HIV-1 exploits CCR5 conformational heterogeneity to escape inhibition by chemokines

Abstract

CC chemokine receptor 5 (CCR5) is a receptor for chemokines and the coreceptor for R5 HIV-1 entry into CD4⁺ T lymphocytes. Chemokines exert anti–HIV-1 activity in vitro, both by displacing the viral envelope glycoprotein gp120 from binding to CCR5 and by promoting CCR5 endocytosis, suggesting that they play a protective role in HIV infection. However, we showed here that different CCR5 conformations at the cell surface are differentially engaged by chemokines and gp120, making chemokines weaker inhibitors of HIV infection than would be expected from their binding affinity constants for CCR5. These distinct CCR5 conformations rely on CCR5 coupling to nucleotide-free G proteins (^NFG proteins). Whereas native CCR5 chemokines bind with subnanomolar affinity to ^NFG protein-coupled CCR5, gp120/HIV-1 does not discriminate between ^NFG protein-coupled and uncoupled CCR5. Interestingly, the antiviral activity of chemokines is G protein independent, suggesting that “low-chemokine affinity” ^NFG protein-uncoupled conformations of CCR5 represent a portal for viral entry. Furthermore, chemokines are weak inducers of CCR5 endocytosis, as is revealed by EC₅₀ values for chemokine-mediated endocytosis reflecting their low-affinity constant value for ^NFG protein-uncoupled CCR5. Abolishing CCR5 interaction with ^NFG proteins eliminates high-affinity binding of CCR5 chemokines but preserves receptor endocytosis, indicating that chemokines preferentially endocytose low-affinity receptors. Finally, we evidenced that chemokine analogs achieve highly potent HIV-1 inhibition due to high-affinity interactions with internalizing and/or gp120-binding receptors. These data are consistent with HIV-1 evading chemokine inhibition by exploiting CCR5 conformational heterogeneity, shed light into the inhibitory mechanisms of anti–HIV-1 chemokine analogs, and provide insights for the development of unique anti–HIV molecules.

C-C chemokine receptor 5 (CCR5) is the principal coreceptor for entry of HIV type 1 (HIV-1), used together with CD4 to enter and infect target cells (1) and a receptor for agonist [C-C chemokine ligand 3 (CCL3)/macrophage inflammatory protein-1α (MIP-1α), CCL4/MIP-1β, CCL5/regulated upon activation, normal T-cell expressed and secreted (RANTES)] and antagonist/weak partial agonist (CCL7/monocyte chemotactic protein 3 (MCP-3)] chemokines (2, 3). The native agonist chemokine ligands of CCR5 induce conformational changes in the receptor that promote activation of pertussis toxin (PTX)-sensitive, heterotrimeric αβγ G proteins (Gi/o-type G proteins) by catalyzing an exchange of GTP for GDP on the Gα subunit. The GTP-bound Gα subunit and the Gβγ dimer then trigger intracellular signaling pathways involved in chemotaxis and activation of leukocytes (4).

Native CCR5 chemokines inhibit infection of R5-tropic HIV-1 in vitro. This occurs via two mechanisms: sterically preventing the viral envelope glycoprotein gp120 from binding to the coreceptor and reducing cell surface coreceptor levels by inducing receptor down-regulation (5–7). They are secreted by a number of cell types and in particular immune cells including R5 HIV-1 target cells (6, 8, 9). The potential role of native CCR5 chemokines in blocking HIV-1 transmission and progression has been extensively studied (9–12), but their efficacy as protective factors remains a matter of debate (13, 14). A major paradox relates to the observation that native CCR5 chemokines show lower antiviral potencies than would be expected based on their CCR5 binding affinity constants (15–18), which are in the subnanomolar range (2, 19, 20), much lower than the corresponding value for the HIV-1 envelope glycoprotein gp120, which is ∼10 nM (19, 21).

A number of CCR5 chemokine analogs with improved antiviral potency have been identified, including N-terminally modified RANTES analogs with agonist [aminooxypentane (AOP)-, PSC- or 6P4-RANTES] or antagonist features (5P12- or 2P3-RANTES), which represent promising molecules as topical microbicides (18, 22). Although the enhanced potency of agonist analogs can be explained in terms of their increased capacity to induce CCR5 down-regulation (23), the inhibitory mechanism of antagonist analogs, which neither activate G protein signaling nor induce receptor down-regulation, is more elusive. It was speculated that it might involve increased steric blockade of CCR5, but competition binding assays using labeled CCL4 as a tracer did not show any significant increase in CCR5 binding affinity (22).

