The attachment inhibitor BMS-378806 inhibits both R5- and X4-tropic HIV-1 isolates [87]. This compound binds to a pocket on gp120 important for binding CD4 and alters the conformation of the protein such that it cannot recognize CD4 [88]. The inhibitor binds with significantly higher affinity than CD4, and is an excellent candidate for therapeutic use. TNX-355 is a humanized anti-CD4 monoclonal antibody that binds to CD4 and inhibits HIV-1 envelope docking, but does not inhibit CD4 function in immunological contexts. Currently no attachment inhibitors are approved for use in patients with HIV-1 infection and none appear to be approaching phase I clinical trials.

The crystal structure of the gp41 ectodomain [29] and of the ectodomain partnered with an inhibitory peptide (C34) [89] revealed that the fusion active conformation of gp41 was a six-helix bundle in which three N-helices form an interior, trimeric coiled-coil onto which three antiparallel C-helices pack. Enveloped viruses use a generally conserved mechanism to initiate the fusion event between viral and host cell membranes, and this six-helix bundle formation has been well studied in the paramyxoviruses family of viruses [10,90]. Peptide fusion inhibitors were designed based on the discovery that two homologous domains in the viral gp41 protein must interact with each other to promote fusion, and that mimicry of one of these domains by a heterologous protein can bind and disrupt the intramolecular interactions of the virus protein. Alpha-helical peptides homologous to the leucine zipper domain of gp41 had significant antiviral activity against HIV-1, and this activity depended upon their ordered solution structure [91]. Rational design ultimately produced a molecule (T-20, enfuvirtide) with potent antiviral activity in vivo [92,93], and many other peptide mimics have been described (Table 2).

Resistance to early alpha-helical inhibitors was mediated by mutations in the N-terminal heptad repeat region of gp41 [94], providing evidence of the specificity of binding of these peptides to the virus. Monotherapy with enfuvirtide resulted in viral load rebounds after 14 days of therapy, and resistance determinants mapped to the HR1 domain (G36D, I37T, V38A, V38M, N42T, N42D, N43K) [95]. Mutations that confer resistance to enfuvirtide also result in reduced replication capacity / replicative fitness of the virus [96]. This is presumably due to adaptations in the gp41 domain that reduce enfuvirtide binding, but consequently reduce the efficiency of six-helix bundle formation and overall fusion rates [97]. These mutations did not confer cross resistance to other types of entry inhibitors (attachment inhibitors or coreceptor inhibitors) [98] but did sensitize viruses to neutralization by monoclonal antibodies that target the gp41 domain, presumably by prolonging the exposure of fusion intermediates that are specifically sensitive to these antibodies [97]. Adaptation to enfuvirtide has even resulted in viruses that require enfuvirtide for fusion [99].

Though it is clear that gp41 resistance mutations, taken out of the context of the envelope under selection, result in decreased fusion efficiency and reduced viral fitness, the function of envelope context seems to modulate the overall effect of these mutations in vivo [100]. Studies of baseline susceptibility to enfuvirtide suggested that large variations in intrinsic susceptibility existed in diverse HIV-1 isolates, and that these variations mapped to regions outside the enfuvirtide binding site [101]. Sequences associated with the V3 loop were correlated with intrinsic enfuvirtide susceptibility, suggesting that interactions with the coreceptor were important determinants of susceptibility of a drug that inhibits virus fusion. A seminal observation in the understanding of entry inhibitor susceptibility was made in 2002 by the discovery that the efficiency of the fusion process was the principal modulator of intrinsic enfuvirtide susceptibility [102]. Mutations in the coreceptor binding site that reduced gp120 affinity for CCR5 resulted in viruses with reduced fusion kinetics [103,104]. Engagement of CD4 by gp120 initiates a process of structural rearrangement in the envelope glycoprotein resulting in fusion. Completion of this process requires engagement of the coreceptor molecule, but enfuvirtide susceptibility is limited to the time between CD4 engagement and six-helix bundle formation. Thus efficiency of this process affects the amount of time that enfuvirtide has to bind gp41 and prevent fusion. Any change in the system that decreases rate of fusion (e.g., reducing the levels of coreceptor expression) also increases susceptibility of virus to inhibition by enfuvirtide. Consistent with this, ENF is synergistic with compounds that inhibit CD4 or coreceptor engagement [105,106,107].

