Research ArticlePharmacology

G protein subtype–specific signaling bias in a series of CCR5 chemokine analogs

See allHide authors and affiliations

Sci. Signal.  16 Oct 2018:
Vol. 11, Issue 552, eaao6152
DOI: 10.1126/scisignal.aao6152

Biasing CCR5 against inflammation

The G protein–coupled chemokine receptor CCR5 is required for the cellular entry of HIV-1. The ideal drug targeting CCR5 to block HIV-1 infection would induce receptor internalization without triggering the undesirable effect of inflammation. Thus, Lorenzen et al. characterized the signaling properties of two endogenous ligands of CCR5, RANTES and MIP-1α, and several RANTES analogs in cells expressing CCR5 with Gi/o or Gq/11 subunits. They found that CCR5 inhibited generation of the second messenger cAMP by stimulating Gi/o and induced Ca2+ signaling through Gq/11 and that the RANTES analogs differentially activated Gi/o and Gq/11. The authors hypothesize that RANTES analogs that preferentially stimulate Gi/o may block HIV-1 infection without inducing inflammation.

Abstract

Chemokines and some chemical analogs of chemokines prevent cellular HIV-1 entry when bound to the HIV-1 coreceptors C-C chemokine receptor 5 (CCR5) or C-X-C chemokine receptor 4 (CXCR4), which are G protein–coupled receptors (GPCRs). The ideal HIV-1 entry blocker targeting the coreceptors would display ligand bias and avoid activating G protein–mediated pathways that lead to inflammation. We compared CCR5-dependent activation of second messenger pathways in a single cell line. We studied two endogenous chemokines [RANTES (also known as CCL5) and MIP-1α (also known as CCL3)] and four chemokine analogs of RANTES (5P12-, 5P14-, 6P4-, and PSC-RANTES). We found that CCR5 signaled through both Gi/o and Gq/11. IP1 accumulation and Ca2+ flux arose from Gq/11 activation, rather than from Gβγ subunit release after Gi/o activation as had been previously proposed. The 6P4- and PSC-RANTES analogs were superagonists for Gq/11 activation, whereas the 5P12- and 5P14-RANTES analogs displayed a signaling bias for Gi/o. These results demonstrate that RANTES analogs elicit G protein subtype–specific signaling bias and can cause CCR5 to couple preferentially to Gq/11 rather than to Gi/o signaling pathways. We propose that G protein subtype–specific signaling bias may be a general feature of GPCRs that can couple to more than one G protein family.

INTRODUCTION

G protein–coupled receptors (GPCRs) can transmit extracellular cues from many structurally distinct ligands. GPCRs have been envisioned as on-off switches that become activated through the binding of an agonist ligand, which, in turn, can couple to and activate a single Gα protein class (Gi/o, Gq/11, Gs/olf, or G12/13). The G protein is a heterotrimeric protein consisting of a Gα subunit in complex with Gβγ subunits, which initially engages with the active GPCR and then dissociates into constituent Gα and Gβγ components. Depending on the particular G protein and its cellular context, either Gα or Gβγ can mediate downstream effector activity through activation or inhibition of enzymes or channels. The active receptor is desensitized through phosphorylation, which, in turn, mediates β-arrestin recruitment. β-Arrestin can facilitate receptor sequestration or internalization in addition to activating noncanonical signaling systems such as mitogen-activated protein kinase pathways. Canonically, GPCR signaling involves sequential activation of a single type of Gα protein class followed by β-arrestin–dependent activation of noncanonical signal pathways. According to the concept of functional selectivity or ligand bias, signaling intensity can be skewed toward either the G protein pathway or the β-arrestin pathway (1, 2). Biased ligands are potentially desirable therapeutic agents that modulate specific signaling pathways relevant to disease processes (3). However, designing biased ligands requires a thorough understanding of receptor pharmacology and the appropriate cell-based assay systems to characterize the properties of drug candidates.

One disease target for biased ligands is HIV-1 infection, an ongoing epidemic with about 2.1 million new cases worldwide in 2015. A strategy to reduce transmission of HIV-1 is to prevent its cellular entry (4). C-C chemokine receptor 5 (CCR5) is an obligate coreceptor required for cellular viral entry. Cells lacking CCR5 on their surface are not susceptible to most strains of HIV-1 infection (5). Although endogenous chemokine agonists reduce the population of CCR5 on the cell surface by inducing receptor internalization, CCR5 agonists are not used as therapeutic agents for two reasons. First, they cause internalization of only a subset of CCR5, allowing HIV-1 to use receptors remaining on the cell surface for cellular entry (6). Second, stimulation of G proteins by activated CCR5 also promotes targeted migration of immune cells, leading to undesirable inflammatory side effects and enhanced HIV-1 infection at the site of inflammation (7, 8). In principle, a biased ligand that causes maximal receptor internalization, in the absence of G protein activation to avoid generating an inflammatory response, would be an ideal HIV-1 cellular entry blocker (9).