In this study, we present evidence that conformationally different CCR5 subpopulations with distinct chemokine binding capacities are present at the surface of HIV-1 target cells. In particular, a fraction of receptors shows strikingly low binding affinity for native CCR5 chemokines, providing an explanation for why native CCR5 chemokines have unexpectedly low anti–HIV-1 potencies. Our results also shed further light on the inhibitory mechanism of chemokine analogs, showing that they overcome the challenge of the chemokine low-affinity CCR5 population through (i) more efficient receptor down-regulation and/or (ii) increased binding affinity for gp120-binding receptors. Overall, these findings explain how R5 HIV-1 could escape from inhibition by native CCR5 chemokines in the course of infection and provide clues for the development of unique chemokine analogs as HIV inhibitors.

Results

Distinct CCR5 Populations Are Differentially Used by HIV-1 gp120 and Chemokines.

Chemokines and the CD4-bound form of HIV-1 envelope glycoprotein gp120 competitively bind to overlapping regions of CCR5 (24). Thus, to investigate whether distinct CCR5 populations interact differently with chemokines and gp120, we first tested unlabeled native chemokines (CCL-3, -4, -5, or -7) or chemokine analogs (AOP-, PSC-, 2P3-, 5P12-, or 6P4-RANTES) for their ability to inhibit binding of either ¹²⁵I-labeled CCL3 (¹²⁵I-CCL3) or the ³⁵S-labeled gp120 from the HIV-1 primary strain Bx08 (³⁵S-gp120_Bx08) on membranes from CCR5-expressing HEK 293T cells (HEK-R5 cells) (Fig. 1 and Table 1). Using the data obtained, we calculated the affinity constant values (K_i) of the competing ligands for receptors using the Cheng and Prusoff equation (SI Materials and Methods).

Fig. 1.

Chemokine- and gp120-binding CCR5 represent different receptor populations. Binding of 0.1 nM ¹²⁵I-CCL3 (A) or 10 nM ³⁵S-gp120_Bx08 (in the presence of 30 nM soluble CD4) (B–D) was displaced by increasing amounts (A, C, and D) or a 100-nM concentration (B) of unlabeled chemokines. Results were normalized for nonspecific binding (0%) and specific binding in the absence of competitors (B₀, 100%) and fitted according to a one-site (A, 6P4 and CCL7 in C, and 2P3 and 5P12 in D) or a two-site (16.2 < F value < 93.3 with P < 0.0001 for CCL3 in C and CCL4 and PSC in D) competitive binding model. In B, data points are means ± SEM of five independent determinations. A, C, and D show representative experiments of three to five performed independently.

Table 1.

CCR5 binding affinity constants of native chemokines and RANTES analogs and their half-maximal inhibitory concentrations (IC₅₀) of HIV-1 infection in CD4⁺ T-cells and HeLa P4C5 cells

Chemokines	KD (nM)	KiCCL3 (nM)	Kigp120 (nM) - Gpp(NH)p	Kigp120 (nM) + Gpp(NH)p	IC50 (nM) T-CD4+ -PTX	IC50 (nM) T-CD4+ +PTX	IC50 (nM) P4C5 -PTX	IC50 (nM) P4C5 +PTX
CCL3	0.25 ± 0.05	0.06 ± 0.002	High 0.6 ± 0.1 (43.6 ± 9.7%); low > 1,000	High 0.58 ± 0.4 (16.4 ± 6.6 %); low 597 ± 185.4	106.9 ± 42.7	143	> 1,000	> 1,000
CCL4	0.37 (a)	0.21 ± 0.1	High 0.43 ± 0.1 (25 ± 3 %); low 44.7 ± 6.5	High und (b) low 86 ± 11 (b)	4.5 ± 1.8	6.7 ± 2.8	444 ± 59	366 ± 62
CCL5	0.75 ± 0.15	3 ± 1.1	> 1,000	-	-	-	> 1,000	> 1,000
CCL7	-	119.4 ± 31.7	34.1 ± 8.2	-	> 1,000	-	> 1,000	> 1,000
AOP	-	1.13 ± 0.18	-	-	-	-	-	-
PSC	-	1.89 ± 0.95	High 1.47 ± 0.03 (31.7 ± 0.5 %); low 215 ± 35.7	-	0.14 ± 0.03	0.12 ± 0.03	0.44 ± 0.2	0.48 ± 0.02
2P3	-	0.31 ± 0.11	2.72 ± 0.21	-	0.96 ± 0.31	-	-	-
5P12	-	0.26 ± 0.13	3.51 ± 1.8	-	0.036 ± 0.015	0.038 ± 0.013	0.87 ± 0.45	0.96 ± 0.45
6P4	-	0.055 ± 0.02	2.93 ± 0.23	-	0.046 ± 0.01	0.047	0.82 ± 0.12	0.64 ± 0.12