Discovery of the chemokine coreceptors was facilitated by the observation that native chemokines can inhibit HIV-1 replication [44]. Thus the first attempt to generate a coreceptor based entry inhibitor was modulation of the endogenous chemokine RANTES (CCL5) to produce a more potent version. Modulation of the N-terminus by deletion of the N-terminal methionine residue resulted in increased potency of the chemokine RANTES [108]. Addition of an aminooxypentane moiety to the N-terminus of either RANTES or MIP-1αP resulted in further increases in potency and increases in affinity for CCR5 [108,109]. Modulation of the N-terminus of the chemokine was specifically chosen because it appears that chemokines interact selectively with specific receptors by a “message” and “address” system. The core domain of the chemokine is specifically responsible for determining receptor specificity (the address), while the N-terminus of the chemokine is involved in elicitation of downstream signal transduction cascades (the message) [110]. An HIV-1 inhibitory molecule should ideally not induce signal transduction cascades or the ensuing lymphocyte activation or chemotaxis. Modifications made to generate met-RANTES and AOP-RANTES resulted in abrogation of chemotaxis signals but not calcium flux [108,111]. This signaling cascade induced by RANTES derivatives can lead to an activation of HIV-1 replication, which can counter the effects of this analog on entry inhibition [108].

One major principal for the development of chemokine-like inhibitors is the ability of chemokines to induce ligand-mediated internalization of their receptor through a clathrin-dependant mechanism. Chemokine receptors, when bound by ligand and in their signaling-active state, activate G protein-coupled receptor kinases (GRKs). GRKs phosphorylate the C-terminal cytoplasmic domain of the chemokine receptor. The phosphorylated form of the receptor is recognized by β-arrestin, which recruits scaffolding proteins and clathrin precursor proteins to the receptor. Clathrin-coated pits form invaginations and the receptor is internalized. CCR5 is specifically endocytosed and moved to a recycling endosomal compartment [112]. Contribution of receptor downregulation in inhibition of HIV-1 entry by chemokines was first observed for inhibition of X4-tropic HIV-1 isolates by SDF-1 [113]. Analysis of a series of RANTES analogs revealed that their potency in blocking HIV-1 entry correlated not with their binding affinity, but with their ability to induce ligand-mediated internalization of CCR5 [114]. Consistent with this, mutagenesis of the CCR5 C-terminus, abrogating its recognition and phosphorylation by GRKs, resulted in a mutant that is not internalized when bound by ligand. Use of this mutant significantly reduced the potency of AOP-MIP-1αP during the course of short term virus exposure (<12hrs) [115]. Beyond simple receptor internalization, it has been found that chemokine analogs result in prolonged sequestration of cell surface receptor [116]. The mechanisms behind prolonged sequestration of CCR5 have been explored but have not been fully elucidated. Further iterative design of the RANTES N-terminus resulted in a third generation compound, PSC-RANTES, with even higher potency than AOP-RANTES [117]. PSC-RANTES has been shown to block vaginal transmission in the SHIV-Macaque model system [118] and is currently in development as a microbicide to prevent sexual transmission of HIV-1.

In some of these macaques treated with high PSC-RANTES concentrations (100μM), a selection of PSC-RANTES resistance variants was confirmed and as described below, resistance was likely conferred by increased coreceptor affinity. In relation to the mechanism of PSC-RANTES inhibition, altered PSC-RANTES or AOP-RANTES sensitivity in primary HIV-1 isolates that remain CCR5-using is likely in the context of either differential binding of virus versus drug on the CCR5 receptor, increased affinity of the virus for the receptor, or more rapid entry kinetics. However, CCR5-dependent resistance as opposed to coreceptor switching indicates that the mechanism of PSC-RANTES derivative inhibition cannot be solely due to receptor-ligand internalization or sequestration from the cell surface. Studies suggest that competitive inhibition may even be a dominant mechanism by which PSC-RANTES or other RANTES derivatives block HIV-1 entry [119].

A major problem associated with chemokine analogs is their residual capacity to signal through CCR5 and through cell surface proteoglycans [111,120]. This residual activity has been associated with stimulation of viral replication by enhancement of viral integration [121] and through enhancement of virus entry [122]. AOP-RANTES, like RANTES, has been shown to activate mitogen-activated protein kinases and pertussis-toxin sensitive signals [121,123]. High concentrations of chemokine analogs clearly have a stimulatory effect on T cells, and are formally considered partial agonists of CCR5. This residual signaling activity may limit the effectiveness and potential use of chemokine analogs in vivo.

