CCR5 Mutations Distinguish N-Terminal Modifications of RANTES (CCL5) with Agonist versus Antagonist Activity

Abstract

CCR5 is the major HIV-1 entry coreceptor. RANTES/CCL5 analogs are more potent inhibitors of infection than native chemokines; one class activates and internalizes CCR5, one neither activates nor internalizes, and a third partially internalizes without activation. Here we show that mutations in CCR5 transmembrane domains differentially impact the activity of these three inhibitor classes, suggesting that the transmembrane region of CCR5, a key interaction site for inhibitors, is a sensitive molecular switch, modulating receptor activity.

TEXT

Multiple studies have examined the impact of mutations in extracellular, transmembrane (TM), and intracellular domains of CCR5 on chemokine binding, receptor activation, HIV-1 entry, and small-molecule inhibitor binding (1, 2, 5, 7, 9, 10, 15). The prevailing two-site model of chemokine binding and receptor activation posits docking of the CC chemokine core on the extracellular domain of CCR5 and insertion of the N terminus of the chemokine into a binding pocket formed by multiple TM domains (2). It is known that small-molecule allosteric CCR5 inhibitors bind in or near the postulated TM binding pocket normally occupied by the N-terminal domain of chemokines (7, 14). Although these studies have identified some sites important for receptor activation (e.g., Thr⁸² [11]), no prior report has examined the binding requirements for chemokine analogs with enhanced anti-HIV activity. The recent description of several classes of N-terminal-modified RANTES/CCL5 analogs (6), all with similarly high HIV-1 inhibitory potencies but showing strikingly different pharmacological profiles (differing in agonist activity and induction of intracellular receptor sequestration), provided the opportunity to determine if mutations in CCR5 TM domains differentially impact the inhibitors from different classes (Fig. 1).

Fig 1.

(A) Sequence of CCR5, with sites of mutations indicated. (B) N-terminal modifications of RANTES used in these studies and their properties. Abbreviations and symbols: TM, transmembrane; ICL, intracellular loop; ECL, extracellular loop; ****, N-nonanoyl, des-Ser¹[l-thioproline², l-cyclohexylglycine³]-RANTES(4-68).

Site-directed mutagenesis (QuikChange; Stratagene, La Jolla, CA) was used to generate alanine substitutions of 7 CCR5 TM residues previously implicated in chemokine and small-molecule inhibitor binding, with Gly163 mutated to Glu (G163E), as described in reference 7 (Fig. 1). The mutated CCR5 expressed in pBABE was transfected (Superfect; Qiagen, Valencia, CA) into NP-2.CD4 glioma cells (16), and cells expressing mutant CCR5 levels comparable to wild-type (WT) CCR5 were selected by cell sorting. All mutations gave CCR5 expression levels within 2-fold of WT CCR5, except Y108A (Fig. 2A). Luciferase-reporter HIV-1 constructs were pseudotyped with the CCR5-using BaL envelope (3, 4) and used for single-cycle infection assays of the NP-2.CD4 cell lines expressing WT or mutated CCR5 (Fig. 2B). Virus entry was in the same range as mediated by WT CCR5 for all mutants except Y108A, which had lower levels of entry function. These results are consistent with prior reports (5, 10, 15) that indicated that these mutations preserve coreceptor function.

Fig 2.

(A) Expression levels of WT and mutated CCR5 on NP-2.CD4 glioma cells. Staining intensity after labeling with the PA12 anti-CCR5 antibody recognizing the N-terminal domain was normalized to the change for WT CCR5 staining. Data are means ± standard errors (SE) of three replicate experiments. (B) Entry function of luciferase reporter HIV-1 pseudotyped with BaL envelope on WT and mutated CCR5. Entry data (luciferase activity, in relative light units) were normalized to the change for entry mediated by WT CCR5. Data are means ± SE of three experiments.

The RANTES/CCL5 analogs illustrated in Fig. 1B were used to inhibit HIV-1 BaL-mediated virus entry into the NP-2.CD4 target cell lines (Fig. 3A). The change in inhibitory potency of each mutation is plotted relative to the pIC₅₀ value (the negative log of the 50% inhibitory concentration [IC₅₀], expressed in moles/liter) for WT CCR5 (Fig. 3B). IC₅₀s were derived from curve-fitting programs (Prism 5; GraphPad, San Diego, CA) with robust statistical support in 3 replicate experiments. 5P12-RANTES, a nonsignaling, nonsequestering analog (6), was highly sensitive to the E283A mutation in TM7, which reduced the inhibitory potency by 80-fold (P < 0.001; two-tailed t test versus WT CCR5). In contrast, 6P4-RANTES, a signaling, sequestering analog (6), showed an increase in potency of greater than 100-fold (P < 0.001) on CCR5 with the E283A mutation and was also sensitive to the Y37A mutation in TM1, with a 100-fold reduction in activity (P = 0.0057). PSC-RANTES (13) which, like 6P4-RANTES, is a signaling, sequestering analog, also showed enhanced potency on CCR5 E283A, but its potency was not reduced by the Y37A mutation. Finally, 5P14-RANTES, a nonsignaling analog that achieves significant receptor sequestration (6), was not sensitive to either the E283A mutation or the Y37A mutation, instead showing a significant increase in inhibitory potency on CCR5 with the N252A mutation in TM6 (Fig. 3A).