K_i^CCL3 and K_i^gp120 represent the equilibrium dissociation constants for interaction of chemokines with CCR5 determined in competition assays using either ¹²⁵I-CCL3 or ³⁵S-gp120 as a tracer, respectively, in the presence or in the absence of Gpp(NH)p. K_D values are the equilibrium dissociation constants of radiolabeled chemokine-CCR5 complexes deduced from saturation binding experiments. The K_D values for ¹²⁵I-CCL3 and ¹²⁵I-CCL5 are deduced from the experiments shown in Figs. 3B and 3C, respectively. For details on the B_max values, see the text covering these figures. (a) The reported K_D value is from ref. 20. (b) The reported K_i values for interaction of CCL4 with ³⁵S-gp120-binding CCR5 in the presence of Gpp(NH)p is from ref. 19. Except for inhibition of infection of T-cells by CCL3 or 6P4 in the presence of PTX, which was performed once, values represent means ± SD of at least three independent determinations. The independent experiments in CD4⁺ T-cells represent experiments run in cells obtained from different donors. Numbers in parentheses represent the percentages of receptors, which are in a high-affinity state for the considered agonist chemokines. Und, undetectable.

Except for CCL7, displacement of ¹²⁵I-CCL3 binding revealed high affinities of competitors for CCR5, with K_i values in the nanomolar range or lower (Table 1). The K_i values obtained for the native CCR5 agonists CCL-3, -4, and -5 are comparable to the K_D values determined for these ligands in saturation binding assays (Table 1), consistent with binding of these chemokines to a similar class of high-affinity receptors in both competition and saturation assays. In contrast, CCL-3, -4, and -5, as well as the chemokine analogs AOP- and PSC-RANTES, only partly displaced ³⁵S-gp120_Bx08 binding when used at a 100-nM concentration (Fig. 1B), suggesting that they have a lower affinity for ³⁵S-gp120_Bx08-binding receptors compared with ¹²⁵I-CCL3-binding receptors. In contrast, the observation that ³⁵S-gp120_Bx08 binds marginally to CCR5 in the presence of 100 nM 2P3-, 5P12-, or 6P4-RANTES (Fig. 1B) suggests that these chemokine analogs preserve high-affinity interactions with ³⁵S-gp120_Bx08-binding CCR5.

In dose–response experiments, 2P3-, 5P12-, or 6P4-RANTES and ³⁵S-gp120_Bx08 competed for binding to an apparent single class of receptors, as is revealed by monophasic competitive binding curves (Fig. 1 C and D). The K_i values in the nanomolar range calculated for these chemokine analogs confirm high-affinity interactions with the ³⁵S-gp120_Bx08-binding receptors (Table 1). CCL7 similarly bound to a single class of ³⁵S-gp120_Bx08-binding CCR5 for which the affinity of the chemokine was higher than that for ¹²⁵I-CCL3-binding CCR5 (K_i = 34 vs. 119 nM). In contrast, displacements of ³⁵S-gp120_Bx08 binding by the agonists CCL3 (Fig. 1C), CCL4, and PSC-RANTES (Fig. 1D) gave biphasic curves, consistent with the presence of two distinct ³⁵S-gp120_Bx08 receptor populations, one with high affinity for these chemokines, the other with significantly lower affinity. CCL-3, -4, and PSC-RANTES had K_i values for interaction with the “high-chemokine affinity” receptor population similar to those determined in ¹²⁵I-CCL3 displacement assays (Table 1), suggesting that the high-chemokine affinity population of ³⁵S-gp120_Bx08-binding receptors and ¹²⁵I-CCL3-binding receptors represent the same receptors. Interestingly, the K_i values obtained for the “low-chemokine affinity” CCR5 population range from a few tens of nM up to more than 10⁻⁶ M (Table 1), thus exceeding the K_D value for the interaction of ³⁵S-gp120_Bx08 with CCR5 determined in saturation assays (∼10 nM) (19). Hence, this fraction of CCR5 has a higher affinity for gp120 than for native CCR5 chemokines.

Poor Ability of Native CCR5 Chemokines to Displace gp120 Binding to CCR5 Correlates with Low Antiviral Activity.