Small molecule CCR5 antagonists bind to hydrophobic pockets within the transmembrane helices of CCR5 [124,125]. This site does not overlap the binding sites of either CCR5 agonists or HIV-1 envelope, but instead induces and stabilizes a receptor conformation that is not recognized by either. Thus, these molecules are considered allosteric inhibitors. Ideally, a small molecule inhibitor of CCR5 would block binding by HIV-1 envelope but continue to bind native chemokines and facilitate signal transduction. Most small molecule inhibitors, however, are pure antagonists of the receptor. Oral administration of small molecule antagonists has been demonstrated to inhibit viral replication in macaque models [126] and to prevent vaginal transmission [127]. Thus far, three antagonists (VCV, MVC, and Aplaviroc) have been shown to inhibit virus replication in humans [128]. The compound maraviroc (MVC) has been tested in phase III efficacy trials and was approved for therapeutic use by the FDA in 2007.

Maraviroc is a selective small molecule CCR5 antagonist currently being used to treat patients with resistance to multiple HIV drugs but has been recently approved for first-line treatment regimens. Identified in a chemokine binding assay screen of a Pfizer compound library, maraviroc is an imidazopyridine that binds a hydrophobic transmembrane cavity of CCR5. Alanine scanning mutagenesis of the transmembrane domains of CCR5 revealed key residues in domains 1, 2, 3 and 7 for small molecule binding [124]. Binding alters the conformation of the second extracellular loop of the receptor, preventing its interaction with the V3 stem loop of HIV’s gp120 envelope glycoprotein [124,129]. Ligand binding competition assays indicated maraviroc prevented C-C chemokine ligand binding to CCR5 with IC₅₀ values (concentration of drug required to inhibit viral replication by 50% in culture) in the nanomolar range. Additionally, it inhibited chemokine-induced signaling of intracellular calcium redistribution, however it did not trigger calcium signaling upon binding up to 10μM [128]. In the same study, maraviroc was shown to have potent antiviral activity against a panel of primary CCR5-tropic HIV-1 isolates with a mean IC₉₀ of 2 nM.

The MOTIVATE 1 (conducted in North America) and MOTIVATE 2 (conducted in Europe, Australia, and the US) phase III clinical trials sought to study the safety and efficacy of maraviroc in multi-drug resistant patients harboring R5 tropic viruses. (MOTIVATE stands for Maraviroc versus Optimized Therapy In Viremic Antiretroviral Treatment Experienced patients) Results from these trials published in 2008 indicated patients receiving maraviroc (either once or twice daily) plus an optimized background therapy (OBT) showed significant reductions in HIV-1 RNA levels and increases in CD4 cell counts versus patients receiving OBT alone [130,131]. Preliminary results from these trials resulted in FDA approval of maraviroc for clinical use in 2007. It is currently marketed as SELZENTRY by Pfizer, Inc. Since its approval, maraviroc has been utilized as a salvage therapy for multi-drug resistant patients with R5 tropic virus.

Vicriviroc is another CCR5 inhibitor by Schering Plough currently in phase III clinical trials that binds a similar transmembrane pocket as maraviroc. Preliminary results from the vicriviroc clinical trials indicate a reduction in viral load in patients taking vicriviroc in combination therapy [132]. In addition to maraviroc and vicriviroc, several other small molecule inhibitors are in various stages of development (Table 2). Aplaviroc was developed as a small molecule CCR5 antagonist to block HIV-1 entry but GlaxoSmithKline halted development in a phase II trial due to strong evidence of hepatic toxicity [133].

The major concern in the therapeutic administration of coreceptor inhibitors is the possibility that resistance will manifest by a change in coreceptor tropism from CCR5 to CXCR4, or that an outgrowth of an X4-tropic virus subset will come to dominate the intrapatient virus population. The mechanisms and consequences of coreceptor switching are poorly understood at this time. It is unclear what factors drive this switch, or why switching is generally limited to late phases of disease progression. Indeed, it is not clear at this time whether a switch from R5 to X4 tropism is a cause or consequence of disease progression.

