Human immunodeficiency virus type 1 biological variation and coreceptor use: from concept to clinical significance


Eva Maria Fenyö, Division of Medical Microbiology, Department of Laboratory Medicine, Lund, Lund University, Sölvegatan23, 223 62 Lund, Sweden.
(fax: +46 46 176033; e-mail:


Abstract.  Fenyö EM, Esbjörnsson J, Medstrand P, Jansson M (Division of Medical Microbiology, Lund; Lund University, Lund; and Karolinska Institutet, Stockholm, Sweden). Human immunodeficiency virus type 1 biological variation and coreceptor use: from concept to clinical significance (Symposium). J Intern Med 2011; doi: 10.1111/j.1365-2796.2011.02455.x

There is ample evidence for intra-patient evolution of the human immunodeficiency virus type 1 (HIV-1) biological phenotype during the pathogenic process. Evolution often involves switch of coreceptor use from CCR5 to CXCR4, but change to more flexible use of CCR5 occurs over time even in patients with maintained CCR5 use. The increasing use of entry inhibitors in the clinic, often specific for one or the other HIV-1 coreceptor or with different binding properties to CCR5, calls for virus testing in patients prior to treatment initiation. Cell lines expressing CCR5/CXCR4 chimeric receptors are tools for testing viruses for mode of CCR5 use. It is conceivable that small-molecule entry inhibitors that differentially bind to CCR5 can be matched for best effect against HIV-1 with different modes of CCR5 use, thereby allowing an individualized drug choice specifically tailored for each patient.


Early in the epidemic, biological variation of human immunodeficiency virus type 1 (HIV-1) has been closely associated with the pathogenic process. During the first decade following the discovery of HIV-1, biological variation was described in terms of replicative capacity and cytopathic effect in primary cells and established cell lines in vitro (Table 1) [1–5].

Table 1. HIV-1 phenotype markers
HIV-1 phenotypeMost often isolated from
  1. SI, syncytium inducing; NSI, nonsyncytium inducing.

  2. aT-lymphoid (CEM, H9, MT-2) and monocytoid (U937) cell lines.

Replication in PBMCsSyncytia induction in PBMCsReplication in established cell linesa 
Rapid/highYes (SI)YesPatients with AIDS
Slow/lowNo (NSI)NoNon-AIDS patients

It then became clear that most viruses isolated from patients with acquired immunodeficiency syndrome (AIDS) replicated rapidly to high titres (rapid/high; R/H) and induced syncytia (SI) in peripheral blood mononuclear cells (PBMCs), whereas viruses from all other HIV-1-infected individuals replicated slowly to low levels (slow/low; S/L) and did not induce syncytia (NSI) [1, 4, 5]. HIV-1 with the latter phenotype failed to replicate in established cell lines [1, 3, 6]. For example, the capacity of the virus to replicate in MT-2 cells [7] positively correlated with the decline in CD4+ T cells in blood and the severity of clinical progression (Fig. 1). A relationship between HIV-1 phenotype and severity of infection was thus established already by the early 1990s.

Figure 1.

Relationship between HIV-1-1 phenotype and severity of infection. Study duration: 80 months. Neg/Neg: patient’s HIV-1 isolates MT-2 negative throughout study; Neg/Pos: phenotype switch from MT-2 negative to MT-2 positive; Pos/Pos: HIV-1 isolates that were MT-2 positive throughout study. Data from Karlsson et al. [28].