The desire for a biased, therapeutically viable HIV-1 drug that targets CCR5 has motivated the development of several analogs of regulated on activation, normal T cell expressed and secreted (RANTES) with N-terminal modifications (Fig. 1A): PSC-, 6P4-, 5P12-, and 5P14-RANTES (PSC, 6P4, 5P12, and 5P14 hereafter). These RANTES analogs appear to show pronounced signaling bias. Optimization of AOP-RANTES, the first potent, nonsignaling chemokine analog that blocks HIV-1 entry (10), has generated PSC, the first highly potent anti–HIV-1 molecule targeting CCR5 to be developed (11). PSC apparently behaves as a pharmacological superagonist because it appears to be more efficacious than RANTES at eliciting the release of intracellular Ca2+ stores and is more effective at causing internalization of CCR5 (9, 12). Recombinant RANTES analogs including 6P4, 5P12, and 5P14 have high anti–HIV-1 potency. Like PSC, 6P4 behaves as a strong agonist that induces both Ca2+ flux and receptor internalization. In contrast, 5P14 does not cause Ca2+ flux signaling but induces receptor internalization, suggestive of an internalization or β-arrestin–biased ligand. Last, 5P12 binds to CCR5 but does not demonstrate apparent Ca2+ flux or receptor internalization (9).

Fig. 1 CCR5-mediated Gi/o protein activation by endogenous chemokines and RANTES analogs.

(A) The RANTES analogs used in this study have modifications to the N-terminal tail region as shown, where p represents pyroglutamate, α represents N-nonanoyl, β represents l-thioproline, and γ represents l-2-cyclohexylglycine. (B) Human embryonic kidney (HEK) 293T cells cotransfected with CCR5 and the EPAC2 reporter plasmid were stimulated with forskolin and incubated with the chemokines RANTES (red, circle), MIP-1α (black, square), 5P12 (blue, diamond), 5P14 (green, hexagon), 6P4 (yellow, triangle), and PSC (purple, star). (C) Transfected cells were pretreated with PTX (100 ng/ml; squares) or control buffer (circles) for 16 hours, stimulated with forskolin, and incubated with 100 nM chemokines. Data are expressed as a percentage of forskolin-stimulated response and are presented as the mean ± SEM for n ≥ 3 three independent experiments performed in at least technical triplicate. Shown are the statistical significance of the differences between PTX-treated and control condition for each chemokine: ****P < 0.0001 (multiple t tests). Each chemokine tested significantly inhibited cAMP generation (P < 0.001). PSC and 6P4 showed enhanced inhibition of forskolin-induced cAMP generation [P < 0.05, one-way analysis of variance (ANOVA), Dunnett’s multiple comparisons test].

These RANTES analogs show promise in being capable of tuning the CCR5 signaling pathways to avoid the undesired side effect of inflammation. However, the signaling pathways that link CCR5 activation to its numerous physiological functions, including inflammation, remain poorly characterized. In addition to inducing Ca2+ flux, activation of G proteins through CCR5 also leads to inhibition of cyclic adenosine monophosphate (cAMP) generation, an effect of Gi/o protein activation. However, it is less clear whether Ca2+ flux is a result of Gq protein activation or release of Gβγ subunits after Gi/o protein activation in the context of CCR5 (13). 5P14, which does not trigger CCR5-mediated Ca2+ flux, does inhibit cAMP production (14). These results suggest that CCR5-mediated Ca2+ flux and inhibition of cAMP generation are not linked and may be mediated by different G proteins. Some GPCRs can signal through more than one Gα protein class depending on the cellular environment and the activating ligand (15). CCR5 has been suggested to switch from Gi/o to Gq signaling, but the relationship between the well-established Gi/o signaling pathway and other G protein signaling pathways remains to be determined (16).

Here, we tested the ability of endogenous chemokines and RANTES analogs to activate different Gα protein classes in the same human cell line in culture. It was important to use the same cell line because compositional differences of effectors among cell types can make it possible for a ligand to appear to be biased when different pathways are evaluated in different cells, a bias that is not detected if the same cell types are used for all of the assays (17). We measured G protein activation through several downstream pathways, including inhibition of cAMP generation, Ca2+ flux, and inositol 1-monophosphate (IP1) accumulation. In addition, we identified the activated Gα protein subtype that was responsible for these downstream events by performing assays in the presence of transfected individual G proteins or in cells treated with pertussis toxin (PTX) to inactivate Gi/o or an inhibitor (YM-254890) to block activation of Gq (18, 19). We found that each of the chemokines induced Gi/o protein activation, but the chemokines varied in their ability to activate Gq proteins. In comparison with the endogenous chemokines RANTES (also known as CCL5) and macrophage inflammatory protein-1α (MIP-1α) (also known as CCL3), 6P4 and PSC acted as superagonists at Gq, whereas 5P12 showed no Gq activation and was Gi/o biased.

