|Home | About | Journals | Submit | Contact Us | Français|
It is widely documented that a complete switch from the predominant CCR5 (R5) to CXCR4 (X4) phenotype is less common for HIV-1 subtype C (HIV-1C) compared to other major subtypes. We investigated whether dualtropic HIV-1C isolates represented dualtropic, mixed R5 and X4 clones or both. Thirty of 35 functional HIV-1 env clones generated by bulk PCR amplification from peripheral blood mononuclear cells (PBMCs) infected with seven dualtropic HIV-1C isolates utilized CXCR4 exclusively. Five of 35 clones displayed dualtropism. Endpoint dilution of one isolate did not yield a substantial proportion of R5-monotropic env clones. Sequence-based predictive algorithms showed that env sequences from PBMCs, CXCR4 or CCR5-expressing cell lines were indistinguishable and all possessed X4/dualtropic characteristics. We describe HIV-1C CXCR4-tropic env sequence features. Our results suggest a dramatic loss of CCR5 monotropism as dualtropism emerges in HIV-1C which has important implications for the use of coreceptor antagonists in therapeutic strategies for this subtype.
HIV/AIDS is the most serious infectious disease challenging the public health sector in sub-Saharan Africa. The disease remains the leading cause of death in the region and in 2007 alone, 76% of deaths were as a result of HIV/AIDS, with 68% of all new HIV infections occurring in this region. Southern Africa is the most seriously affected sub-region and in 2007 accounted for 32% of all new infections and AIDS-related deaths worldwide. South Africa constitutes the highest number of HIV infections globally (UNAIDS, 2007). The main circulating HIV-1 subtype in South Africa is HIV-1 subtype C (HIV-1C), which accounts for approximately 56% of infections worldwide (Esparza, 2005; Hemelaar et al., 2006; Visawapoka et al., 2006). Subtype groupings are based on sequence variations that occur within all gene and non-protein coding regulatory regions, but the most dramatic differences are found in the envelope (env) gene (Gao et al., 1998). The envelope is an important target of humoral immune responses and is a crucial determinant of overall viral fitness (Ball et al., 2003; Marozsan et al., 2005). The env gene plays an important role in viral transmission by determining which coreceptor the virus uses to mediate entry. During transmission and subsequent to infection, viral fitness and target cell tropism properties are thought to be important determinants of infectivity and the rate of disease progression (Troyer et al., 2005). The importance of the envelope as a major target of humoral immunity, its contribution to overall fitness during transmission, and its role in the rate of disease progression make it a particularly attractive target for vaccine and drug development. However, progress towards these goals has been greatly hindered by the extreme genetic variability of the env gene.
HIV-1 utilize members of the seven transmembrane chemokine receptor family as coreceptors for entry into target cells (de Roda Husman et al., 1999; Oppermann, 2004; Ross and Cullen, 1998; Vila-Coro et al., 2000; Xiao et al., 1999). The virus gp120 envelope glycoprotein first binds to the primary CD4 receptor on target cells, which induces conformational changes on the envelope exposing the coreceptor binding site (Rizzuto et al., 1998; Wyatt et al., 1995). The two main coreceptors that the HIV-1 envelope binds to subsequent to the conformational change are CCR5 or CXCR4 (Alkhatib et al., 1996; Choe et al., 1996; Deng and Liu, 1996; Doranz et al., 1996; Dragic and Litwen, 1996; Feng and Broder, 1996). HIV-1 strains can be classified based on their coreceptor utilization, with CCR5 tropic viruses termed R5, CXCR4 tropic viruses termed X4 and viruses that use both coreceptors (dualtropic viruses) termed R5X4 (Berger, 1998; Berger, Murphy, and Farber, 1999). R5 viruses predominate in the early stages of HIV-1 infection, whereas dualtropic and X4 variants, which are associated with rapid disease progression, emerge in the late chronic phase of disease in a significant proportion of patients (Connor et al., 1997; Scarlatti et al., 1997). It is well established that while all subtypes are capable of undergoing coreceptor utilization switch from CCR5 to CXCR4, this is less frequently encountered in HIV-1 subtype C infections, even in late stages of disease (Bjorndal and Sonnerborg, 1999; Cecilia et al., 2000; Cilliers et al., 2003; Ndung’u et al., 2006; Tscherning et al., 1998). Furthermore, expanded coreceptor usage beyond CCR5 and CXCR4 has also been occasionally reported but its significance for HIV-1 replication in vivo and disease progression is unclear (Aasa-Chapman et al., 2006; Cilliers et al., 2005; Dash et al., 2008).
Coreceptor utilization is genetically determined by sequence characteristics within the env gene, primarily specific amino acid changes within three of the five hypervariable regions namely the V1/V2 and V3 loops, as well as the number and pattern of predicted N-linked glycosylation sites (Fouchier et al., 1992; Pastore and Nedellec, 2006; Pollakis et al., 2001). Additional sequence changes within the env gene have also been implicated in coreceptor determination or the coreceptor switching process (Aasa-Chapman et al., 2006; Coetzer et al., 2008).