Chemokine receptor use corresponds to HIV-1 biological phenotype

An explanation for HIV-1 biological phenotypes was provided by the recognition that HIV-1 entry into cells requires, in addition to CD4, a secondary structure, which generally belongs to the chemokine receptor family [8–12]. Binding of the viral envelope to CD4 serves as a key to open up the closed envelope structure, thereby exposing the coreceptor-binding site buried within the envelope structure (reviewed in [13]). In the case of HIV-1, the major coreceptors are CCR5 and CXCR4 (reviewed in [14]). Because S/L or NSI viruses bind to CCR5, whereas R/H or SI viruses bind to CXCR4 exclusively or in addition to CCR5, a new nomenclature based on receptor use was introduced for HIV-1 phenotypes [15–17]. Accordingly, it is possible to distinguish between HIV-1 with R5 and X4 phenotypes (Fig. 2). Dual-tropic viruses, R5X4, able to bind both receptors have been observed in vitro but in vivo use of receptors by these apparently dual-tropic viruses seems to be highly variable [18]. In addition, HIV-1 isolates may be composed of mixed populations of monotropic R5 and X4 viruses, as well as true dual-tropic R5X4 virus variants and are therefore often referred to as dual-tropic/mixed isolates. The constitutive expression of CXCR4 in PBMCs and T-lymphoid and monocytoid cell lines and, particularly, the strong upregulation of CXCR4 following in vitro PBMC stimulation, allows immediate and vigorous replication of HIV-1 using this receptor. By contrast, the slow appearance of CCR5 in PBMC cultures or the lack of expression in cell lines slows down or prevents replication of HIV-1 with the R5 phenotype [19].

Figure 2.

Differentiation between HIV-1 biological phenotypes by coreceptor use and proposed signal transduction pathways.

Chemokines are small protein molecules (65–95 amino acids) that, based on their primary structural similarities and chromosomal localization, belong to the same superfamily. They contain four cysteine residues, and the proximity of the first two cysteines determines the specificity of receptor binding. Accordingly, CC chemokines (no intercalating amino acid) bind to CC receptors, whereas CXC chemokines (where X denotes an intercalating amino acid) bind to CXC receptors. CC and CXC receptors are structurally similar 7-transmembrane G-protein-coupled receptors, but have only 32–34% amino acid similarity and differ in function (reviewed in [20]). CXC chemokines and receptors have housekeeping functions and are expressed constitutively, whereas CC chemokines and receptors are pro-inflammatory and inducible. CC chemokines, CCL3, CCL4 and CCL5 (previously known as MIP-1α, MIP-1β and RANTES, respectively), are ligands of CCR5 and were first shown to be important in HIV-1 biology when it was reported that these specific chemokines could inhibit in vitro replication of certain HIV-1 strains [21]. Furthermore, studies of genetic polymorphisms in regard to segmental duplication of the gene encoding CCL3L1, an HIV-1 inhibitory CCR5 ligand, have suggested that elevated expression levels of CCR5 ligand chemokines also may play role in vivo by reducing susceptibility to both HIV-1 infection and AIDS development [22]. The important role of CCR5 for HIV-1 host cell entry in vivo, i.e. HIV-1 resistance, has also been extensively demonstrated by genetic polymorphisms linked to expression of CCR5, most prominently the 32 bp deletion in the CCR5 gene (known as CCR5Δ32; reviewed in [23]).

Chemokine receptor signalling in HIV-1 infection

Much is known about the signal transduction elicited by the binding of natural ligands to their cognate receptors, i.e. CCL3, CCL4 and CCL5 to CCR5 and CXCL12 (SDF-1) to CXCR4. Chemokine receptor signalling is known to be diverse and is coupled to distinct signalling pathways that mediate cell migration, transcriptional activation, and cell growth and differentiation (reviewed in [24]). Whether signalling after HIV-1 binding to either CCR5 or CXCR4 is required for productive infection has long been a controversial issue. Recent results show that the activation state of the cell used for experiments influences the results such that resting CD4+ T cells have a signalling requirement for latent HIV-1 infection [25]. In addition to the functions of viral binding and entry, CXCR4 signalling appears critical for viral DNA stability and nuclear migration. HIV-1 envelope–CXCR4 interaction signals through one of the G-protein signalling pathways (Gαi) and thereby promotes cortical actin dynamics. This is achieved through activation of an actin depolymerizing factor, cofilin. T cell migration and activation are critically regulated by actin dynamics, and HIV-1 exploits this cellular machinery. Yoder and colleagues have extended their in vitro studies to include resting T cells from HIV-1-infected patients, and show that these cells carry elevated levels of active cofilin, suggesting that the cells have been primed in vivo with regard to the level of active cofilin in the cells, to facilitate HIV-1 infection [26]. Indeed, it seems that the virus-hijacked signalling process may facilitate pathogenesis [27]. Similarly, data are accumulating in favour of the importance of signalling in the viral life cycle at HIV-1 envelope–CCR5 interactions as well. The current understanding in the field has recently been reviewed [24].