RESULTS

RANTES analogs induce CCR5-mediated inhibition of cAMP production, which depends on Gi/o protein activation

We first measured the ability of the RANTES analogs to induce Gi/o activation, which is the canonical signaling pathway of CCR5. Gi/o activation leads to the inhibition of adenylyl cyclase and a decrease in the amount of cellular cAMP. We tested the ligand-dependent inhibition of forskolin-stimulated cAMP production in cells transfected with CCR5 vector. Each ligand displayed a robust dose-dependent increase in inhibition of cAMP production (Fig. 1B). The EC50 (the concentration of ligand required for half-maximal signal) values for inhibition of forskolin-stimulated cAMP accumulation determined for the ligands were all within 20-fold of each other, whereas the Emax (maximum signal) values were within about a 2-fold range (Table 1). At the highest concentration of ligand used (100 nM), we found that the range of Emax values was 14 to 31% inhibition of the forskolin response. The rank order of maximum signal was 5P14 > RANTES > MIP-1α = 6P4 = 5P12 > PSC. The EC50 values of ligands ranged from 0.1 to 2.1 nM with the following rank order: 6P4 > 5P14 > RANTES = PSC > MIP-1α = 5P12 (Table 1). Signaling occurred specifically through CCR5, because empty vector–transfected cells did not show inhibition of cAMP generation in response to any of the chemokines tested (fig. S1). As a control for the effect of changes in CCR5 amount, experiments were also performed 48 hours after transfection (fig. S2), in addition to experiment performed 24 hours after transfection (cf. Fig. 1B).

Table 1 Summary of fitted curve parameters for inhibition of cAMP accumulation.

EC50, pEC50, and Emax values with SEM are given for Fig. 1B.

View this table:

To confirm that inhibition of cAMP accumulation was due to Gi/o protein activation, we measured cAMP accumulation in cells pretreated with PTX to prevent Gi/o proteins from interacting with receptors. Similar to the results from the dose-response experiment, each chemokine demonstrated statistically significant CCR5-mediated inhibition of cAMP accumulation in forskolin-treated cells. As expected, PTX pretreatment abrogated the ability of all ligands to inhibit forskolin-stimulated cAMP accumulation (Fig. 1C). 6P4 and PSC also induced significant enhancement of cAMP accumulation above forskolin-stimulated cells after treatment of PTX, as seen as negative percentage values of forskolin inhibition. However, 6P4 and PSC did not induce cAMP accumulation after treatment of PTX when forskolin was not added (fig. S3). We also present raw bioluminescence resonance energy transfer 2 (BRET2) values of the complete time course experiments (fig. S4, A to D).

RANTES analogs have different effects on CCR5-dependent Ca2+ flux

In addition to inhibiting cAMP accumulation, CCR5 activation also leads to Ca2+ flux. In contrast to inhibiting cAMP accumulation, not all of the ligands stimulated Ca2+ flux through CCR5 (Fig. 2A). The RANTES analogs induced varying amounts of Ca2+ flux with the rank order being 6P4 = PSC > RANTES > MIP-1α > 5P14 > 5P12. Signaling occurred specifically through CCR5, because empty vector–transfected cells did not show Ca2+ fluxes in response to any of the chemokines tested (fig. S5). Thus, for Ca2+ flux, 6P4 and PSC are superagonists that are more efficacious than the endogenous CCR5 chemokines RANTES and MIP-1α, whereas 5P14 acts as a weak partial agonist and 5P12 does not have agonist activity.

Fig. 2 The effects of G protein–subtype cotransfection on CCR5-mediated Ca2+ flux and IP1 accumulation in response to endogenous chemokines and RANTES analogs.

Ca2+ flux in HEK293T cells transfected with (A) CCR5 alone or with G protein α subunits (B) Gi2, (C) Gq, or (D) Gqi5 was measured in response to 100 nM chemokine. Fluorescence data (maximal minus basal) are presented as mean relative fluorescence units (RFU) ± SEM for n = 4 independent experiments performed in technical duplicate. IP1 accumulation in HEK293T cells transfected with (E) CCR5 alone or with G protein α subunits (F) Gi2, (G) Gq, or (H) Gqi5 was measured in response to 100 nM chemokine. Data are expressed as picomoles of IP1 formed per well and are presented as the mean ± SEM for n = 4 independent experiments performed in technical duplicate. Shown are the statistical significance of the differences between chemokine treatment and buffer control and between chemokine treatment and RANTES treatment: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (two-way ANOVA, Tukey’s multiple comparisons test).