In several instances where HIV-1 subtype C isolates able to mediate cell entry via CXCR4 have been described, dualtropic (R5X4) strains that utilize both CCR5 and CXCR4 have been more frequently encountered compared to X4 monotropic viruses (Cilliers et al., 2003; Coetzer et al., 2006; Dash et al., 2008; Johnston et al., 2003; Ndung’u et al., 2006; van Rensburg et al., 2002). However, despite the occasional isolation of dualtropic HIV-1C viruses, such viruses have rarely been extensively characterized at both the functional and genetic clonal level. It is therefore largely unknown whether dualtropic HIV-1C strains represent a mixture of R5 and X4 viruses or truly dualtropic strains (or both) at the clonal level. Furthermore, genetic determinants associated with change in coreceptor usage have rarely been described for HIV-1 subtype C, particularly those that may reside outside of the V3 loop region. In this study we investigated whether dualtropic HIV-1C primary isolates represented truly dualtropic viruses at the clonal level, or mixed R5 and X4 clones. We describe the generation of functional envelope clones from dualtropic HIV-1C isolates and the sequence characteristics in the HIV-1C env gene, both within and outside of the V3 region that are associated with coreceptor utilization phenotype.
Seven HIV-1 subtype C dualtropic isolates from individuals in South Africa (Cilliers et al., 2003) and Botswana (Ndung’u et al., 2006) were selected for this study. In addition, a well-characterized CCR5-only utilizing HIV-1 subtype C isolate, BWM01_5 was used as a positive control (Ndung’u et al., 2006). Infection of the stimulated PBMCs was assessed by HIV-1 p24 antigen ELISA over a 14-day culture period. As shown in Fig. 1, p24 antigen concentration increased in culture supernatant for all the isolates. The isolates replicated to different levels and with different replication kinetics. On day 14, culture supernatants were removed and genomic DNA was extracted from the cells for env gene amplification.
We first analyzed the seven primary viral isolates propagated in PBMCs for their ability to use multiple coreceptors on cell lines. Specifically, we analyzed for ability to mediate cell entry via CCR5 or CXCR4 because these are the main coreceptors previously described for a significant proportion of HIV-1 primary isolates. In order to assess the ability of the isolates to utilize these coreceptors, virus equivalent to 2 ng of p24 antigen each was used to infect U87.CD4 glioma cell lines with or without the co-expression of the coreceptors. Table 1 shows the highest amounts of p24 antigen reached by the primary isolates over a 10-day period in culture. All 7 primary viral isolates replicated in cells expressing CXCR4 and CCR5. It was noteworthy that while all the isolates replicated efficiently in CXCR4 expressing cells, 3 isolates (RP1, SW30 and CM9) replicated to relatively low titers in cells expressing CCR5. Isolates CM9 and SW30 were previously shown to replicate efficiently in both CCR5 and CXCR4 expressing cell lines (Cilliers et al., 2003) and therefore our results could indicate that in vitro passages of the isolates is selecting against CCR5 utilization.
The env gene (approximately 3 kb) was then amplified from PBMC genomic DNA by PCR, gel purified and cloned into a mammalian cell expression plasmid vector (pcDNA3.1D/V5-His-TOPO). Five clones for each patient isolate were generated. Pseudoviruses were prepared by cotransfecting 293T cells with each of the 35 env plasmid clones with the pNL4-3.Luc.R-E-construct (Connor et al., 1995; He et al., 1995). This construct contains the infectious NL4-3 provirus backbone but is env and vpr deleted and nef has been replaced by the firefly luciferase gene. Virus supernatant from the transfected 293T cells equivalent to 2 ng p24 antigen was then used to infect U87 cells expressing the chemokine receptors CCR5 or CXCR4. A previously characterized dualtropic env clone Du179 (Coetzer et al., 2006) was used in parallel as a positive control. Productive entry of target cells was determined by measurement of firefly luciferase activity in cell lysates of infected U87 cells. All 35 env clones were able to efficiently infect cells expressing CXCR4 (Table 2). Five of 35 (14.3%) clones tested were also able to mediate entry via the CCR5 receptor, indicating that they were dualtropic. Two of the clones able to utilize CCR5 were from isolate RP1 while the other three were from isolate CM9. Surprisingly, there were no CCR5 monotropic clones detected from the bulk PCR envelope analysis of dualtropic isolates in this study.
We then reasoned that there were two possibilities to explain these results; one is that CCR5 monotropic envelopes are present at very low frequencies and therefore are virtually undetectable as clones from amongst the primary isolate viral quasispecies or that the CXCR4 viruses may be more fit and outcompeting CCR5 clones when propagated in PBMCs. In order to differentiate between these two possibilities, we infected U87.CD4.CCR5 cells with each of the primary viral isolates, with the objective of amplifying or favoring the CCR5-tropic clones in CCR5-only expressing cells. RNA was then extracted from the viral supernatant from U87.CD4.CCR5 cells, functional env clones generated and coreceptor usage determined as described for the PBMC-derived clones. All clones generated from U87.CD4.CCR5 cell supernatants showed dualtropism (data not shown). This result suggests that env clones using CCR5 only were virtually absent or present at very low frequencies within the quasispecies of the 7 primary isolates analyzed here.