We next investigated whether low anti-HIV potency of native chemokines is related to low-affinity interactions with gp120-binding receptors. For that purpose, we infected activated CD4⁺ T lymphocytes, which represent the major target cells for R5 HIV-1, and HeLa P4C5 cells with infectious NL4-3Ren-based HIV-1 expressing the envelope glycoprotein gp160 of the Bx08 strain and Renilla luciferase as a reporter gene (Bx08Ren viruses; see SI Materials and Methods for details) in the presence of native CCR5 chemokines or chemokine analogs (Table 1 and Fig. S1 A and B). Except for the chemokine analog PSC-RANTES, the antiviral potencies of all other chemokines correlated better with their ability to displace the binding of ³⁵S-gp120_Bx08 than that of ¹²⁵I-CCL3. In particular, the native CCR5 agonists (CCL-3, -4, and -5) that show low-affinity interactions with a proportion of gp120-binding receptors had many more weaker antiviral activities than chemokine analogs, even though both groups of molecules have comparable affinities for ¹²⁵I-CCL3-binding CCR5 (Table 1). Similar results were obtained using five other viruses that were generated by inserting the R5 env gp120 or env gp160 sequences from laboratory-adapted (JRRen) or primary (25Ren, 34Ren, 50Ren, or 58Ren) viruses in place of the corresponding regions of the NL4-3 env sequence in the pNL4-3Ren plasmid containing the Renilla luciferase gene in place of the nef gene (see SI Materials and Methods for details) (Fig. S2).

The potent antiviral activities of the antagonist RANTES analogs 5P12 and 2P3, which do not down-regulate CCR5 (see Fig. 5B and ref. 22), reflect their increased binding affinities detected using ³⁵S-gp120_Bx08 as a tracer (Fig. 1). The enhanced potency of PSC-RANTES, which occurs despite its relatively low capacity to compete with ³⁵S-gp120_Bx08 for binding to CCR5, is likely to be due to its enhanced capacity to induce CCR5 down-regulation (23). To test this hypothesis, we performed infection inhibition experiments under conditions where receptor down-regulation is suppressed. HeLa P4C5 cells were preincubated for 2 h with PSC-RANTES (40 nM), 5P12-RANTES (40 nM), or the CCR5 inverse agonist maraviroc (MVC, 20 μM) at either 37 °C or at 4 °C, a temperature at which receptor endocytosis does not occur. Bx08Ren virus was then added to the cells, which were incubated for a further 2 h at 4 °C, then washed in cold PBS, warmed to 37 °C for 15 min to allow entry of attached viruses, trypsin treated to remove residual viruses, and incubated for 48 h at 37 °C (Fig. 2). Under conditions where CCR5 down-regulation is suppressed, the antiviral activity of PSC-RANTES was almost completely abrogated, but the inhibitory potency of 5P12-RANTES or MVC was unaffected, in accordance with our results in Fig. 1 showing that PSC- and 5P12-RANTES are weak and potent inhibitors of gp120 binding to CCR5. Hence unlike 5P12-RANTES and maraviroc, PSC-RANTES owes a large part of its inhibitory activity to its capacity to induce CCR5 down-regulation. Importantly, these results also validate the notion that the receptors interacting with monomeric gp120/soluble CD4 complexes in the binding assays presented here (Fig. 1 B–D) and those that are used by infectious virus particles at the surface of intact cells in infection assays extensively overlap and would represent similar receptor populations.

Fig. 5.

Native chemokines and RANTES analogs induce CCR5 down-regulation in a G protein independent manner. Native chemokines have a low potency in internalizing CCR5. (A–D) Cell surface expression of Flag-tagged receptors stably expressed in HEK 293 cells was detected by flow cytometry (SI Materials and Methods). (A) CCR5 cell surface expression level (in mean fluorescence intensity) was measured after stimulation or not by 300 nM PSC-RANTES (90 min, 37 °C) of cells treated or not with PTX. In B and C, CCR5 down-regulation is expressed as percent of maximum effect induced by PSC- or 6P4-RANTES, which showed equal potency (Table S1) and efficacy (A) in the assay. In B, cells were stimulated by chemokines for 90 min at 37 °C. Data were fitted to a sigmoidal dose–response model with a variable slope, with Lower and Upper values constrained to equal 0% and 100%, respectively. C shows time-dependent down-regulation of CCR5 induced by 100 nM PSC-RANTES or CCL4. Data were analyzed using a one-phase exponential association function. In D, the expression level of either WT- or 349-CCR5 at the surface of cells stimulated by CCL4 or PSC-RANTES (90 min, 37 °C) is expressed as percent of receptor expression level at the surface of untreated cells (100%). Results are representative of two (C) or three (A, B, and D) independent experiments.

Fig. 2.

Steric inhibition of gp120 binding to CCR5 does not contribute to the anti–HIV-1 activity of PSC-RANTES. HeLa P4C5 cells were incubated with MVC, 5P12-, or PSC-RANTES at 37 °C or 4 °C before being infected by Bx08Ren viruses and treated as indicated in the text. Results represent the luciferase activity in the cell lysates, expressed as relative light units (RLUs). One representative experiment of five independent determinations is shown. Uninfected cells (NI) represent negative controls.