On the other hand, resistance to a variety of small molecule CCR5 inhibitors has been generated by passage of inhibitor-susceptible virus isolates in sequential dose escalations of drug. These propagations have been performed in PBMC cultures, which express CCR5 and CXCR4, as well as a variety of other chemokine receptors that have been previously implicated as potential HIV-1 coreceptors. In these experiments, inhibitor-resistant viruses continue to require CCR5 for entry and cannot utilize either CXCR4 or any alternative coreceptors [134,135,136,137] (Figure 3). Furthermore, evaluation of coreceptor tropism of viruses derived from patients who failed maraviroc therapy during the MOTIVATE trials indicate that tropism changes occurred only when X4 tropic viruses were pre-existing in the patient quasispecies [138]. Of 1042 patients with R5 tropic virus at the time of enrollment, 228 patients reportedly failed treatment. Of these failures, 133 were receiving maraviroc and of these 76 patients were found to harbor dual or mixed tropic virus at the time of failure [131]. Tropism testing revealed that these patients had pre-existing populations of X4-tropic virus prior to the start of the trial, which was below the level of detection of the tropism assay used at the time of enrollment. Another trial specifically studying effects of maraviroc on patients with non-R5 tropic virus found no benefit in maraviroc treatment versus placebo [139]. Thus it appears that, though outgrowth of pre-existing X4-tropic viruses in the quasispecies remains a problem for the therapeutic administration of CCR5 antagonists, de novo mutations conferring altered coreceptor usage may not be the favored pathway for resistance in vitro or in vivo.

It is important to note that a coreceptor switch or evidence of pre-existing CXCR4 tropic virus only provide a reason for 76 of 133 MVC failures. As described below, four of the MVC treatment failures may relate to use of CCR5 even in the presence of MVC. However, 55 MVC treatment failures remain unexplained at the time of this publication. Recent studies suggest a possible discrepancy between differential sensitivity to CCR5 antagonists in various phenotypic assays. For example, susceptibility to entry inhibitors, specifically CCR5 antagonists, can be modulated dependent on cell type, state of cellular activation, and number of virus replication cycles [119,134,136,140]. Also, different primary HIV-1 isolates can vary in sensitivity to entry inhibitors as much as 100-fold in IC₅₀ values [119,128]. This difference is much more demonstrable with infection assays using replication competent primary HIV-1 isolates as compared with defective viruses limited to single cycle replication. In 2005, Dorr et

al. reported less than a 10-fold variation in MVC sensitivity with envelope pseudotyped viruses. In contrast, primary isolates in multiple cycle assays displayed a 100-fold variation in MVC sensitivity but only a 10-fold inhibition difference in single cycle assays [128]. Thus, it is quite possible that the remaining 55 failures in the MOTIVATE trials might relate to sensitivity of the infecting virus to MVC prior to treatment or simply, the inability to detect MVC resistance at the time of treatment failure with existing assays.

If coreceptor switching is not the main mechanism of resistance to small molecule coreceptor inhibitors, then it is important to understand by what mechanism resistance does arise. Two models of resistance have been proposed: competitive and noncompetitive resistance (Figure 4). In the competitive resistance model, gp120 interacts with CCR5 with a given affinity. In the presence of an inhibitor (green), the high affinity interaction between receptor and inhibitor blocks the lower affinity interaction between the receptor and gp120 (Figure 4c). Adaptations in gp120 increase the affinity relationship between envelope and coreceptor, and this change in receptor affinity is sufficient for gp120 to compete inhibitor off of the receptor (Figure 4f). Experimental evaluation of this resistance mechanism would appear as a shift in IC₅₀ value and complete inhibition of virus entry can be achieved by increasing the concentration of inhibitor (Figure 4g). Many questions about this potential form of resistance can be raised: 1) what are the baseline affinities of each contact point between gp120 and coreceptor? 2) How great are the variations among HIV-1 isolated in these baseline affinities? 3) What types of change in the gp160 sequence would result in greater coreceptor affinity? 4) How great a change in affinity is required to result in differential inhibition by a given drug? 5) Will gp120 adapt to different CCR5 inhibitors by a similar pathway, or by alternative mutations for each drug (i.e., will cross resistance be a problem)? 6) What are the consequences of affinity changes on overall viral replication? 7) What are the consequences of affinity changes on viral dynamics in vivo and disease progression? Much work is still required to substantially address these issues.