We also considered the possibility that bulk PCR env amplification and cloning could result in resampling bias and explain the absence of CCR5-only env clones. We thus used a single genome amplification approach to generate diverse clones from isolate Du36_5. This clone was selected for this analysis because it showed a bias towards CCR5 utilization and yet we had failed to identify CCR5-only tropic clones from this isolate. Thirty clones of DU36_5 were amplified and cloned by this limiting endpoint dilution PCR approach. These clones were then tested for coreceptor usage in the U87.CD4.CCR5 and U87.CD4.CXCR4 cell lines. Of the 26 functional clones, 24 were dualtropic, one clone used CXCR4 exclusively and one clone showed exclusive R5-usage.
All 35 env full-length clones generated in this study by bulk PCR were sequenced to investigate phylogenetic relationships and to correlate coreceptor usage phenotype to genotype data. Phylogenetic analysis showed that all clones clustered with subtype C references with a high degree of confidence (Fig. 2). Furthermore, the clones from each primary viral isolate clustered together. As described above, sequences were also generated from U87.CD4. CCR5 and U87.CD4.CXCR4 cells infected with each of the primary isolates. These clones utilized both CCR5 and CXCR4, and their sequences were virtually phylogenetically indistinguishable from those obtained from PBMC cultures. Results obtained by using position-specific scoring matrix for HIV-1 subtype C (C-PSSM), a phenotype predictive tool based on HIV-1 subtype C sequences (Jensen et al., 2006) indicated CXCR4 or dualtropic phenotype and high net V3 charges (data not shown).
The envelope V3 loop is an important determinant of coreceptor utilization (Briggs et al., 2000; Cann et al., 1992; Fouchier et al., 1992; Rizzuto et al., 1998; Wu et al., 2006). Therefore, we further analyzed the V3 loop of the functionally characterized env clones in order to identify and describe sequences associated with dualtropism and CXCR4 utilization (Fig. 3 and Table 3). Of particular interest was the crown motif, a conserved tetrapeptide located at the tip of the V3 loop. Changes within this region may influence coreceptor usage. The consensus crown motif for clones from isolate RP1 was GPGQ, which is the conventional V3 loop crown sequence observed in CCR5-tropic subtype C sequences. The crown motifs for clones generated from Du36_5 and CM9 were GPGR and GPRY respectively, sequence substitutions that are indicative of CXCR4 tropism (Coetzer et al., 2006). Clones from BW17, TM1B and SW20 each displayed consensus crown motif sequences that read GRGQ. The consensus crown motifs of SW30, Du36_5 and CM9 read GRGH, GPGR and GPRY respectively. Thus CXCR4 utilization in HIV-1 subtype C is commonly associated with a basic amino acid substitution in the V3 tetrapeptide although this is not an absolute requirement.
Another feature of the env V3 loop associated with tropism determination is the property of amino acids at positions 11 and/or 25 (Fouchier et al., 1992). The consensus sequences for all isolates with the exception of SW20 showed a positively charged amino acid substitution at one or both of these positions. BW17 has serine (S) (neutral charge) and arginine (R) (positively charged); TM1B has asparagine (N) (neutral) and arginine, SW20 has serine and glutamine (Q) both of which carry neutral charges, SW30 has serine and lysine (K), Du36_5 has arginine and glycine (G) and CM9 has arginine and threonine (T) at positions 11 and 25 respectively. RP1 has arginine at both positions. The number of amino acids in the V3 loop can also be indicative of coreceptor usage. The typical V3 loop from CCR5 tropic viruses has 35 amino acids. Clones from CM9 were 35 amino acids long in the V3 loop, whereas clones from Du36_5 were 36 amino acids long. Clones from isolates TM1B, SW30 and BW17 had 2-amino acid insertions, increasing the length of the V3 loop to 37 amino acids. The insertions occurred at positions 13 and 14 of the V3 loop for clones from isolates RP1 and SW20 and at positions 6 and 7 for clones from Du36_5. Clones from TM1B, SW30 and BW17 had insertions between positions 15 and 16. Amino acid insertions in the V3 loop, particularly at positions 13 and 14 are features consistent with CXCR4 utilization as previously described (Coetzer et al., 2006). None of the insertions observed in the V3 loop of the clones from this study was noted in HIV-1 subtype C R5 sequences downloaded from the Los Alamos database (www.hiv.lanl.gov) (data not shown). The V3 region was also analyzed by manually calculating the overall net amino acid charge, another indicator of env coreceptor tropism (Table 3). C-PSSM, a web-based bioinformatic tool used for predicting HIV-1C coreceptor usage from the amino acid sequences of the V3 loop (Jensen et al., 2006) was also used. Both manual and C-PSSM calculations were comparable except for the clones from SW30 where calculated scores were slightly higher than C-PSSM generated scores. Higher overall net V3 charges are associated with X4-usage. A charge less than +4.5 is regarded as R5-using and charges above +4.5 are regarded as X4-using (Coetzer et al., 2006; Fouchier et al., 1995; Fouchier et al., 1992; Kuiken et al., 1992). Therefore, based on the multiple V3 loop sequence based algorithms available for phenotype prediction, all clones generated in this study were either only CXCR4-using or dualtropic, consistent with the functional data.