Evolution of HIV-1 coreceptor use during pathogenesis

Early studies established intra-patient evolution of the HIV-1 biological phenotype [28–30]. As soon as HIV-1 coreceptor use was discovered, phenotypic variation seen earlier could be explained by coreceptor use. Accordingly, early in infection HIV-1 almost exclusively uses CCR5 as a coreceptor, whereas the virus population may switch or broaden its coreceptor use to include CXCR4 during the disease course [16, 31]. In most cases, CXCR4 use appears in late-stage disease, close to the onset of AIDS. The prevalence of HIV-1 CXCR4 tropism has been debated, and it is commonly suggested that it appears in approximately 50% of infected individuals [32]. However, this is a very rough estimate based on early reports investigating the dynamics of coreceptor tropism. The approximation of 50% is biased by either the low numbers of study subjects or by combined analyses of subjects at different disease stages [33, 34]. In addition, these early studies were mainly performed in individuals infected with HIV-1 of the genetic subtype B (the dominant form in Europe and North America, which represents approximately 10% of global HIV-1 infections) [35]. More recent studies have indeed demonstrated that the genetic subtype is important for the coreceptor evolution throughout the disease course [36–40]. At least two subtypes, C and D (contributing to approximately 50% and 3%, respectively, of infections worldwide [35]), seem to stand out compared to other subtypes or circulating recombinant forms (CRFs). Whilst subtype C has shown low levels of CXCR4 tropism (0–30%) amongst patients with late-stage disease, subtype D has been reported to display a higher extent of CXCR4 tropism early in infection [36, 38, 39, 41]. The underlying mechanisms of these subtype-specific differences are largely unresolved and need to be further explored.

To further investigate HIV-1 subtype-specific differences in the ability to use CXCR4 during the early asymptomatic disease stage (defined by CD4+ T cell counts >200 cells μL−1) compared to late-stage disease or AIDS (defined either clinically or by absolute levels of CD4+ T cells <200 cells μL−1), we conducted an extensive literature analysis (List of references provided as supporting information). The analysis included samples from 1134 patients (seven different HIV-1 subtypes collected over 25 years 1983–2008). This analysis revealed a high frequency of HIV-1 R5X4 or X4 populations in late-stage disease amongst all subtypes analysed, except for subtypes C and G (Fig. 3). During the asymptomatic phase of infection, none of the subtype C viruses used CXCR4, whilst CXCR4 use was more frequent (34%) amongst subtype D viruses than amongst viruses of other subtypes (10–19%). In late-stage disease, both subtype C and G displayed CXCR4 tropism with low frequency (13% and 12%, respectively). By comparison, all other subtypes displayed a much higher rate of CXCR4 tropism (54–76%) in late-stage disease. In conclusion, these results demonstrate that evolution of HIV-1 coreceptor tropism over time differs between different subtypes and CRFs.

Figure 3.

Prevalence of HIV-1 CXCR4 tropism in the asymptomatic phase and in late-stage disease. Results of the literature analysis show the prevalence of HIV-1 CXCR4 tropism amongst individuals in either the asymptomatic phase or late-stage infection. The number of individuals with known HIV-1 coreceptor tropism that is used in each diagram is specified within brackets.

What determines HIV-1 coreceptor switch?