Ca2+ flux mediated by CCR5 is inhibited by Gi2 cotransfection and increased by Gq or Gqi5 cotransfection

Because of the differences in signaling profiles of the RANTES analogs between Ca2+ flux and inhibition of cAMP accumulation, we next investigated the G protein subtype that was responsible for eliciting Ca2+ flux through CCR5 activation. Classically, Ca2+ flux is a result of Gq protein activation, but chemokine receptors can also induce Ca2+ flux through the release of Gβγ subunits after Gi/o protein activation (20, 21). To investigate the individual contributions of each G protein subtype to the Ca2+ flux signal, we repeated the Ca2+ flux experiments in cells in which CCR5 was cotransfected with Gq, Gi2, or Gqi5 protein. Gqi5 is an engineered Gq protein containing the last five amino acid residues of Gi2, which allows Gi/o-coupled GPCRs to signal through Gq downstream signaling pathways (22). Cotransfection of CCR5 with the three G protein subunits did not affect the amount of CCR5 on the cell surface (fig. S6). Also, Ca2+ flux in response to adenosine 5′-triphosphate (ATP) in HEK293T cells was not affected by cotransfection of G protein subunits (fig. S7). ATP was used as a positive control to activate endogenous purinergic receptors in HEK293T cells.

Cotransfection of Gi2 with CCR5 markedly inhibited the ability of RANTES and MIP-1α to induce Ca2+ flux through CCR5 (Fig. 2B). Gi2 cotransfection also dampened the maximum Ca2+ flux signal elicited in response to PSC. In contrast, cotransfection of CCR5 and Gq enhanced the maximal Ca2+ flux signal for all ligands, except 5P12 (Fig. 2C). In cells cotransfected with CCR5 and a promiscuous G protein construct, Gqi5, each RANTES analog induced robust Ca2+ flux (Fig. 2D), confirming the ability of all the RANTES analogs to activate Gi/o through CCR5. The data were replotted to highlight differences in signal caused by cotransfection of CCR5 with Gq, Gi2, or Gqi5 (fig. S8A).

Because of differences in the binding kinetics of ligands, measurements of ligand bias are sensitive to the experimental time scale (23, 24). For example, Ca2+ flux assays measure a transient signal that is complete in 2 min, whereas cAMP accumulation is measured over durations of 15 min or more. To accurately compare the ability of each chemokine to activate CCR5-mediated Ca2+ flux and inhibition of cAMP accumulation, it was necessary to measure Ca2+ flux with an accumulation assay. Inositol 1,4,5-trisphosphate (IP3) is a product upstream of Ca2+ flux, and a degradation product of IP3 can be measured by an accumulation assay. Although IP3 is rapidly degraded, its degradation product IP1 is stabilized by the addition of lithium chloride, allowing it to accumulate in the cell after receptor stimulation. IP1 accumulation above that induced by buffer was not seen in vector-transfected negative control cells treated with any ligand (fig. S9). In CCR5-expressing cells, we observed IP1 accumulation in response to only PSC (Fig. 2E), demonstrating that PSC functions as a superagonist as measured by IP1 accumulation and Ca2+ flux.

Cotransfection of Gi2 with CCR5 reduced IP1 accumulation in response to PSC (Fig. 2F). In contrast, Gq cotransfection with CCR5 enhanced IP1 accumulation in response to each of the chemokines, except 5P12 (Fig. 2G). Moreover, Gq cotransfection with CCR5 revealed agonist activity as measured by IP1 accumulation in response to 6P4, RANTES, MIP-1α, and 5P14, an effect not seen in cells transfected with only CCR5. In addition, the rank order of maximal IP1 accumulation signal among the RANTES analogs was similar to that shown for Ca2+ flux, with 6P4 = PSC > RANTES > 5P14 > 5P12. Moreover, each of the RANTES analogs showed robust IP1 accumulation in cells cotransfected with Gqi5, confirming that each ligand activates Gi/o (Fig. 2H). The data were replotted to highlight differences in signal caused by cotransfection of CCR5 with Gq, Gi2, or Gqi5 (fig. S8B).

Together, our IP1 accumulation and Ca2+ flux data show that PSC and 6P4 are superagonists, whereas 5P12 lacks agonist activity in these assays. Furthermore, Gi2 cotransfection disrupts the ability of both endogenous CCR5 chemokines and RANTES analogs to induce Ca2+ flux and IP1 accumulation, whereas Gq cotransfection has the opposite effect.