CCR5 Coupling to Nucleotide-Free G Proteins Differentially Regulates Native Agonist Chemokine and gp120 Binding.

It has been established that conformations of G protein-coupled receptors with high-affinity for agonists are stabilized by coupling to guanine nucleotide-free G proteins (^NFG proteins), and that the receptors are induced to shift toward low-affinity conformations as soon as G proteins are occupied by nucleotides (25). Similarly to what we and others showed in mammalian cell lines (19, 26), we observed that CCR5 coupling to ^NFG proteins also stabilizes the receptor in a high-affinity conformation for agonists in HIV-1 target cells. Indeed, the non hydrolysable GTP analogs guanosine 5′-O-(γ-thio)-triphosphate (GTPγS) and guanosine 5′-(β,γ-imido)triphosphate [Gpp(NH)p] or PTX, which inactivates G_i/o-proteins, decreased ¹²⁵I-CCL3 binding to CCR5 expressed in human lymphoblastoid CD4⁺ T-cell lines (A3.01-R5 cells) or primary T lymphocytes to levels approaching that of nonspecific binding (Fig. 3A). Saturation binding of ¹²⁵I-CCL3 to membranes from HEK-R5 cells further revealed that Gpp(NH)p decreases the maximum number of binding sites for the chemokine (B_max) from 10.7 ± 1.3–2.9 ± 0.7 pmol/mg of protein (Fig. 3B), while only slightly affecting the K_D value from 0.25 ± 0.05–0.46 ± 0.08 nM, indicating that Gpp(NH)p reduces the number of receptors that are of high affinity for ¹²⁵I-CCL3. In line with this, 0.1 nM ¹²⁵I-CCL3 showed only background levels of binding to membranes from HEK cells expressing the R126N-CCR5 mutant, which does not activate G proteins (27) (Fig. 3D). Similarly to CCL3 and CCL4 (26), high-affinity binding of CCL5 also required CCR5 coupling to ^NFG proteins (Fig. 3C).

Fig. 3.

CCR5 coupling to ^NFG proteins differentially influences native agonist chemokine and gp120 binding. (A) Total binding of 0.2 nM ¹²⁵I-CCL3 to 5.10⁵ A3.01-R5 cells (Left) or membranes from CD4⁺ T cells (15 μg of proteins) (Right) was measured in the presence or absence (control) of GTPγS or Gpp(NH)p or after treatment of cells with PTX. Nonspecific binding was determined using the antagonist TAK779 or MVC. *P < 0.05; **P < 0.01 compared with controls in unpaired two-tailed Student t test. B and C are saturation experiments of ¹²⁵I-CCL3 and ¹²⁵I-CCL5 binding to HEK-R5 cell membranes, respectively. Specific binding was measured in the presence or absence of Gpp(NH)p. Total binding of 0.1 nM ¹²⁵I-CCL3 (D) or 10 nM ³⁵S-gp120 from the HIV-1 strains 25, 34, or Bx08 in complex with sCD4 (E) to membranes from HEK 293T cells expressing WT-CCR5 or R126N-CCR5 (R/N) was measured in the presence or absence (control) of Gpp(NH)p and/or MVC (nonspecific binding). Equal amounts of WT-CCR5 and R126N-CCR5 at the cell surface were confirmed by flow cytometry. Saturation binding experiments of ³⁵S-gp120/sCD4 complexes revealed K_D values (in nanomoles) of 7.5, 8.3, and 9.9 for gp120₂₅, gp120₃₄, and gp120_Bx08, respectively. (F) Displacement of ³⁵S-gp120_Bx08 binding by CCL3 was measured in the absence or presence of Gpp(NH)p. Data were fitted according to a two-site competitive binding model (F = 42 with P < 0.0001 and F = 5.5 with P = 0.0062 for data in the absence and presence of Gpp(NH)p, respectively). Representative experiments of at least three independent experiments are shown.

In contrast to native CCR5 agonist chemokines, R5 HIV-1 gp120 acts as an antagonist/weak partial agonist for CCR5, as it does not discriminate between ^NFG protein-coupled or uncoupled CCR5 and binds equally well to R126N-CCR5 and wild-type CCR5, both in the presence and absence of Gpp(NH)p (Fig. 3E). This led us to hypothesize that the biphasic competitive binding curves obtained with native CCR5 agonist chemokines using ³⁵S-gp120_Bx08 as a tracer is a reflection of (i) the existence of populations of both ^NFG protein-coupled and ^NFG protein-uncoupled receptors with respectively high and low affinity for these chemokines and (ii) the capacity of gp120 to bind indiscriminately to either population. We tested this hypothesis by repeating the competition experiments of ³⁵S-gp120_Bx08 binding by CCL3, in the presence and absence of Gpp(NH)p (Fig. 3F). Treatment with Gpp(NH)p decreased the proportion of high-chemokine affinity receptors versus low-chemokine affinity receptors from 43% to 16% (P = 0.016 in unpaired, two-tailed Student t test), without affecting the K_i value of the low-chemokine affinity receptor population (Table 1 and Fig. 3F). This result is consistent with our previous observation that Gpp(NH)p eliminates the fraction of ³⁵S-gp120_Bx08-binding CCR5 that binds CCL4 with high affinity (19).