Despite intensive research efforts, the exact cellular and molecular mechanisms responsible for the coreceptor switch remain unclear (reviewed in [42]). Three main hypotheses have been considered, based on (i) transmission and mutation, (ii) immunity or (iii) the target cell. First, the transmission–mutation hypothesis proposes that CCR5 tropism is fundamental for transmission, and that the emergence of CXCR4 tropism is a result of a gradual mutation of the transmitted CCR5-tropic founder strains. Second, the immune-based hypothesis emphasizes that X4 viruses are more sensitive to a strong and well-functioning immune system than R5 viruses and, as a consequence, are better recognized and suppressed early in infection. In late-stage disease, when the host immune system is severely dysfunctional, X4 viruses are allowed to emerge. Third, the target cell-based hypothesis states that the viral phenotype is affected by the composition of the target cell pool. This proposal was based on the differential expression of CCR5 and CXCR4 on naïve and memory CD4+ T cells and differences in the population dynamics of these cells during the disease course [43]. The increasing proliferation rate of naïve CD4+ T cells (expressing high levels of CXCR4) would result in selection in favour of X4 viruses [44]. Each of these hypotheses presents a different view of the selection pressures acting on the viral population and the interaction of the virus with the immune system of the host, and it is both possible and likely that multiple mechanisms co-exist. The results of a recent study lend further support to the complexity of the coreceptor switch by suggesting that recombination between co-existing R5 and X4 populations may be involved [45]. The recombinants consisted of an R5-derived env V1–V2 region (which has been suggested to be a regulator of resistance against neutralizing antibodies) and an X4-derived env V3 loop (the major coreceptor determinant), and it has been proposed that a coreceptor switch could be facilitated by the recombination between an immune-resistant R5 virus and an immune-sensitive X4 virus [46].

In view of the fact that subtype C is the most prevalent HIV-1 subtype of the global epidemic (>50%), it is important to understand the differences in viral phenotypes described above. The possibility that the epidemiological dominance of subtype C might be related to a higher replicative capacity compared to other subtypes has been considered [47, 48]. However, recent in vitro data suggest that subtype C has lower replicative capacity than anticipated. Using a competition assay of primary HIV-1 subtype B and C isolates in PBMCs, Ball et al. [49] showed that all subtype C isolates were outcompeted by the subtype B isolates. Moreover, Tebit et al. [50] identified that subtype C is unique compared to all other studied subtypes (A, B and D) in terms of lower replicative fitness based on more than 2000 competition assays in PBMCs. If subtype C displays a lower replicative fitness, this may in turn be reflected by a generally lower evolutionary rate (possibly due to longer generation times) and a lower likelihood of accumulating mutations needed for the transition from CCR5 to CXCR4 use. Coetzer et al. [51] recently demonstrated that the transition from CCR5 to CXCR4 use in subtype C requires more mutations compared to other subtypes. Taken together, lower replicative fitness resulting in a lower evolutionary rate in combination with the requirement of more mutations for switch to CXCR4 use of subtype C may explain the low frequency of CXCR4 use in late-stage disease. It is tempting to speculate that the coreceptor-binding site of the subtype C envelope gp120 is structurally different from the envelope of other subtypes.

From a population dynamics perspective, it is reasonable to suppose that recombination or other evolutionary bottlenecks would result in fitness advantages. If not, the effect of such events would not be as distinct at the population level. In 2006, results from replicative competition assays between the CRF02_AG and the putative parental strains, subtype A and G, were presented by two independent studies [52, 53]. Both studies showed that the CRF02_AG was more replicative than subtype A or G. This indicates that recombination between two subtypes produced a more replicative virus. Higher replicative fitness has been shown to inversely correlate with CD4+ T cell counts [54], thereby linking replicative capacity of the virus to severity of disease.

Increased emergence of CXCR4 use: sign of an evolving epidemic?