CCR5-mediated Ca2+ flux is reduced by inhibiting Gi/o protein activation and abolished by inhibiting Gq protein activation

To further differentiate between the contribution of Gi/o and Gq proteins to CCR5-mediated signaling, we treated cells with G protein subtype–specific inhibitors. PTX prevents Gi/o protein coupling to GPCRs by catalyzing the adenosine 5′-diphosphate ribosylation of a Cys residue in the C-terminal tail of Gi/o α subunits (25). YM-254890 prevents Gq activation by acting as a guanine-nucleotide dissociation inhibitor of Gq (26). Both PTX and YM-254890 decreased CCR5-mediated Ca2+ flux (Fig. 3, A to D). However, YM-254890 abolished CCR5-mediated Ca2+ flux stimulated by all chemokines, whereas the effects of PTX were less pronounced. In CCR5-expressing cells, PTX reduced Ca2+ flux induced by the superagonists 6P4 and PSC and completely inhibited that induced by RANTES and MIP-1α (Fig. 3A). In cells cotransfected with CCR5 and Gi2, both PTX and YM-254890 decreased the remaining signal from PSC and 6P4 (Fig. 3B). However, in cells cotransfected with CCR5 and Gq, which would be expected to increase CCR5 signaling, RANTES, MIP-1α, and 5P14 triggered a partially PTX-insensitive Ca2+ flux response (Fig. 3C). 6P4 and PSC also showed more PTX insensitivity in cells cotransfected with Gq and CCR5 than in cells expressing CCR5 alone. Last, the signal elicited by chemokines in cells cotransfected with CCR5 and the Gqi5 chimera was reduced as a result of PTX treatment, except for that triggered by 5P12 (Fig. 3D). Overall, in contrast to PTX, disruption of Gq activation by YM-254890 abolished Ca2+ signaling by each ligand even when Gq or Gqi5 was cotransfected.

Fig. 3 The effects of PTX and YM-254890 treatment on CCR5-mediated Ca2+ flux in response to endogenous chemokines and RANTES analogs.

Ca2+ flux in HEK293T cells transfected with (A) CCR5 alone or with G protein α subunits (B) Gi2, (C) Gq, or (D) Gqi5 was measured in response to 100 nM chemokine. Cells were pretreated with buffer (white bars), PTX (100 ng/ml; gray bars), or 1 μM YM-254890 (black bars). Fluorescence data (maximal minus basal) are presented as mean RFU ± SEM for n = 4 independent experiments performed in technical duplicate. Shown are the statistical significance of the differences between buffer control, PTX, and YM-254890: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (two-way ANOVA, Tukey’s multiple comparisons test).

CCR5-mediated IP1 accumulation is reduced by inhibiting Gi/o protein activation and abolished by inhibiting Gq

Consistent with the Ca2+ flux experiments, YM-254890 prevented all of the chemokines from causing IP1 accumulation, regardless of the Gα protein subunit cotransfected (Fig. 4, A to D). In cells transfected with CCR5 alone, the PSC-induced IP1 signal was attenuated by PTX (Fig. 4A). PTX treatment of cells cotransfected with CCR5 and Gq only minimally reduced the IP1 signal induced by RANTES, MIP-1α, 6P4, and PSC, but had a greater effect on that induced by 5P14 (Fig. 4C). As with Ca2+ flux, PTX did not strongly reduce the signal elicited by the RANTES analogs when CCR5 was cotransfected with Gqi5 (Fig. 4D). In summary, although PTX caused only a modest decrease in chemokine responses, YM-254890 abolished the ability of each of the ligands tested to elicit IP1 accumulation and Ca2+ flux. Thus, both cotransfection and inhibition experiments suggest that Gq is the dominant G protein in stimulating Ca2+ flux after CCR5 activation, with Gi/o playing a much smaller role.

Fig. 4 The effects of PTX and YM-254890 treatment on CCR5-mediated IP1 accumulation in response to endogenous chemokines and RANTES analogs.

IP1 accumulation in HEK293T cells transfected with (A) CCR5 alone or with G protein α subunits (B) Gi2, (C) Gq, or (D) Gqi5 was measured in response to 100 nM chemokine. Cells were pretreated with buffer (white bars), PTX (100 ng/ml; gray bars), or 1 μM YM-254890 (black bars). Data are expressed as picomoles of IP1 formed per well and are presented as the mean ± SEM for n = 4 independent experiments performed in technical duplicate. Shown are the statistical significances of the differences between buffer control, PTX, and YM-254890: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 (two-way ANOVA, Tukey’s multiple comparisons test).