The second model of resistance to coreceptor inhibitors is noncompetitive resistance (Figure 4). In this scenario, a small molecule inhibitor binds CCR5 in an allosteric site, which induces and/or stabilizes a receptor configuration that is not recognized by HIV-1 envelope (or native chemokines). Due to the lack of recognition, receptor docking is blocked and the HIV-1 entry process cannot proceed (Figure 4b). Adaptation to this type of inhibition is mediated by changes in gp120 that allow for recognition of the CCR5 configuration induced by inhibitor binding in the allosteric site. The overall affinity for the inhibitor-free receptor may not be relevant in this circumstance, and the envelope can complex with the inhibitor-bound configuration of CCR5 (Figure 4e). Experimental evaluation of noncompetitive resistance is manifested by a plateau effect on HIV-1 entry as inhibitor concentration increases (Figure 4g). In this case, the overall extent of inhibition never reaches 100% independent of the concentration of inhibitor. Evaluation of inhibition curves would reveal that the EC₅₀ for an inhibitor resistant virus (defined as the half-plateau height, or the concentration at which the inhibitor achieves 50% of its maximal effect) is equal to the IC₅₀ of the inhibitor against the inhibitor-sensitive variant of the virus. The plateau effect is caused by differences in the efficiency of usage of the inhibitor-free and inhibitor-bound form of the receptor. Thus a completely sensitive virus uses the inhibitor-bound form of CCR5 with 0% efficiency. A virus that uses the inhibitor-bound form of CCR5 with equal efficiency as the inhibitor-free form of CCR5 would not indicate any level of inhibition. Consequently, any virus adaptation that would cause an increase in the efficiency of use of the inhibitor-bound receptor over the efficiency of use of the inhibitor-free receptor would yield a negative inhibitory value, or an apparent enhancement of infection in the presence of saturating levels of inhibitor. This has led to redefining phenotypic resistance as a reduction in the maximal percent inhibition (MPI) <95% rather than a classical shift in IC₅₀ values. The pattern of mutations in env conferring this drug resistant phenotype, i.e., entry through drug-bound receptor, is largely unknown but could be related to V3 loop mutations (Table 3). In vivo, this resistance mechanism has been observed in several studies. For example, an HIV-1 subtype C infected patient treated with vicriviroc failed treatment and was shown to harbor virus capable of replicating even at the highest drug concentration (1μM) [141]. With this patient’s HIV-1 envelope, there was no apparent shift in IC₅₀ values to VCV. Resistance to PSC-RANTES was observed in macaque microbicide studies and the SHIV env variant was capable of replicating even at 100 nM of drug, a concentration 100-fold above the IC₅₀ for wild type SHIV env [142]. In contrast to the observation with VCV resistance, this PSC-RANTES resistant virus displayed a shift in drug susceptibility curve suggesting that increased binding affinity for native CCR5 and displacement of PSC-RANTES versus PSC-RANTES bound CCR5 may be more dominant in the drug resistant phenotype. This combination in mechanisms has also been observed in resistant HIV-1 isolates selected under VCV selective pressure in vitro [134].

During the MOTIVATE clinical trials, several patients failed treatment yet maintained CCR5 tropism. While some of these patients experienced treatment failure as a result of opportunistic infections, a subset experienced failure as a result of apparent resistance to maraviroc. Clonal analyses of the env regions of viruses from these patients indicated changes had occurred in the V3 loop of gp120 as compared to viruses isolated at the time of enrollment. However, the pattern of mutations was different in each patient implying multiple pathways to resistance were possible [143].

Currently, there is limited data available from in vitro resistance studies. However, a few studies have described in vitro generated resistant variants of primary HIV-1 isolates. In 2006 maraviroc resistant variants were generated from two different primary HIV-1 isolates, RU570 (clade G) and CC1/85 (clade B), by sequential passage in the presence of maraviroc [136]. Both resistant viruses possessed distinct amino acid changes in the V3 loop of gp120 as compared to control virus passages (Table 3). For CC1/85_res virus A316T and I323V mutations were observed in the V3 loop while the RU570_res virus had a three amino acid (QAI) deletion from residues 315-317. Both resistant viruses had additional mutations in other regions of env outside of V3 but site-directed mutagenesis studies of molecular clones created from CC1/85_res indicated that the V3 loop changes were responsible for conferring maraviroc resistance. The reduction in MPI observed for the resistant viruses indicated these viruses were able to utilize inhibitor-bound receptor for entry. To further support this hypothesis, maraviroc resistant viruses that were susceptible to aplaviroc (a CCR5 inhibitor that binds to a similar region as maraviroc) were unable to be inhibited by aplaviroc in the presence of high concentrations of maraviroc.