Recently, evolving patterns of different viral traits at the population level have been presented [55, 56]. Connell et al. [41] analysed the results of 19 subtype C isolates collected during late-stage disease in 2005, and found a higher prevalence (30%) of CXCR4 tropism than in previous studies of this subtype. They suggested that HIV-1 subtype C might be an evolving epidemic in terms of coreceptor tropism, with an increase over time of prevalence of CXCR4 use in South Africa. To investigate whether a similar evolving pattern during late-stage disease was evident also for CRF02_AG in Guinea-Bissau, West Africa, we compared the coreceptor tropism of samples collected during two different time periods (1997–2001 and 2003–2007) [57]. In the earlier time period five of eight viruses used CXCR4, which was significantly different from the later time period in which all 14 viruses studied were able to use CXCR4. Our study as well as that of Connell et al. was based on a small number of samples; therefore we re-analysed both of these datasets together with data from Genbank and confirmed the pattern of an evolving epidemic for subtypes C and CRF02_AG. Subtype C samples collected during late-stage disease before 2000 had a CXCR4 prevalence of 8% (11/145), compared with 33% (21/63) for samples collected after 2000 (< 0.001). Likewise, CRF02_AG samples collected during the period 1997–2001 had a rate of CXCR4 use of 56% (28/50), compared with 93% (14/15) for samples collected between 2003 and 2007 (= 0.012, two-tailed Fisher’s exact test). In competition experiments, CXCR4-using HIV-1 replicated faster than HIV-1 with the R5 phenotype [58]. This higher replicative capacity of CXCR4-using viruses together with a rate in sequence change of 1% per year in the HIV-1 env, reflecting the constant genetic evolution of HIV-1 [59], could in part explain such an evolving epidemic at a population level. Further studies are, however, needed to explain why HIV-1 might be evolving towards an increased likelihood of changing into the CXCR4 phenotype, and what the consequences might be in terms of viral transmission, pathogenesis and disease progression.

Intra-patient evolution of the CCR5-restricted HIV-1 phenotype

As discussed above, coreceptor switch from CCR5 to CXCR4 use represents one pathway of evolution of the HIV-1 biological phenotype during disease progression. However, many patients develop AIDS without such a switch and maintain viruses displaying strict CCR5 dependence [60, 61]. HIV-1 subtype C infections dominate globally (>50%), but only 15% of individuals carrying HIV-1 subtype C develop CXCR4-using virus variants during progression to AIDS [33, 36, 37, 57]. Collective data, mostly from studies of HIV-1 subtype B infections, suggest that HIV-1 infection may result in two distinct phenotypic pathways, either coreceptor switch or adaptive changes of the R5 virus population, including both phenotypic and molecular alterations of envelope glycoprotein (Env) structures (summarized in Fig. 4), in parallel with disease progression [61–73].

Figure 4.

Evolution of the HIV-1 phenotype. Two distinct pathways are shown.

Studies of HIV-1 samples obtained sequentially during disease progression showed that R5 virus variants emerging after onset of AIDS display augmented in vitro fitness. Such viruses had elevated levels of infectivity in cultures of cell lines and primary T lymphocytes, as well as increased fusogenicity and dominance in competition assays [62, 63, 69, 71]. In parallel, it has been reported that late R5 viruses may display increased macrophage tropism, as a result of their ability to exploit low levels of CCR5 for host cell entry (reviewed in [74]). Likewise, late viruses or derived Envs showed increased cytopathic effect and apoptosis, along with reduced entry inhibitor sensitivity [67, 73]. Hence, these viruses could contribute to the depletion of CD4+ T cells. In an attempt to explain the immunodeficiency caused by R5 viruses in patients who do not switch to CXCR4 use (nonswitch virus patients), HIV-1 pathogenic processes were studied in ex vivo lymphoid tissue [75]. The results showed that R5 viruses isolated from nonswitch virus patients depleted more target cells (CD4+ CCR5+) than R5 viruses from switch virus patients. The high depletion of CCR5+ cells by HIV-1 isolates from nonswitch virus patients may explain the steady decline in CD4+ T cells in patients with continuous dominance of R5 HIV-1. Whether the observed differences in depletion of target cells apply to a fraction of patients or delineate a more general rule remains to be clarified.