Cotransfection of CCR5 with Gi/o, Gq, or Gqi5 causes different effects on Emax values and EC50 values

The effects of G protein cotransfection with CCR5 could be a result of a change in ligand Emax or EC50. Performing and fitting dose-response curves for Ca2+ flux (Table 2) showed that in cells transfected with CCR5, EC50 values ranged from 21.0 to 240 nM with the rank order of PSC = 6P4 = RANTES = 5P14 > MIP-1α (Fig. 5A). Emax values ranged from 2700 to 45,000 RFU, and the rank order was 6P4 > PSC > RANTES = MIP-1α > 5P14. Similar results were obtained 48 hours after transfection (fig. S10). Cotransfection of CCR5 and Gi2 decreased Ca2+ flux such that only PSC and 6P4 could elicit any Ca2+ flux response (Fig. 5B), which was due to reduced Emax values because EC50 values in cells transfected with CCR5 were similar to those in cells cotransfected with both CCR5 and Gi2. Similarly, the increase in response from each chemokine in cells cotransfected with CCR5 and Gq was not a result of altered EC50 values, but rather due to an increase in Emax values (Fig. 5C). In contrast to Gi2 and Gq, cotransfection of Gqi5 and CCR5 altered both EC50 and Emax values in dose-response Ca2+ flux experiments (Fig. 5D). The range of EC50 values among the ligands tested decreased, but the rank order was similar, with 6P4 = RANTES > PSC > 5P14 = MIP-1α > 5P12. In addition, the Emax value of each ligand increased, but the rank order also remained the same.

Table 2 Summary of fitted curve parameters for Ca2+ flux.

EC50, pEC50, and Emax values with SEM are given for Fig. 5, left column. ND, not detectable.

View this table:
Fig. 5 Dose-response curves for CCR5-mediated Ca2+ flux and IP1 accumulation with or without cotransfection of different G protein α subunits.

Ca2+ flux in HEK293T cells transfected with (A) CCR5 alone or with G protein α subunits (B) Gi2, (C) Gq, or (D) Gqi5 was measured in response to RANTES (red, circle), MIP-1α (black, square), 5P12 (blue, diamond), 5P14 (green, hexagon), 6P4 (yellow, triangle), and PSC (purple, star). Fluorescence data (maximal minus basal) are presented as mean RFU ± SEM for n = 4 independent experiments performed in technical triplicate. IP1 accumulation in HEK293T cells transfected with (E) CCR5 alone or with G protein α subunits (F) Gi2, (G) Gq, or (H) Gqi5 was measured in response to RANTES (red, circle), MIP-1α (black, square), 5P12 (blue, diamond), 5P14 (green, hexagon), 6P4 (yellow, triangle), and PSC (purple, star). Data are expressed as picomoles of IP1 formed per well and are presented as the mean ± SEM for n = 3 independent experiments performed in technical triplicate.

In cells transfected with CCR5 alone, only PSC showed any dose-dependent IP1 accumulation, with an EC50 of 190 nM (Fig. 5E) (Table 3). No agonist-stimulated IP1 accumulation was observed in cells cotransfected with CCR5 and Gi2 (Fig. 5F). Cells cotransfected with CCR5 and Gq displayed a range of EC50 values for IP1 accumulation of 4.3 to 72 nM with a rank order of 5P14 > 6P4 > RANTES = PSC. These EC50 values were similar to those for Ca2+ flux (Fig. 5, C and G). The Emax values ranged from 1.0 to 8.4 pmol per well with a rank order of PSC > 6P4 > 5P14 > RANTES, although the curve fits for 5P12 and MIP-1α were somewhat ambiguous. This rank order of Emax values was similar to that observed with Ca2+ flux in CCR5-transfected cells both with and without cotransfection of Gq. As seen previously, cotransfection of Gqi5 and CCR5 allowed each RANTES analog to cause IP1 accumulation (Fig. 5H). This increase in signal caused by Gqi5 cotransfection with CCR5 could be attributed to a decrease in EC50 values for all ligands tested. In addition, an increase in Emax contributed to the higher signals for RANTES, MIP-1α, 5P12, and 5P14 in cells cotransfected with Gqi5 and CCR5.

Table 3 Summary of fitted curve parameters for IP1 accumulation.

EC50, pEC50, and Emax values are given for Fig. 5, right column. ND, not detectable.

View this table:

In summary, Emax values were altered when Gi/o or Gq was cotransfected with CCR5, whereas both Emax and EC50 values were altered when Gqi5 was cotransfected with CCR5. These results show that when CCR5 is cotransfected with either Gi2 or Gq, the Emax values, but not the EC50 values, for agonist-dependent IP1 accumulation and Ca2+ flux were altered. In contrast, cotransfection of CCR5 and Gqi5 caused a decrease in EC50 values for IP1 accumulation and Ca2+ flux for all ligands tested. In addition, coexpression of CCR5 and Gqi5 led to an increase in Emax values for Ca2+ flux for all ligands and an increase in Emax values for IP1 accumulation for most ligands tested.