Studies describing other resistant primary isolates vary widely in adaptive amino acid changes observed in the envelope proteins. Genotypic and phenotypic analysis of CC1/85 passage variants that ultimately gave rise to resistance to AD101 indicated that resistance arose via stepwise accumulation of mutations in the V3 region of gp120 [140]. A single mutation was selected for in V3 after four to six passages in the presence of AD101 that resulted in modest resistance (three-fold) through an apparent competitive mechanism. This variant had increased capacity to utilize low levels of CCR5 [135] and is hypothesized to have greater overall CCR5 affinity. Upon continued passage, three more mutations arose in the V3 which led to complete insensitivity to AD101. Inhibition studies of virus clones bearing these four mutations resulted in a plateau effect which indicated the virus was using the inhibitor-bound CCR5 complex with approximately 90% efficiency of the inhibitor-free form of CCR5, though cell type differences in plateau height did exist [144]. However, adaptation to vicriviroc of this same primary isolate did not involve changes in the V3 region, suggesting that resistance can arise through alterations in envelope geometry that are not restricted to specific sites within the envelope glycoprotein [134,145]. These studies suggest that viruses from the same primary isolate can follow multiple genetic pathways to resistance.

Additional reports of chemokine receptor directed small molecule resistant viruses have largely implicated mutations in the V3 region as necessary and sufficient for resistance. Mutations in the V3 loop sequence of virus from the subtype C infected individual harboring VCV resistant virus were found sufficient to confer vicriviroc resistance [141]. More specifically the emergence of the S306P mutation was associated with complete resistance to vicriviroc in chimeric clones. Utilizing the clade G primary isolate RU570, a vicriviroc resistant variant was isolated and mutations in the V3 loop and C4 region were identified as responsible for conferring resistance [146]. Resistance to TAK-652 manifested as changes in the V3 loop as well as other regions of gp160 [137]. A single K315R mutation in the V3 loop of SHIV_SF162-p3 conferred at least 10-fold resistance to PSC-RANTES, which was selected under a single dose of drug applied as a microbicide [142]. More importantly, virus harboring the K315R mutation was still capable of replicating in the presence of 100 nM PSC-RANTES or a concentration at least 100-fold above the IC₅₀ value for the wild type virus. Ultimately, the changes required in gp120 to allow for recognition of inhibitor-bound forms of CCR5 are unclear, and the alterations in the geometry of the envelope – receptor complex remain a mystery.

Of note is that several of the studies describing generation of resistant viruses have utilized primary isolate CC1/85 or a derivative clone to generate resistance to maraviroc, vicriviroc, or AD101 [134,140,147]. This parental virus was derived from a subtype B individual in 1985 that had viruses only able to use CCR5. Use of this single primary isolate has resulted in multiple sets of mutations conferring resistance to each of these drugs implying multiple pathways of resistance can be followed by the same virus. Only two studies have described resistance developed in a non-subtype B virus (clade G RU570) against maraviroc and vicriviroc [136,146]. There is a clear need for more studies characterizing small molecule resistance in non-subtype B isolates.

Cross-resistance has been described for some but not all of the in vitro derived resistant variants. Two maraviroc resistant variants remained susceptible to other small molecule CCR5 antagonists and the fusion inhibitor enfurvitide [136]. In contrast, vicriviroc resistant variants acquired resistance to the modified chemokine derivatives PSC-RANTES and AOP-RANTES as well as other small molecule CCR5 inhibitors, namely AD101, SCH-C, and CMPD 167 [134]. An AD101 resistant variant acquired resistance to the small molecule SCH-C yet remained susceptible to inhibition with the chemokine ligand RANTES [140]. Therefore the implications of the development of resistance to one small molecule CCR5 inhibitor to cross-resistance to other members of this class are not yet clear.

In summary, mutations associated with resistance to CCR5 antagonists in in vitro derived resistant isolates have largely mapped to the V3 region of gp120 with the exception of two viruses with resistant mutations in other regions of gp160. However, resistance in clinical trial patients failing treatment with R5 tropic viruses have all been attributed to mutations in the V3 loop arising during therapy [141,143]. While cross-resistance has been reported for some resistant variants, others have remained susceptible to small molecule inhibitors.