We next analyzed the V1/V2 and V4/V5 regions of the env gene as these regions have also been implicated in playing a role in viral tropism. Sequence features in these regions that may influence coreceptor utilization are the amino acid length and the number of predicted N-linked glycosylation sites (Chohan et al., 2005; Coetzer et al., 2007; Coetzer et al., 2008; Masciotra et al., 2002; Pollakis et al., 2001). The number of predicted N-linked glycosylation sites in clones from this study varied from 23 to 33. Clones for RP1, SW20 and SW30 all had 30 predicted N-linked glycosylation sites. Within the V1/V2 and V3 regions, the N-linked glycosylation sites varied between isolates but occurred at the same positions for all clones of the same isolate irrespective of whether they were X4-using or dualtropic except for one clone from TM1B which had 2 predicted N-linked glycosylation sites in the V2 region whereas the other 4 clones of this isolate had 3. The sites within the V4/V5 regions for all clones of all isolates showed slight variations in position. However, all clones from Du36_5 exhibited CXCR4-usage and showed variation in the positions of the sites in all five hypervariable regions. The positions of N-linked glycosylation sites varied from clone to clone and based on these positions no pattern emerged that could distinguish CXCR4-using clones from those that used both CCR5 and CXCR4.
When the total number of predicted N-linked glycosylation sites within the env as well as within the V1/V2 and V4/V5 regions was analyzed, no significant difference was observed between the CXCR4-using clones and dualtropic clones. However, the median number of N-linked glycosylation sites for X4/X4R5 clones from this study was significantly higher at (30) compared to (21) for R5 clones (30 sequences downloaded from the Los Alamos HIV-1 database) (p<0.0001) (Figs. 4A-C). R5 sequences showed a lower number of predicted N-linked glycosylation sites within the entire env as well as within the V1/V2 and V4/V5 regions when compared to R5X4/X4 clonal sequences.
Previous reports have suggested that a lack of predicted N-linked glycosylation sites at positions 6–8 of the V3 loop may be indicative of CXCR4-usage (Coetzer et al., 2006). We found this site to be conserved in the clones analyzed in this study, despite the utilization of CXCR4 by all the clones. All clones (except those from isolate Du36_5) contained a predicted N-linked glycosylation site at position 6 although they were CXCR4-using. This was also observed in a previous study by Johnston et al. (2003) where all but one X4 sequence maintained this site. We found a significant reduction in the number of predicted N-linked glycosylation sites within the V3 region of clones generated in this study as compared to the R5 sequences from the database (Fig. 4D).
The entire env sequence i.e. gp160 of all CXCR4- and CCR5/CXCR4-using clones were compared to determine if any distinguishing features could be identified. Specifically, we analyzed for unique signature patterns such as conservation of amino acids with a specific charge or physical property at a particular position, putative N-linked glycosylation sites, deletions, insertions or number of amino acids. The 2 clones of RP1 displaying dualtropism (i.e. clones #10 and 13) had leucine (L) at position 373 whereas the CXCR4-using clones of this isolate (clones #5; 6; 8) had proline (P) at this position (data not shown). The other isolate that produced clones exhibiting dualtropism was CM9. No distinguishing signature sequences were noted that could differentiate between X4 and X4R5 sequences.
We next analyzed for differences in the loop lengths between X4, X4R5 and R5 sequences. Most variation was seen in V1 which ranged from 16 to 27 amino acids. V2 had a relatively constant loop length (40-45). The combined V4/V5 loop length ranged from 37 to 46. The V1/V2 and V4/V5 loop lengths of the clones produced in this study were plotted against R5 sequences from the Los Alamos database. No significant differences were observed between the V1/V2 sequences of clones generated in this study and the R5 sequences from the Los Alamos database (Fig. 4E). However, for the V4/V5 region, there was a significant difference between the generated clones using the CXCR4 coreceptor for viral entry and the dualtropic clones (p =0.015) (Fig. 4F), with the dualtropic clones having an increased V4/V5 loop length. For the V3 loop, all analyzed R5 sequences had a loop length of 36 amino acids whereas X4 and R5X4 clones from this study showed variability with a range from 35 to 37 amino acids (Fig. 4G).