To further characterize the evolution of the R5 phenotype we used cell lines with chimeric CCR5/CXCR4 receptors, in which successively larger portions of CCR5 have been replaced by the corresponding regions of CXCR4 (Fig. 5) [76]. The chimeric receptors were stably expressed in U87.CD4 cells. In particular, three chimeras, involving exchange of the N-terminal and first extracellular loop regions, were instrumental in demonstrating evolution of CCR5 use over time (Fig. 5). Early isolates exclusively used wild-type CCR5 (phenotype R5narrow), whereas use of an increasingly higher number of chimeric receptors (phenotype R5broad1–3) could be demonstrated in vitro for later viruses, indicating intra-patient evolution over time to a more flexible use of CCR5 [65, 77]. This could be explained by the fact that the virus becomes more resistant to inhibition by the natural CCR5 ligands owing to differential binding to CCR5 [78]. Similar to the CCR5 ligand CCL5, HIV-1 with the R5narrow phenotype (early viruses from patients with high CD4 counts) binds exclusively to the N-terminal of CCR5. Consequently, these viruses are sensitive to inhibition by CCL5, because they compete for binding to the same structure on CCR5. Evolution will, however, lead to viruses gradually more resistant to CCL5 inhibition [61, 64, 65]. Such viruses are no longer dependent on binding to the N-terminal alone but can also use the extracellular loops of CCR5 for cell entry [65, 77, 79–81].

Figure 5.

Tools for testing evolution of the R5 phenotype. (a) Chimeras between CCR5 and CXCR4 are particularly useful for testing intra-patient evolution of CCR5 use. (b) Classification of R5 phenotype according to chimeric receptor use.

In the case of mother-to-child transmission of HIV-1, the maternal R5 phenotype, whether R5narrow or R5broad, is predictive of the virus phenotype of the newborn [80]. The presence of HIV-1 with R5broad phenotype at birth is in turn predictive of early immunological failure in the child [79, 80]. Accordingly, children with HIV-1 R5broad phenotype suffer earlier and with more severe disease than children with the R5narrow phenotype virus. Also in the latter case, the R5narrow phenotype will be subjected to evolution towards a more flexible use of the CCR5 receptor paralleled by increasing resistance to inhibition by CCL5, similarly to the evolution observed in adults during disease progression [79].

The idea that the mode of binding to the 7-transmembrane receptor (in the case of HIV-1 to CCR5 and CXCR4) might result in differences in G-protein-coupled signalling and thereby trigger different physiological cellular processes has been put forward recently [24]. If this hypothesis is correct, it could explain for example how variations in binding of different R5 phenotypes to the CCR5 receptor may modulate cell survival. Indeed, when the distribution of 173 R5 isolates of different phenotypic categories were examined, there was a significant difference between the groups (= 0.01). The majority of viruses in the nonswitch virus group had the R5narrow or R5broad1 phenotype and, conversely, most of the viruses in the switch group had the R5broad2 or R5broad3 phenotype [65]. Moreover, viruses belonging to the R5broad categories had higher degree of infectivity than viruses with R5narrow phenotype, lending further support to the idea that a link exists between the mode of receptor use, virus infectivity, depletion of target cells and the pathogenic process. Early testing and follow-up of the mode of CCR5 use may enable the severity of HIV-1 infection to be predicted in individual patients.

Modifications of Env structures as the molecular rationale for in vivo R5 phenotypic evolution

The molecular rationale for R5 phenotypic evolution has been linked to modifications of Env structures. One such modification emerging after the development of AIDS was reduced gp120 glycosylation observed in R5 virus variants [62, 82, 83]. In general, reduced glycosylation results in particles that are less shielded from neutralizing antibodies. Indeed, such particles were shown to appear simultaneously with the drop in neutralizing antibody responses late in disease. Conversely, high-titre neutralizing activity in sera obtained early in the infection correlated with accumulation of potential N-linked glycosylation sites (PNGSs) [83]. Another interesting finding was the association between loss of specific Env PNGSs and reduced ability of end-stage disease R5 viruses to utilize DC-SIGN (Dendritic cell-specific intercellular adhesion molecule-3-grabbing nonintegrin) binding for trans-infection of CD4+ T cells [62]. Although the efficiency of DC-SIGN use was sustained from primary infection throughout the chronic phase, and believed to facilitate virus spread (reviewed in [84]), efficient DC-SIGN use did not seem to be required by R5 virus variants emerging late in the disease course. Instead, after AIDS onset, changes in the viral envelope favoured virus variants with increased fitness for direct target cell infection, suggesting that DC-SIGN is part of an immune evasion mechanism that is no longer necessary late in infection. DC-SIGN may indeed contribute to evasion of the neutralizing effect of antibodies, as HIV-1 already bound by neutralizing antibodies can be efficiently captured by cells expressing DC-SIGN and result in transfer of infectious virus to T cells [85].