Chemokine-Mediated Inhibition of HIV Infection and CCR5 Endocytosis Are G Protein-Independent Processes.

Based on the observation that HIV envelope binds indiscriminately to high-chemokine affinity ^NFG protein-coupled CCR5 and low-chemokine affinity ^NFG protein-uncoupled receptors, we hypothesized that infection in the presence of chemokine ligands would be more likely to occur via the low-chemokine affinity ^NFG protein-uncoupled receptors, and that the chemokine ligands would be required to engage this population of receptors to achieve inhibition of infection. This would explain why native CCR5 agonist chemokines have low potency as HIV inhibitors.

To test this hypothesis, we treated activated CD4⁺ T cells, A3.01-R5 or HeLa P4C5 cells with PTX and then infected them with R5 HIV-1 in the presence of native CCR5 chemokines or chemokine analogs (Fig. 4 and Figs. S2 and S3). PTX attenuated ¹²⁵I-CCL3 binding to target cells (Fig. S3A and Fig. 3A) and abrogated chemokine-induced chemotaxis (Fig. 4A), indicating that CCR5 coupling with G_i/o proteins is required for both high-affinity binding of the ligands and signal transduction. In contrast, PTX changed neither viral infectivity (Fig. S2) nor the potency of chemokines to block infection (Fig. S2 and S3B, and Fig. 4B and Table 1). This suggests that CCR5 engagement by HIV-1 is independent of G proteins and that high-affinity binding of ligands to ^NFG protein-coupled CCR5 does not make a significant contribution to their capacity to inhibit infection.

Fig. 4.

Chemokine-mediated inhibition of HIV-1 infection, but not chemotaxis, is independent of G proteins. PTX-treatment impaired PSC-RANTES–mediated chemotaxis of A3.01-R5 cells (A) but not the ability of the chemokine analog to inhibit infection of these cells by the Bx08Ren viruses (B). Representative experiments of at least three independent experiments are shown.

These results also imply that CCR5 down-regulation, which contributes to the anti-HIV activity of agonist chemokines, is not dependent on engagement of high-chemokine affinity ^NFG protein-coupled CCR5. To address this possibility, we tested the ability of CCL-3, -4, PSC-, 6P4-, and 5P12-RANTES to down-regulate Flag-tagged CCR5 in HEK 293 cells (28) with and without PTX, which decreases high-affinity binding of ¹²⁵I-CCL3 by 87.4 ± 11.8% (Fig. 5 A–C and Table S1). PTX influenced neither the efficacy nor the potency (EC₅₀) of chemokines to down-regulate CCR5 (Fig. 5 A and B and Table S1). Moreover, the kinetic rates of CCR5 down-regulation induced by PSC-RANTES and CCL4 were unchanged by PTX treatment (t_1/2 (min) = 3.3 ± 0.6 vs. 3.2 ± 0.9 and 8.9 ± 0.8 vs. 8.7 ± 1.7 for PSC and CCL4 in the absence or in the presence of PTX, respectively) (Fig. 5C). Hence interaction with high-chemokine affinity ^NFG protein-coupled CCR5 is not a requirement for the induction of CCR5 down-regulation by its ligands.

CCR5 Down-Regulation Involves Low-Affinity Interactions of Native CCR5 Agonists with Internalizing Receptors.

PSC- and 6P4-RANTES had nanomolar EC₅₀ values for CCR5 down-regulation (Table S1), whereas the EC₅₀ values for CCL-3 and -4 exceeded by more than two or three orders of magnitude their K_i value for ^NFG protein-coupled CCR5 (Table 1 and Table S1). The differential abilities of chemokines to down-regulate CCR5 could be due to internalization-competent CCR5 that might represent a receptor subpopulation to which CCL-3 and -4, but not PSC- and 6P4-RANTES, bind with a low affinity. Alternatively, but not exclusively, RANTES analog-induced inhibition of receptor recycling could also contribute to their potent ability to down-regulate CCR5, as previously suggested (23).