The requirement by HIV-1 for specific cellular interacting factors during the entry step offers an opportunity for the development of vaccines and drugs that target this crucial step in the virus replication cycle (Dhami et al., 2009; Hunt and Romanelli, 2009; Pantophlet and Burton, 2006; Phogat et al., 2007). Coreceptors play an important role in initiating infection at the cellular level. Additionally, coreceptor utilization is an important determinant of the rate of disease progression. The emerging availability of entry inhibitors such as the CCR5 antagonists underlines the importance of better characterization of coreceptor utilization and cellular tropism by HIV-1 isolates particularly in heavily burdened countries where the drugs are likely to be required on a large scale for the clinical management of HIV/AIDS. In this study, we generated 35 full-length env clones from seven dualtropic isolates of HIV-1 subtype C, in order to determine whether they were a mixture of CCR5 and CXCR4 quasispecies or dualtropic viruses at the clonal level. We also interrogated the sequence characteristics of these clones in order to better elucidate the genetic determinants of coreceptor utilization by HIV-1 subtype C viruses. We found that CXCR4-tropic clones dominated within the dualtropic viral isolates quasispecies. A minority proportion of dualtropic clones were also identified. Unexpectedly, we found that there was not a single CCR5-monotropic env clone from the seven primary isolates analyzed in this study. This is an unusual finding considering that many studies have shown that HIV-1 subtype C viruses even in late stages of disease utilize CCR5-only predominantly for cell entry. We thus expected to find a significant proportion of the remnants of these viruses among the quasispecies of the dualtropic isolates. Instead, all the clones detected in this study used CXCR4 as the coreceptor for cell entry, with a minority of these (14.3%) also able to mediate entry via the CCR5 receptor. Our results may explain why in previous studies of some of the dualtropic isolates described here (CM9, SW20 and SW30); the isolates could be strongly inhibited by CXCR4 inhibitors but only modestly by CCR5 inhibitors (Cilliers et al., 2003). These earlier results can now be explained by the observation that although these isolates are dualtropic, they are dominated by X4 variant clones.
An alternative explanation of our findings is that these isolates changed their coreceptor preference during in vitro passages in PBMC co-cultures as has been previously described (Voronin et al., 2007). This possible explanation is supported by the finding that isolates CM9 and SW30 displayed remarkably lower CCR5 utilization capacity (Table 1) than was previously described (Cilliers et al., 2003). It is also worth noting that although isolates BW17, SW20 and Du36_5 showed a possible bias towards CCR5 utilization and were clearly dualtropic, all the env molecular clones generated from these isolates by bulk PCR amplification were CXCR4-only using. This finding strongly suggested that the bulk PCR could be biased towards X4 viruses. We therefore performed limiting endpoint dilution PCR on one dualtropic viral isolate (Du36_5) which was biased towards CCR5 utilization (Table 1). Remarkably, of 26 functional env clones generated by this approach, 24 exhibited dualtropism, one used CXCR4 exclusively and one used CCR5 exclusively. We therefore conclude that dualtropic HIV-1 subtype C isolates are dominated by X4 and X4R5 clones with negligible proportion of R5 monotropic clones.
It has been recently proposed that coreceptor switching is associated with deleterious mutations in env that diminish CCR5-tropism as mutations associated with CXCR4 utilization accumulate (Coetzer et al., 2008). Although we did not directly test for coreceptor binding in this study, our results are consistent with the proposal by Coetzer et al. and with their observation that coreceptor switching is associated with a rapid decrease in the ability to use CCR5. Our results may suggest that in HIV-1 subtype C, the mutations required for adaptation to CXCR4 utilization significantly reduce the ability of env to utilize CCR5, thus resulting in reduced fitness of CCR5 utilizing viruses. This could in turn lead to the selection and amplification of clones able to utilize CXCR4. We can speculate that given the low frequency of HIV-1 subtype C CXCR4 utilizing viruses reported in various studies, more accumulated mutations are required for switching to CXCR4 utilization for this subtype. Alternatively, the changes required for a switch to CXCR4 utilization may result in a bigger fitness deficit for HIV-1 subtype C CCR5-tropic variants thus leading to selection against these viruses once adaptation to CXCR4 utilization has been accomplished. Further studies will be required to carefully investigate the specific localization and nature of complementary mutations required for HIV-1 subtype C env coreceptor switch.
We also investigated the genetic characteristics associated with CXCR4-usage or dualtropism for HIV-1 subtype C viruses. Our results may be limited by founder effects since we could not generate R5 sequences from the study isolates but we nevertheless had HIV-1 subtype C R5 sequences available from the Los Alamos database that facilitated this comparative analysis. As described for HIV-1 subtype B, the subtype C third variable loop of gp120 (V3 region) is a major determinant of whether CXCR4 or CCR5 will be the accessory protein used by the virus for membrane fusion (Cilliers et al., 2003; Coetzer et al., 2007; Coetzer et al., 2006; Fouchier et al., 1992; Morris et al., 2007; Ndung’u et al., 2006). Typically, the V3 region consists of approximately 35 amino acids in CCR5-tropic viruses (Coetzer et al., 2006). Consistent with earlier studies, we found that V3 loop amino acid characteristics are important determinants of coreceptor tropism. In most cases, we found that the V3 loop crown of CXCR4-utilizing clones had basic amino acid substitutions which differed from the canonical GPGQ sequence found in CCR5 HIV-1 subtype C viruses to GPGX (where X is any other amino acid), GRGH, GPGR or GPRY. X4 variants are more variable than R5 viruses in the V3 region particularly at positions 11 and 25 which tend to be mostly positively charged amino acids, often arginine (R), lysine (K) or histidine (H). We found the presence of a basic amino acid at both or one of these positions for the majority of clones in this study (85%). In addition, in X4 variants there may be insertions particularly between positions 13 and 14 of the V3 loop contributing to an increased length. We found amino acid insertions in 71% of clones in this study. X4 variants may also be distinguished from R5 viruses as they usually have an increased net V3 charge. Consistent with these observations, we found that 100% of X4-utilizing clones had V3 loop amino acid charges of +5 or more.