Fig 3.

(A) Changes in inhibitory potencies of 5P12-, 5P14-, 6P4-, and PSC-RANTES caused by individual transmembrane mutations in CCR5. Data are expressed as the log change in pIC₅₀ values (the inverse log of the IC₅₀ in moles/liter [M], e.g., 9 = −9 M, = 1 nM) compared to unmutated WT CCR5. (B) The pIC₅₀ values of 5P12-RANTES, 5P14-RANTES, 6P4-RANTES, and PSC-RANTES compared to WT CCR5. Data are the means ± SE of three replicate experiments. (C) IC₅₀s (in nM) for half-maximal inhibition of [¹²⁵I]CCL3 binding (means ± SE) to WT CCR5, CCR5 with the Y37A TM1 mutation, or CCR5 with the E283A TM7 mutation.

Although the E283A and Y37A mutations had the most dramatic impact on inhibitor potency, all 8 TM mutations significantly altered inhibition by at least one of the RANTES analogs (Fig. 3A).

The relative potency of each inhibitor on native CCR5 is shown for reference in Fig. 3B. Hence, prototypic analogs from the three different classes, (i) nonsignaling and nonsequestering, (ii) signaling and sequestering, and (iii) nonsignaling and sequestering, were affected differently by certain TM domain mutants investigated in this study. These results highlight key points of interaction between the different classes of anti-HIV chemokines and CCR5. Glu²⁸³, which has been previously shown to be of importance for both the anti-HIV activity of maraviroc (7, 14) and the signaling activity of native RANTES/CCL5 (17), appears be very important. Elimination of the side chain negative charge here (E283A) enhanced the inhibitory effect of the signaling and internalizing molecules, PSC-RANTES and 6P4-RANTES. Interestingly, these analogs both carry a negatively charged residue in the modified N-terminal region (Asp⁶ in PSC-RANTES and Asp⁵ in 6P4-RANTES), as do four of the other highly potent analogs with this property (identified in reference 6), while none of the 15 nonsignaling analogs had a negatively charged residue in this region. In contrast, the E283A mutation strongly reduced the inhibitory potency of the nonsignaling, nonsequestering analog 5P12-RANTES. Interaction of 5P12-RANTES with this site is likely to involve structures located in positions 6 through 9, because these are the only positions at which 5P12-RANTES differs from 5P14-RANTES (Fig. 1B), which was unaffected by the E283A mutation. Similarly, the corresponding region of 5P14-RANTES is likely to be responsible for a key interaction with Asn²⁵², since 5P12-RANTES was not affected by the N252A mutation. The Y37A mutation reduces the potency of 6P4-RANTES but not PSC-RANTES, the other prototypic signaling and internalizing analog. This TM1 site may interact with the part of the N-terminal pharmacophore region located beyond the first four residues, since these are shared by 5P12-RANTES and 5P14-RANTES, whose activities were not affected by the Y37A mutation.

Prior studies (6, 12) have shown that the antiviral potencies of the RANTES analogs are not correlated with their binding affinities for native CCR5. To investigate whether or not CCR5 TM mutations change the receptor binding affinity for the RANTES analogs, HEK293T cells were transfected with plasmids encoding WT CCR5 or the two mutants with most extreme phenotype, Y37A and E283A. Expression of the WT or mutated CCR5 was assessed after 48 h by flow cytometry, and membranes for receptor binding assays were prepared as described in reference 8. The IC₅₀s for inhibition of [¹²⁵I]CCL3 binding by each of the 4 RANTES analogs differed by less than 2-fold when we compared WT CCR5 to Y37A or E283A mutant CCR5 (Fig. 3C). These small differences in binding affinities contrast with the 80- to 100-fold differences in inhibitory potencies and thus suggest that binding affinity makes a minor contribution to the results presented in Fig. 3A.

Overall, these data support the hypothesis (6) that the TM domains of CCR5 are a finely tuned molecular switch for modulating receptor activity in different ways and that by accessing this region via residues in their modified N termini, chemokine analogs achieve exquisite modulation of CCR5 function that impacts receptor signaling and/or sequestration and the potency of coreceptor inhibition.