To assess these hypotheses, we compared the abilities of PSC-RANTES and CCL4 to down-regulate WT-CCR5 or the 349-CCR5 mutant, which does not recycle back to the cell surface (28) (Fig. 5D). Compared with WT-CCR5, CCL4 down-regulated 349-CCR5 with higher potency (sixfold, Table S1) and efficacy, confirming that receptor recycling interferes to some extent with the ability of CCL4 to down-regulate CCR5. PSC-RANTES also down-regulated 349-CCR5 more efficiently than WT-CCR5 (93.9 ± 1.6% vs. 72.6 ± 3.3%, respectively), albeit with comparable potencies (Table S1), suggesting that PSC-RANTES slows down CCR5 recycling but does not prevent it. However, CCL4 induced endocytosis of 349-CCR5 with a 18-fold higher EC₅₀ value compared with PSC-RANTES, indicating that low potency of CCL4 in down-regulating CCR5 is modestly due to its inability to prevent receptor recycling. Rather, this EC₅₀ value for endocytosis of 349-CCR5 by CCL4 (47 nM) is similar to its K_i value for interaction with the low-chemokine affinity ^NFG protein-uncoupled population of CCR5 (44.7 nM). This suggests that native CCR5 agonist chemokines have a low potency to down-regulate CCR5, owing to their inability to prevent CCR5 recycling and, above all, to their low affinity for ^NFG protein uncoupled CCR5 undergoing endocytosis.

Discussion

Our findings indicate that inhibition by native CCR5 chemokines of HIV-1 infection is hindered by a proportion of receptors that exists in a low-chemokine affinity conformation at the target cell surface. This likely explains the discrepancy between the apparently high CCR5 affinities measured previously for native chemokine ligands (2, 20) and their relatively modest potency as entry inhibitors (15–18). Different CCR5 conformations with distinct pharmacological and antigenic properties have been described (27, 29). Here, we found that the apparent affinity of native chemokines and RANTES analogs for CCR5 varies depending on whether ¹²⁵I-CCL3 or ³⁵S-gp120 is used as a tracer in competition experiments (Fig. 1), identifying that distinct receptor populations interact with ¹²⁵I-CCL3- and ³⁵S-gp120-binding receptors. Indeed, we further showed that whereas high-affinity binding of ¹²⁵I-CCL3 requires CCR5 to be coupled to ^NFG proteins, ³⁵S-gp120 binds with the same affinity to both high-chemokine affinity ^NFG protein-coupled CCR5 and low-chemokine affinity ^NFG protein-uncoupled CCR5. Although native CCR5 agonist chemokines interact with subnanomolar affinities with ¹²⁵I-CCL3-binding receptors, they bind to the low-chemokine affinity population of ³⁵S-gp120-binding receptors with affinities lower than those of primary gp120 (Fig. 3E, legend), thereby contributing to limiting their antiviral potency.

PTX treatments of HIV-1 target cells had no effect on virus entry and replication (Fig. S2), suggesting that similarly to gp120, HIV-1 attachment to CCR5 is independent of G proteins, in agreement with our previous data showing that the non-G protein coupling mutant receptor R126N-CCR5 supports HIV entry (30). The observations that CCR5 is constitutively active (19, 27) and that preformed receptor/G protein complexes exist in living cells (31) suggest that an equilibrium may exist between ^NFG protein-coupled and -uncoupled CCR5 in HIV target cells. On the other hand, ^NFG proteins that stabilize high-agonist affinity conformations of CCR5 likely represent a minor fraction of total G proteins in intact cells (25). In line with this, our observations that PTX does not change the anti-HIV potency of chemokines (Table 1) suggest that high-chemokine affinity ^NFG protein-coupled receptors play a minor role in the antiviral activity of chemokines and that low-chemokine affinity ^NFG protein-uncoupled CCR5 represent a portal for HIV entry into target cells. Interaction with ^NFG protein-uncoupled CCR5 could allow HIV to evade inhibition by the chemokines secreted in the surrounding environment. At the same time, through high-affinity interactions with receptors coupled to ^NFG proteins, these chemokines would still be capable of activating target cells, facilitating viral replication (32), and recruiting target cells into sites of HIV replication.