Reduced glycosylation is, however, not the only Env modification in the R5 virus population that develops in the immune-deficient host. Increased positive charge in Env gp120 was observed and the mutations leading to an altered amino acid charge were mapped mainly to the variable regions excluding the V3 loop [70, 86]. Increased positive charge of R5 Env correlated with elevated infectivity and increased cell attachment [70]. It is likely that an overall increased positive charge of gp120 may result from the combined effect of an elevated number of positively charged amino acids and reduced numbers of negatively charged sugar residues. Interestingly, a physical model of the initial steps of retrovirus infections suggests that the initial force for virus adsorption is provided by nonspecific electrostatic interactions [87]. An increase in positive charge would reduce electrostatic repulsion between the viral and cell membranes lending late R5 viruses an increased ability to nonspecifically adhere to the cell surface (Fig. 6). Thus, as a consequence of failing immunity, HIV-1 R5 variants display adaptive Env changes that involve glycosylation as well as variations in charge and may contribute to increased pathogenic properties in an opportunistic manner.

Figure 6.

Schematic representation of the reduced electrostatic repulsion of end-stage R5 variants. At the stage of severe immunodeficiency, shielding from host immunity is not necessary and the number of R5 Env glycans is reduced. Instead, in an opportunistic manner, viruses with Env displaying increased positively charged amino acids are selected for. It is conceivable that the combined loss of glycans and the addition of positively charged amino acids in the R5 Env reduce repulsion between the virus envelope and host cell membrane and contribute to the increased ability to attach to cells, altered mode of receptor use, and better fusogenicity of end-stage R5 viruses.

Chimeric CCR5/CXCR4 receptors as a tool to predict HIV-1 sensitivity to entry inhibitors

Shortly after the discovery of HIV-1 coreceptor use, it became clear that (i) interactions between HIV-1 and entry cofactors are conformationally complex [88, 89], (ii) distinct CCR5 domains can mediate coreceptor usage [90] and (iii) chemokine and coreceptor function maps to distinct but overlapping structures on CCR5 [91]. Moreover, in transient transfection experiments, different HIV-1 envelopes appeared to use different domains of CCR5 [92]. We have shown that the mode of CCR5 use by HIV-1 undergoes evolution during the pathogenic process both in adults and children [65, 77, 79–81]. This evolution parallels CD4+ T cell decline, such that patients with the lowest CD4 counts yield viruses with the most flexible CCR5 use (R5broad3). This would imply intra-patient evolution of sensitivity to entry inhibitors. Indeed, in patients naïve to entry inhibitor treatment, the R5 variants that emerged after AIDS onset displayed greatly reduced sensitivity to inhibition by both natural CCR5 ligands, such as CCL5, and the small-molecule CCR5 antagonist TAK-779 [61, 64, 69]. CCL5 and TAK-779 resistance was paralleled by an altered mode of CCR5 use, involving more flexibility [65, 81]. In this case, it appears that reduced sensitivity to CCR5-ligand inhibition before initiation of anti-retroviral treatment is associated with R5 viruses that may enter host cells independent of the CCR5 N-terminus.

With regard to using entry inhibitors in the clinic as anti-retroviral treatment, and the possibility of developing new small molecules with different CCR5-binding properties [93], it has become increasingly important to be able to predict whether a patient’s virus is sensitive or resistant to the intended entry inhibitor. The chimeric CCR5/CXCR4 receptors can be useful tools for matching the virus with the inhibitor and thereby increase effectiveness of treatment.