The V3 region however, is not the exclusive determinant of coreceptor usage and other regions within the env gene may also contribute to viral tropism. The V1/V2 and V4/V5 regions have been implicated in playing a role in determining the biological phenotype of the virus. Specifically, the number of N-linked carbohydrate moieties in these variable loops has been associated with coreceptor determination (Chohan et al., 2005; Coetzer et al., 2008; Masciotra et al., 2002; Pollakis et al., 2001). Here we found a strong association between the number of N-linked glycosylation sites and coreceptor utilization with X4 clones having a significantly higher number of these sites than R5 clones from the database overall and in the V1/V2 or V4/V5 regions (Figs. 4A-C). In contrast in the V3 region, the number of sites was significantly higher in R5 sequences than X4/X4R5 sequences (Fig. 4D). A previous longitudinal study of HIV-1 env evolution showed no significant changes in N-linked glycosylation sites of 23 viral isolates from 5 patients followed for 2-4 years (Coetzer et al., 2007). Therefore our findings may suggest a rapid accumulation of N-linked glycosylation sites as coreceptor tropism switches, as opposed to a slow accumulation of these sites over time. This is consistent with recent findings of rapid decline in CCR5 utilization as alternate coreceptor utilization emerges in HIV-1 subtype B infection (Coetzer et al., 2008). In both HIV-1 subtypes A and C, shorter V1/V2 loop sequences and fewer predicted N-linked glycosylation sites have been correlated with preferential heterosexual viral transmission (Chohan et al., 2005; Derdeyn et al., 2004). We did not find significant differences in V1/V2 length between R5 and X4 clones in this study but a trend towards shorter V4/V5 for X4 clones was noted (Figs. 4E-F). Further longitudinal studies will be necessary in order to better understand HIV-1 subtype C transmission, coreceptor switching and the env genetic characteristics associated with these processes. Overall, our results suggest that sequence characteristics in the V3 loop, the V4/V5 loop length as well as the number of env predicted N-linked glycosylation sites are the primary genotypic determinants for viral tropism in HIV-1 subtype C.
It is worth noting that we did not perform limiting endpoint dilution of samples in this study except for isolate Du36_5. Therefore we cannot completely rule out the presence of substantial frequencies of R5-monotropic viruses in the quasispecies of the isolates where endpoint dilution was not used. However, the absence of these clones in bulk amplified clones, in CCR5 only expressing cells and in endpoint diluted Du36_5 isolate that is biased towards CCR5 is all suggestive of absence of such quasispecies or presence at very low frequency. Our results appear to contradict the recent findings of Irlbeck et al. (2008) but it must be emphasized that in that study, samples were analyzed directly from plasma in contrast to our study in which we examined in vitro propagated isolates. Further studies will be needed to determine whether env clones directly obtained from patients with dualtropic HIV-1 subtype C viruses have a bias towards CCR5 or CXCR4 tropism.
In conclusion, we show in this study that dualtropic viral isolates consist of predominantly X4 and X4R5 clones. Thirty of 35 env clones analyzed from PBMCs utilized X4 only as the coreceptor for entry into cells, whereas 5 of 35 clones tested displayed dualtropism and no CCR5-only utilizing clones were identified. R5 monotropic clones could not be detected even when the isolates were cultured in cells expressing CCR5 coreceptor only. We also failed to detect a significant number of R5 monotropic clones when we changed our approach of viral amplification from bulk PCR to limiting endpoint dilution PCR for one dualtropic isolate showing bias towards CCR5 tropism. Viral env sequences from both CXCR4 and CCR5-expressing cells were indistinguishable and possessed X4/dualtropic characteristics. Furthermore, we describe env sequence characteristics associated with CXCR4 utilization in HIV-1 subtype C. In addition to sequence changes in the env V3 region, we identify the number of N-linked glycosylation sites in the V1/V2, V3 and V4/V5 regions as major determinants of coreceptor utilization in HIV-1 subtype C. We also show that the length of the V4/V5 is a possible determinant of coreceptor utilization. We note that our results are consistent with recent findings of the rapid loss of fitness of CCR5 envelope as coreceptor switching emerges and suggest that the sequence characteristics associated with coreceptor switch must occur rapidly in vivo. Further studies are needed to better characterize coreceptor switching, particularly in the context of HIV-1 subtype C, the predominant subtype in the world. We have generated 35 full-length CXCR4- or dualtropic clones of HIV-1 subtype C, important reagents that will facilitate further functional studies of this globally predominant subtype. Our results have important implications for coreceptor antagonist design and application, and further contribute to better understanding of HIV-1 pathogenesis.
Seven primary viral isolates were analyzed in this study. BW17 is a dualtropic HIV-1C virus isolated in 1996 from an infected person with acquired immunodeficiency syndrome (AIDS) in Botswana (Ndung’u et al., 2006). TM1B; RP1; SW20; SW30; CM9 and Du36_5 were obtained from the AIDS Virus Research Unit, National Institute for Communicable Diseases, Johannesburg, South Africa and were from patients at various disease stages: acute infection (Du36_5), slow progressor (TM1B), rapid progressor (RP1) and AIDS (SW20; SW30; CM9) (Choge et al., 2006; Cilliers et al., 2003; Coetzer et al., 2006).