Whereas it is commonly accepted that coreceptor down-regulation contributes to chemokine inhibition of HIV-1 infection (5–7), we showed that native CCR5 agonist chemokines exhibit a weak ability to down-regulate CCR5, as is indicated by EC₅₀ values for CCR5 down-regulation by the chemokines that are close to their K_i values for interaction with the low-chemokine affinity population of CCR5. Preventing CCR5 recycling only modestly increases the ability of CCL4 to down-regulate CCR5, but several observations suggest that CCR5 down-regulation involves low-affinity interactions of native chemokines with ^NFG protein uncoupled CCR5. Indeed, we previously demonstrated that R126N-CCR5 does not trigger G protein signaling but retains β-arrestin–dependent endocytosis, indicating that both processes are independent functions of CCR5 mediated by different receptor conformations (27). R126N-CCR5 is also altered in its ability to bind ¹²⁵I-CCL3 (Fig. 3D) and ¹²⁵I-CCL4 (26), indicating that the ^NFG protein-coupled conformation of CCR5 required for high-affinity binding of agonist chemokines is distinct from the CCR5 conformation undergoing endocytosis. This conclusion agrees with our present results that PTX that inhibits CCR5/G protein coupling and high-affinity binding of native agonist chemokines preserves CCR5 endocytosis (Fig. 5). Overall, these data support the view that natural agonist chemokines engage low-affinity interactions with internalizing CCR5, hence explaining why they are weak inducers of CCR5 endocytosis and inhibitors of HIV infection.

Structurally different agonists can stabilize distinct receptor conformations with distinct signaling outcomes (33). In particular, ligands referred to as biased ligands differentially stimulate G protein- and β-arrestin–dependent signaling pathways (34). Similarly, PSC-RANTES and CCL4 have comparable binding affinities for CCR5 (Fig. 1) and potencies for activating G proteins in a ³⁵S-GTPγS binding assay (EC₅₀ = 4.1 ± 0.9 and 6.3 ± 0.4 nM for CCL4 and PSC, respectively, n = 2), whereas PSC-RANTES is substantially more potent in internalizing CCR5 (Fig. 5), suggesting that the two ligands stabilize distinct CCR5 conformations. In fact, the EC₅₀ value for PSC-RANTES to down-regulate CCR5 is roughly equal to its K_i value for interaction with ^NFG protein-coupled, ¹²⁵I-CCL3-binding CCR5, suggesting that PSC-RANTES preserves high-affinity interactions with internalizing CCR5, despite the fact that these receptors are not coupled to ^NFG proteins. It could be that PSC-RANTES stabilizes a β-arrestin–coupled conformation of CCR5 for which it maintains a high affinity, similarly to other receptors, which are in a high-affinity state for agonists when complexed with arrestins (35). Finally, the robust CCR5 down-regulation induced by PSC-RANTES explains why the molecule preserves a strong antiviral activity despite having a low affinity for gp120-binding receptors. Indeed, preventing CCR5 endocytosis virtually abrogates PSC-RANTES–mediated inhibition of HIV infection (Fig. 2), indicating that steric inhibition of gp120 binding to CCR5 plays a marginal role in the antiviral activity of PSC-RANTES.

The antagonists 5P12- and 2P3-RANTES appeared instead to act solely by potently blocking the interaction between gp120 and CCR5 (Fig. 1). Using these antagonists together with CCR5-internalizing molecules such as PSC-RANTES could in principle represent an interesting therapeutic perspective, albeit no studies have shown yet whether these different analogs have additive inhibitory effects in HIV infection. Interestingly however, we showed here that 6P4-RANTES resembles both 5P12- and PSC-RANTES in that it preserves high affinity for gp120 binding receptors (Fig. 1C) and down-regulates CCR5 at nanomolar concentrations. Considered altogether, these results are consistent with 5P12-, 6P4-, and PSC-RANTES stabilizing different CCR5 conformations. In line with this, mutations in the transmembrane domains of CCR5 were found to modulate in different ways their ability to inhibit HIV infection, indicating that they have different structural constraints for HIV-1 inhibition (36). Notably, these mutations did not change the ability of the RANTES analogs to inhibit ¹²⁵I-CCL3 binding to CCR5 (36), again strengthening the notion that chemokines have different structural requirements for interacting with ^NFG protein coupled, CCL3-binding receptors, and inhibiting gp120 binding and HIV infection.

Overall, our findings document that both mechanisms whereby native CCR5 chemokines exert their anti-HIV activity, inhibition of gp120/CCR5 interactions, and CCR5 down-regulation, are strongly limited by virtue of their low-affinity interactions with a proportion of CCR5 conformations. Overcoming these limitations explains why RANTES analogs show improved antiviral potencies compared with their natural counterparts and should help guide the development of new anti-HIV agents. Finally, these limitations could make it difficult to accomplish the blockade of R5 HIV-1 isolates by chemokines in vivo and contribute to their preferential transmission and propagation in the early stages of infection.

Materials and Methods

Radioactive chemokines were from PerkinElmer Life Sciences. CCL3 and CCL7 were purchased from R&D Systems. Other information regarding materials (chemokines, HIV-1 glycoproteins, viruses, and cells) and experimental procedures (radioligand binding, chemotaxis, receptor down-regulation, and infection inhibition assays) is provided in SI Materials and Methods.