DISCUSSION

Several GPCR-targeted ligands have been described to cause signaling bias toward β-arrestin or G protein signaling pathways. However, reports of ligands that bias GPCR signaling toward a specific Gα protein class are rare, in part, because of the difficulty in interpreting signaling assays in which cross-talk between different G protein signaling pathways occurs. In this study, we determined the specific Gα protein class that was responsible for CCR5-mediated inhibition of cAMP accumulation, Ca2+ flux, and IP3 production. Analogs of the endogenous chemokine RANTES with N-terminal modifications showed similar abilities to activate CCR5-mediated Gi/o signaling, but either enhanced or decreased abilities to activate CCR5-mediated Gq signaling.

For inhibition of cAMP accumulation, each of the endogenous chemokines and the RANTES analogs tested acted as agonists with similar Emax values and EC50 values of 2.1 nM or lower. EC50 values of at least one order of magnitude higher were obtained earlier from CCR5-expressing Chinese hamster ovary (CHO-CCR5) cells, which also showed a lack of detectable signal for 5P12 (14). In addition to the use of different cells, the source of variation between these results could be due to the use of different ligand incubation times or different assays to detect cAMP [real-time changes of cAMP in live cells compared with measurement of cAMP after a single incubation time point (14)].

The RANTES analogs showed different abilities to induce Ca2+ flux and IP1 accumulation through CCR5. In comparison to the endogenous chemokines RANTES and MIP-1α, the RANTES analogs 6P4 and PSC acted as superagonists, whereas 5P12 did not have agonist activity in the Ca2+ flux assays. Our measurements of EC50 and Emax values recapitulated the results from Ca2+ flux assays performed in CCR5-transfected Hela-P5L and phytohemagglutinin- and interleukin-2–activated T blasts (9). Cells expressing only CCR5 demonstrated IP1 accumulation in response only to PSC, whereas CCR5-mediated Ca2+ flux was measured in response to RANTES, MIP-1α, 6P4, and PSC. The differences between results when assessing Ca2+ flux and IP1 accumulation are likely due to assay sensitivity (27). The ability of the RANTES analogs to induce CCR5-mediated Gq signaling appears to correlate with their ability to induce β-arrestin recruitment and internalization. Analogous to the Ca2+ flux results, only 5P12 was unable to induce CCR5 internalization (28). In addition, cells treated with PSC demonstrated the strongest recruitment of β-arrestin to CCR5, followed by RANTES, and then 5P14. 5P12 did not cause β-arrestin recruitment to CCR5 (14, 29, 30).

The differences in the signaling profiles of the RANTES analogs suggest that different effectors are responsible for CCR5-mediated inhibition of cAMP accumulation or induction of Ca2+ flux. Gi/o protein activation inhibits cAMP accumulation, a pathway further confirmed by our results. 6P4 and PSC can stimulate cAMP production above levels induced by forskolin in the presence of PTX, suggesting that these ligands might also stimulate CCR5-mediated Gs protein activation. However, 6P4 and PSC did not demonstrate CCR5-mediated cAMP accumulation in cells treated with PTX, but not with forskolin. The differential regulation of adenylated cyclase subtypes in HEK293 cells could also be the source of this PTX-dependent cAMP production by 6P4 and PSC. Ca2+ can activate certain subtypes of adenylate cyclase, causing cross-talk between Ca2+ signaling and cAMP production (31). Alternatively, 6P4 and PSC in the presence of PTX could inhibit the degradation of forskolin-dependent cAMP, and these ligands would not induce cAMP production in the presence of PTX, but the absence of forskolin.

The Gα protein subunit that is responsible for Ca2+ flux and IP3 formation is less clear, but could be Gq or Gi/o. Cotransfection of Gq augments CCR5-mediated Ca2+ flux and IP1 accumulation, responses that are inhibited by cotransfection of Gi2, which supports the hypothesis that Gq is responsible for Ca2+ flux. Gi2 transfection could inhibit Ca2+ flux and IP1 accumulation through CCR5 either by preventing Gq recruitment by competitively binding to CCR5 or by acting as a Gβγ subunit scavenger, which would prevent the formation of heterotrimeric Gq proteins required for G protein activity.

We further differentiated the contributions of Gi/o and Gq proteins to CCR5-mediated Ca2+ flux and IP3 signaling using inhibitors. The Gq inhibitor YM-254890 prevented all of the RANTES analogs from inducing CCR5-mediated Ca2+ flux and IP1 accumulation. However, the ability of PTX to at least partially inhibit Ca2+ flux and IP1 accumulation does not support the hypothesis that Gq mediates these signaling pathways through CCR5. This seemingly paradoxical result can be explained by the indirect effect of PTX decreasing binding of chemokines to CCR5 (6). Reduction of agonist affinity is likely due to a reduction of Gi/o protein precoupling to CCR5. Structural studies of β2 adrenergic receptor interacting with G protein has confirmed the paradigm that a receptor must be bound to a transducer to stabilize the agonist-receptor interaction (32). Likewise, the conformation of CCR5 molecules on the cellular surface is altered upon PTX-induced abrogation of Gi/o precoupling (33). Similar to our results, other reports have also described incomplete inhibition of CCR5-mediated Ca2+ flux by PTX, suggesting that Gi/o plays a role in agonist binding but not in Ca2+ flux (3436). Last, it is unlikely that Gβγ subunits could induce Ca2+ flux in HEK293T cells, because the closely related HEK293 cell line does not produce the phospholipase C subunit that links Gβγ release to Ca2+ flux (37). Although we cannot rule out a contribution from Gβγ, our data suggest that Ca2+ flux signaling through CCR5 in this cellular system is mainly, if not completely, mediated by Gq protein activation.

