreceptor inhibitors.
NGS of the V3 region of plasma vRNA.Plasma vRNA was reverse transcribed to cDNA using the SuperScript III first-strand synthesis system (Thermo Fisher Scientific) with random hexamers as previously described. The V3-spanning region was amplified using PrimeStar HS (hot-start) DNA polymerase (TaKaRa, Tokyo, Japan) with forward primer 5′-TCACARACAATGCC-3′ and reverse primer 5′-CCCTCTACAATTAAAATGATG-3′ for the first PCR and forward primer 5′-CCATAATAGTGCACCT-3′ and reverse primer 5′-TAGATCTCCTCCTGAGG-3′ for the second PCR. Amplified products were purified using the QIAex II gel extraction kit (Qiagen) and quantified with Qubit 2.0 using a double-stranded DNA (dsDNA) HS (high-sensitivity) assay kit (Thermo Fisher Scientific). The indexed library was prepared using the NEBNext ultra DNA library preparation kit for Illumina and NEBNext multiplex oligonucleotides for Illumina (New England Biolabs, MA, USA) and requantified using the Agilent 2100 bioanalyzer (Agilent, CA, USA). Sequence reads were obtained by MiSeq using reagent kit version 2 and 500 cycles (Illumina, CA, USA). Clean reads were obtained by removing low-quality reads (<Q30) through trimming using a quality trimming tool, Sickle (https://www.github.com/ucdavis-bioinformatics/sickle). A quality score (Q-score) is a prediction of the probability of an error in base calling. Q30 means an error probability of 0.001 (1 in 1,000). To determine the frequency of each virus with different coreceptor usage in the same plasma vRNA, the number of reads carrying the nucleotide sequence at the 5′ site of the V3 region that was specific to the usage of each coreceptor of the virus was first counted, and then the number of nucleotide sequences at the 3′ site of the V3 region specific to the same coreceptor usage among the reads having the first sequences was further counted.
ACKNOWLEDGMENTS
We thank the AIDS Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, for kindly providing TZM-bl cells, maraviroc, and AMD3100. We also thank Yutze Yuan and Keisuke Yusa for helpful discussions.
This research was supported by a JSPS KAKENHI grant-in-aid for scientific research (C) (grant number 16K09938), Japan, and by a grant from the Joint Usage/Research Center on Tropical Diseases, Institute of Tropical Medicine, Nagasaki University (grant number 2017-Kyoten-02).
Y.M. and M.T. contributed the study’s concept and design. K.V.N., T.V.N., and M.T. selected and enrolled patients. G.V.T., T.A., H.M., N.K., and Y.Z. processed blood and extracted vRNA from the plasma. Y.M. isolated HIV-1 from the plasma. T.T., T.C., C.H.P., and A.H.Q.P. performed and analyzed next-generation sequencing. T.K. and H.T. cloned the env region. Y.M., H.T., R.F., and N.K. performed laboratory experiments. Y.M. and M.T. analyzed and interpreted the data. Y.M. wrote the manuscript with support from K.M., T.S., S.M., F.H., and T.Y. All authors reviewed and contributed to the final manuscript.
FOOTNOTES
- Received 5 February 2020.
- Accepted 4 April 2020.
- Accepted manuscript posted online 15 April 2020.
- Copyright © 2020 American Society for Microbiology.
