In this study we have explored the functional properties of Env proteins from contemporaneous HIV-1 biological clones isolated from the plasma of chronically infected patients. Even when comparisons were restricted to viruses with similar tropism, these genetically diverse Env proteins were found to exhibit striking functional diversity, including, i) a wide range of infectivities for a given target cell, ii) differences in relative ability to infect different target cells, and iii) differences in sensitivity to certain entry inhibitors. In addition to the functional diversity observed among viruses from a single patient, significant inter-patient differences were also observed for some characteristics of Env proteins, including sensitivity to several different entry inhibitors and the relative ability of viruses to infect different target cells. Interestingly, no correlation was observed between viral infectivity and sensitivity to entry inhibitors, indicating that these properties are, to some extent, dissociable. These findings demonstrate the marked functional heterogeneity of HIV-1 Env proteins expressed by contemporaneous circulating viruses, and underscore the advantage of clonal analyses in characterizing such viral populations.
The observation that Env proteins expressed by contemporaneous viruses display a broad spectrum of infectivity towards a given target cell raises the question of the forces responsible for this functional diversity. One factor that can drive the development of diversity is the selection in a single patient of numerous viral subpopulations with distinct Env functional properties. Genetically distinct viral populations replicating in different tissues or cellular compartments have been identified in numerous studies [50-54]. Such populations can also result in functional diversity, as illustrated by the obvious example of the coexistence of viruses with different tropism. Indeed, we found that the coexistence of viruses with R5 and X4 tropism could make an important contribution to overall genetic diversity in patients from whom both populations were isolated.
The high mutation and recombination rates of HIV-1 are also likely to have an important impact on infectivity. Because of the numerous structural constraints in viral proteins, including Env, many mutations will prove to be deleterious for fitness. Similarly, recombination events that shuffle env segments can modify the function of Env proteins [55,56]. Indeed, genetic evidence supports the idea that purifying selection against deleterious mutations is occurring in the env region . In this regard, we observed that 3 of 4 viruses in which V3 sequences contained a single amino acid polymorphism not seen in any other sequence from that patient had low infectivity. The association of such rare polymorphisms with poor infectivity is consistent with the possibility that the deleterious mutations contribute to the wide spectrum infectivity observed in these studies, although further studies are required to directly demonstrate the impact of such specific genetic differences on viral infectivity.
The extent that genetic differences deleterious for viral replication accumulate because they are associated with escape from immune responses is also an important issue. In this regard, we observed that a V2 haplotype associated with reduced viral infectivity emerged in patient 2 over an 8 month interval, and had become the dominant haplotype. This haplotype differed by a single amino acid change from another haplotype observed in this individual (Fig. 8). Such a modification would be typical of escape from neutralizing antibodies or cytotoxic T-cells targeting this epitope, although other explanations are also possible (e.g., adaptation of the Env proteins to improve the targeting of a specific cell type or fixation of a deleterious mutation through stochastic processes). Further studies comparing the spectrum of infectivity of Env proteins present in the genetically homogeneous viral populations of viruses present immediately before seroconversion with that observed following the development of the anti-viral immune response will help define the extent that selection of variants associated with immune escape affects Env function.
The possibility that viral evolution during in vitro culture contributed to diversity should also be considered. For example, greatest viral diversity was observed for patient 3, and clones from this patient emerged later than those from the other patients. However, control experiments (see Materials and Methods) showed no evidence for viral evolution during culture. In particular, the absence of polymorphic bases in the sequences of the viral clones indicates that variants were not emerging to the extent that they were detectable by bulk sequencing, and therefore were not sufficiently abundant to influence either viral diversity or the assessment of infectivity.
The spectrum of Env infectivity in vivo is almost certainly greater than observed in our study. The Env proteins studied by us were derived from viruses capable of productively infecting MT4-R5 cells. We must assume that the env sequences of these viruses were enriched for those with high infectivity in this cell system, and Env proteins carrying lethal mutations or mutations preventing efficient propagation in MT4-R5 cell cultures would not be sampled. We found that recombinant viruses produced using bulk env sequences amplified directly from plasma by RT-PCR had an infectivity that was lower than that of recombinant viruses carrying Env proteins from the clonal viruses derived from the same sample, as would be expected if a significant proportion of viruses present in plasma carry Env proteins that are noninfectious or whose infectivity is too low to permit emergence of clones during in vitro culture. The finding in previous studies that a substantial portion of viruses expressing env sequences amplified from plasma by RT-PCR show low or undetectable infectivity is also compatible with this interpretation [19-23,53,58]. It should be emphasized, however, that additional artifacts may also contribute to the low infectivity observed in our study for viruses carrying bulk env sequences amplified from plasma by RT-PCR, such that the infectivity of the recombinant viruses would not be reflective of viruses in plasma. First, recombinant viruses formed using mixtures of sequences amplified from plasma will express heterotrimeric Env proteins, including, for some samples, trimers containing sequences with both X4 and R5 tropism. In this regard, our preliminary results suggest that the infectivity of viruses generated using mixtures of clonal env sequences may be lower than the mean infectivity of viruses expressing each of these env sequences as homotrimers. In addition, during the amplification of env sequences from plasma, recombination between the heterogeneous target sequences can occur, and may form sequences in which incompatibilities between Env segments are deleterious to infectivity. In view of these uncertainties, further studies will be required to fully define the spectrum of infectivity of Env proteins expressed by viruses in plasma.
An interesting observation from our study was the absence of correlation between the infectivity of the recombinant viruses studied and their susceptibility to several different entry inhibitors or neutralizing antibodies. The affinity of Env-coreceptor interactions is one factor that can influence both Env fusion kinetics and the sensitivity of Env to the inhibition by enfuvirtide and TAK-779 [35,36,56,59], although mutations that modify sensitivity to co-receptor antagonists without modifying fusion kinetics have also been described [36,60]. The failure to find a correlation between Env infectivity and sensitivity to these inhibitors suggests that differences affecting membrane fusion that are independent of Env-coreceptor affinity (e.g., "fusogenicity") or differences affecting other Env properties (e.g., the expression or stability of Env trimers) make an important contribution to the wide spectrum of Env infectivities observed in this study.
We did observe, however, that sensitivity to inhibition by soluble CD4, enfuvirtide, and three different neutralizing antibodies were properties that were shared by viruses from a given patient, independent of their infectivity. Similarly, Ray et al. found differences in enfuvirtide sensitivity comparing Env clones isolated from different patients . These findings suggest that genetic determinants important in defining the sensitivity to these entry inhibitors lie in regions that are not necessarily subject to extensive diversity. For example, determinants in the relatively well conserved membrane proximal ectodomain of gp41 appear to be important in determining sensitivity to both enfuvirtide and the monoclonal antibody 2F5 [61,62]. Determinants of sensitivity to inhibitors of coreceptor binding appear to be subject to greater intra-patient variability. No patient-specific differences in sensitivity to TAK-779 was observed in our study, and considerable variability in sensitivity of individual Env clones from a given patient has been reported for several other coreceptor antagonists . Thus, the impact of viral diversity on sensitivity to entry inhibitors is likely to differ for inhibitors with different modes of action. Our findings suggest, however, that resistance to entry inhibitors is not likely to be a useful surrogate marker for viral infectivity.