The way to individualized drug choice

The emergence of multidrug-resistant virus variants in those treated with the current standard anti-retroviral therapy has spurred the development of new treatment strategies based on inhibition of virus entry [94]. Using site-directed mutagenesis and CCR5 homology modelling, a large number of small-molecule CCR5 antagonists that show potent activities in blocking chemokine function and HIV-1 entry have been developed [93]. Five antagonists, including maraviroc, the only currently licensed CCR5 antagonist, were used to establish the nature of the binding pocket in CCR5. Although the five antagonists are very different in structure, shape and electrostatic potential, they were able to fit in the same binding pocket formed by the transmembrane domains of CCR5 [93]. Of importance, each antagonist displayed a unique interaction profile with the amino acids lining the pocket. It is possible that the unique binding profiles may reflect unique alterations in the conformation of CCR5 after antagonist binding and forecast variation in inhibition sensitivity of different R5 virus variants. Accordingly, this would support the need for individualized drug choice for each patient in HIV-1 therapy. Consequently, knowledge of baseline characteristics of the binding of HIV-1 variants naturally emerging in the infected host is of utmost importance.

In the clinic, maraviroc is used as a complement to other anti-retroviral agents in patients with late-stage disease who have failed other regimens [95]. Clinical studies have also shown that the emergence of X4 viruses predominates in individuals experiencing virological failure during maraviroc treatment [95–97]. It is likely that maraviroc treatment in late-stage disease selects for CXCR4-using viruses that are present in the patient, whereas early treatment in the asymptomatic phase rarely yields resistant viruses. Indeed, maraviroc has been shown to efficiently block R5 viruses derived from the asymptomatic phase (reviewed in [95]). This also suggests that CCR5 antagonist treatment may be more effective in subtype C and G than subtype B infections, as CXCR4-using viruses appear to emerge less frequently in individuals infected with C and G subtypes of HIV-1 (Fig. 3).

Nevertheless, in vitro R5 virus passage experiments have revealed that maraviroc-resistant variants can be selected for. In this case, resistance was not because of CXCR4 use, but involved changes in the virus CCR5-binding properties [98, 99]. In the case of selection with maraviroc, vicriviroc and AD101, this change has been identified as high dependence on the CCR5 N-terminus for binding [100–102].

In summary, small-molecule entry inhibitors are a promising way of suppressing HIV-1 replication in the infected host. It is, however, important to keep in mind that the baseline mode of CCR5 use and resistance to CCR5 ligands may complicate the efficacy of this treatment, particularly in late-stage HIV-1 disease. We have the means today – including the use of chimeric CCR5/CXCR4 receptors – to assess the mode of CCR5 use as well as both baseline sensitivity and changes in the sensitivity of the virus during treatment. We believe that enhanced knowledge of the evolution of coreceptor use during progressive HIV-1 infection has opened up new avenues leading towards individualized and optimized anti-retroviral treatment.

Conflict of interest statement

No conflicts of interest to declared.


We would like to thank all our colleagues who have participated in discussions and contributed material, results and ideas during the past 25 years. Eva Maria Fenyö would especially like to thank the authors of the first publications on HIV-1 biological phenotype, in the 1980s and early 1990s: Birgitta Åsjö, Linda Morfeldt, Jan Albert, Gunnel Biberfeld, Anders Karlsson, the late Knut Lidman, Francesca Chiodi, Birgitta Lind, Agneta von Gegerfelt, Eva Olausson, Katarina Parsmyr, Kajsa Aperia and Eric Sandström. Additionally, we would like to thank Åsa Björndal, Gabriella Scarlatti, Robert Fredriksson, Charlotte Tscherning, Eva Backström, Hans Wigzell and Mikulas Popovic for scientific input during the time immediately after the discovery of HIV-1 coreceptors. In the past 10 years, Ingrid Karlsson, Anna Laurén, Mattias Mild, Liselotte Antonsson, Johanna Repits, Marie Borggren, Ulf Karlsson, Monica Öberg and Elzbieta Vincic have contributed to our knowledge of HIV-1 coreceptor use. Last but not least we would like to thank colleagues with whom collaboration was essential for reaching the results described here, including Dan Littman at New York University, Christer Owman at Lund University, Leonid Margolis at NIH, Bethesda and Paul Gorry at the Burnet Institute, Melbourne. Grants were received from the Swedish Research Council, the Swedish International Development Cooperation Agency/Department for Research Cooperation (SIDA/SAREC) and the European Commission.