U87.CD4 cells with or without the expression of the chemokine receptors CCR5 or CXCR4 were obtained from the NIH AIDS Research and Reference Reagent Program (Rockville, MD). U87.CD4 cells without chemokine receptors were cultured in Dulbecco’s modified eagles medium (DMEM) containing l-glutamine (Gibco, NY, USA) supplemented with 15% heat inactivated fetal bovine serum (FBS) (Gibco, NY, USA), 300 μg/ml G418 (Sigma, Germany) and 50 μg/ml penicillin-streptomycin (Gibco, NY, USA). U87.CD4 cells expressing CCR5 or CXCR4 were propagated in the same medium but additionally supplemented with 1 μg/ml puromycin (Sigma, Germany). 0.5×106 cells were cultured in 6-well flat-bottomed plates in a total of 2 ml culture medium at 37 °C and 5% CO2.
293T cells were obtained from the NIH AIDS Research and Reference Reagent Program (Rockville, MD) and were cultured in DMEM containing l-glutamine supplemented with 10% heat inactivated FBS and 50 μg/ml gentamicin (Sigma, Germany). 50,000 293T cells per well were seeded in a total volume of 0.3 ml per well of a 48-well flat-bottomed plate. These cells were incubated at 37 °C in 5% CO2 overnight before transfection.
PBMCs from anonymous low risk HIV-negative volunteers were separated by density-gradient centrifugation on Ficoll-Paque (Amersham Biosciences, Uppsala, Sweden). We confirmed that the samples were HIV-negative by performing HIV RNA testing on the plasma (Ampliscreen v1.5, Roche Diagnostics, Rotkreuz, Switzerland). 5×106 PBMCs from 2 donors were combined and stimulated by culturing in RPMI 1640 with penicillin–streptomycin (50 μg/ml and 50 U/ml), 10% heat inactivated FBS, 5 μg/ml phytohaemagglutinin (PHA) (Sigma, Germany) and 20 U/ml interleukin-2 (IL-2) (Roche Applied Science, Germany) at 37 °C and 5% CO2 for 72 h in a T-25 flask. For infection of the stimulated PBMCs, 5 ng p24 antigen equivalent of virus was used. On days 1, 4, 7, and 10, 50% of the media was removed and replaced with fresh medium. Aliquoted supernatant was retained for quantification of p24 antigen as previously described. On day 14, supernatant was removed and preserved for p24 antigen quantification. Cells were harvested and resuspended in 200 μl PBS. DNA was then extracted using the QiaAmp DNA Blood Mini kit (Qiagen, Germany).
PBMC-grown virus corresponding to 2 ng of p24 HIV-1 antigen was used for infection of U87.CD4 cells expressing either CCR5 or CXCR4. On days 0, 4, 7 and 10 half of the media was removed and replaced with fresh medium. The removed supernatant was retained for quantification of p24 antigen using the Vironostika HIV-1 Antigen Microelisa system (Biomerieux, Boxtel, Netherlands). Previously well-characterized dualtropic (Du179) and CCR5-tropic (BWM01_5) primary isolates were used as positive controls.
The 3 kb env gene was amplified by polymerase chain reaction (PCR) using Phusion Hot Start High-Fidelity DNA Polymerase (Finnzymes, Finland) with the following primers: Env1Adir 5′-CACCGGCTTAGGCATCTCCTATGGCAGGAAGAA-3′ and EnvM 5′-TAGCCCTTCCAGTCCCCCCTTTTCTTTTA-3′. The forward primer Env1A-dir was designed to include the 4 base pair sequence (CACC) necessary for directional cloning on the 5′ end. Cycling conditions were as follows: a 5 minute denaturation at 94 °C, followed by 35 cycles of 30 s at 94 °C, 30 s at 55 °C, and 4 min at 72 °C. The final extension was at 72 °C for 10 min. The amplified product was then run on a 1% agarose gel and gel purified using the Qiaquick gel extraction kit (Qiagen, Germany).
Once the DNA was purified, the env of each primary viral isolate was cloned into the pcDNA3.1D/V5-His-TOPO vector (Invitrogen, Carlsbad, CA). For the transformation procedure, Stratagene XL-10 Gold Ultracompetent cells (Stratagene, USA) were used. Molecular clones were screened as follows. A colony PCR was performed after the colony was incubated for 1.5 h at 37 °C with continuous shaking (225 rpm) in a 96-well plate containing 100 μl Luria Bertani (LB) media (Sigma, Germany) and ampicillin (100 μg/ml) (Calibiochem, Darmstadt, Germany) to determine positivity of the cloned insert. This is a directional env insert-specific PCR as it uses the forward primer T7 (5′ TAATACGACTCACTATAGGG 3′) found on the vector and reverse primer env M which is env specific. SuperTherm Taq Polymerase (Southern Cross Biotechnology, Cape Town, South Africa) was used and cycling conditions were as follows: denaturation for 5 min at 94 °C, followed by 35 cycles of 30 s at 94 °C, 30 s at 55 °C, and 4 min at 68 °C. The final extension was for 10 min at 68 °C. The amplified products were then run on a 1% agarose gel. Clones were considered positive if they yielded a 3 kb band on an agarose gel. These clones were then grown up at 37 °C overnight with shaking in 3 ml LB broth containing 100 μg/ml ampicillin. Plasmid DNA was then isolated using Qiaprep Spin Miniprep kit by following the manufacturer’s instructions (Qiagen, Germany). The first five functional env clones identified from each isolate were selected for further analysis.
A limiting endpoint dilution PCR was performed on one primary viral isolate, Du36_5 in order to determine whether the bulk PCR resampling bias resulted in clones biased towards either CCR5 or CXCR4 coreceptor usage. Single genome amplification was undertaken as previously described (Salazar-Gonzalez et al., 2008). Primers and cycle conditions were the same as used in the bulk PCR reactions. Once confirmed by agarose gel electrophoresis, 30 PCR products were purified and cloned and coreceptor usage was tested for the 26 clones that were functional.
Cotransfection was carried out by first combining 50 μl serum free DMEM and 2.5 μl Fugene reagent (Roche Applied Science, Germany) and incubating for 5 min at room temperature. This was then incubated at room temperature together with 0.6 μg gp160 env DNA (i.e. cloned product) and 0.3 μg pNL4-3.Luc.R-E-(Connor et al., 1995; He et al., 1995). pNL4-3.Luc.R-E- is a full-length HIV plasmid with two frameshifts that render the clone env and vpr deleted. The reporter firefly luciferase gene has been inserted into the nef gene. The transfection mixture was incubated for 30 min, and then added to assigned wells of the plate seeded with 293T and incubated at 37 °C and 5% CO2 for 48 h. The supernatant together with 7.5 μg/ml DEAE-Dextran (Sigma, Germany) was added to U87. CD4 cells as well as U87.CD4 cells expressing the coreceptors CXCR4 or CCR5. This was incubated at 37 °C and 5% CO2 for 48 h. The cells were lysed using Glo Lysis buffer (Promega, Madison, WI, USA) and incubated with Bright-Glo Assay reagent (Promega, Madison, WI, USA). The luciferase activity was then determined using the Turner-Biosystems Modulus Microplate instrument (Promega, Madison, WI, USA). A negative control consisting of the plasmid pNL4-3.Luc.R-E- and various positive controls were used in each coreceptor expressing cell line. The positive controls were previously characterized env clones Du179 (dualtropic), 96BWM01_5 (R5), and 96BW17#10 (X4). In addition, for each assay plate, 100 nM of CCR5 inhibitor RANTES and 500 nM of CXCR4 inhibitor AMD3100 were used with the respective controls to confirm specificity of entry into target cells. Experiments were done in duplicate and the average relative luminescence units (RLUs) for each clone were calculated. A positive result was considered to be twice the average of the negative control plus standard deviation.
The env gene was sequenced after cloning using the ABI PRISM Big Dye Terminator Cycle Sequencing Ready Reaction kit version 3.1 (Applied Biosystems, CA, USA). Sequences were assembled and edited using Sequencher 4.8. They were then aligned with Mega 4. Phylogenetic trees were constructed in Paup 4.0 to evaluate the clustering of these sequences with each other and with subtype references. Phylogenetic trees were then visualized using Treeview 1.6.6. The consensus sequence for clones from each isolate was generated using BioEdit Sequence Alignment Editor Software (Tom Hall, North Carolina State University). Coreceptor utilization was predicted using the web-based subtype C-specific position-specific scoring matrix (C-PSSM) programme (http://indra.mullins.microbiol.washington.edu/pssm/), a bioinformatics tool that reliably predicts coreceptor phenotype using V3 loop sequences (Jensen et al., 2006). Predicted N-linked glycosylation sites were examined using the web-based programme N-GLYCOSITE (www.hiv.lanl.gov). All reference sequences were obtained from the Los Alamos database (www.hiv.lanl.gov).
The sequence data obtained from this study have been submitted to Genbank under the following accession numbers: FJ846629-FJ846662 and selected clones will be deposited into the NIH AIDS Research and Reference Reagent Program repository.
This study was supported by grants from the Hasso Plattner Foundation, the South African Department of Science and Technology/National Research Foundation (DST/NRF) Research Chair Initiative and the South African AIDS Vaccine Initiative (SAAVI). AS was supported by the Columbia University-Southern African Fogarty AIDS International Training and Research Programme (AITRP) funded by the Fogarty International Center, National Institutes of Health (grant # D43TW00231) and by a training grant from the East Coast Biotechnology Research Innovation Centre (Lifelab), funded by the South African Department of Science and Technology. TN holds the South African DST/NRF Research Chair in Systems Biology of HIV/AIDS. We thank Dr Johannes Viljoen and the Africa Centre for providing us access to the sequencing facility and the biosafety level 3 tissue culture laboratory. We thank Walter Campos for the kind donation of AMD3100 and RANTES. We thank Dr Zabrina Brumme for critical review of this manuscript.