Chemokine receptors are historically believed to couple to and signal through Gi/o proteins exclusively. The Ca2+ flux response elicited by many chemokine receptors has been thought to be a result of Gβγ subunit release after activation of Gi/o (20). However, we provide evidence that CCR5 signals through both Gi/o and Gq/11 proteins. We also demonstrate that IP1 accumulation and Ca2+ flux are products of Gq protein signaling in HEK293T cells. CCR5 mediates PTX-insensitive signaling responses, suggesting that CCR5 couples to and signals through other G proteins (16, 34, 35). CCR5 couples to and coimmunoprecipitates with Gq (38, 39). Both Gq and Gi/o protein signaling may contribute to the activities of other chemokine receptors. CXCR4 is proposed to switch from Gi/o signaling to Gq signaling, leading to reduced migration and activation of the T cells (16, 40). In addition, CXCR4- and CCR7-dependent chemotaxis of dendritic cells requires activation of both Gi/o and Gq pathways (41). Theoretically, the capacity of a receptor to couple to certain Gα protein subunits should be apparent from its primary structure. However, although residues in similar positions in different GPCRs recognize G proteins, other distinct residues make structural binding contacts. Therefore, without the availability of comparable structures of Gi/o and Gq bound to activated GPCRs, it is difficult to surmise whether Gq coupling to other chemokine receptors can be predicted or validated by structural information (42).

The capacity of each chemokine to bind to different conformations of CCR5 likely accounts for apparent differences to induce CCR5-mediated signaling in either Gi/o- or Gq-dependent pathways. The “spare receptor” concept posits that there are at least two pools of inactive CCR5: receptors precoupled to Gi/o proteins and “naked” receptors not precoupled to Gi/o protein (6, 43). CCR5 can assume multiple conformations, as shown by monoclonal antibody detection of different populations of CCR5 (44). The endogenous chemokines RANTES and MIP-1α bind with high affinity to only Gi/o protein–coupled CCR5, whereas PSC has a high affinity for both Gi/o protein–coupled CCR5 and naked CCR5 (6). Single-molecule binding experiments have also shown that 5P12, 5P14, PSC, and 6P4 bind to CCR5 with affinities of 374 ± 24 pM, 494 ± 38 pM, 10.1 ± 0.4 pM, and 6.6 ± 0.3 pM, respectively, in detergent solution and in absence of G protein. However, MIP-1α and RANTES do not show detectable binding to naked receptor up to concentrations of 10 μM under these conditions (45). Thus, PSC can activate a subset of CCR5 molecules that remains inactive after stimulation with RANTES (43). Therefore, the increased Ca2+ flux in response to PSC or 6P4 is likely due to the activation of naked CCR5, in addition to Gi/o protein–precoupled CCR5. In line with this hypothesis, PSC and 6P4 do not have substantially lower EC50 values than RANTES, but do have higher Emax values, suggesting that these chemokines are superagonists because they bind to more receptors, rather than because they bind with higher affinity.

We present a model to contextualize the results of our experiments and hypothesize about how RANTES analogs are capable of showing G protein subtype–specific signaling bias (Fig. 6A). For RANTES and MIP-1α, Gi/o subunit precoupling promotes binding of the ligand to the receptor, upon which Gi/o is activated and dissociates from the receptor. A ligand that remains bound to the receptor can recruit and activate Gq or additional Gi/o proteins. In contrast to RANTES and MIP-1α, PSC and 6P4 can bind to and activate CCR5-dependent signaling pathways, independently of whether the receptor is precoupled to Gi/o or naked. By signaling through two pools of the receptor, PSC and 6P4 can produce more signals and thus act as superagonists. Although not apparent from our data, 5P12 binds to CCR5 in a G protein–independent fashion (6), and thus, binding of 5P12 to naked CCR5 preferentially activates Gi/o (Fig. 6B).

Fig. 6 Conceptual model depicting CCR5-mediated signaling through Gi/o and Gq by endogenous chemokines and RANTES analogs.

The CCR5 receptor exists in at least two populations, with a group of CCR5 molecules precoupled to Gi/o (gray background) and another group in an uncoupled or naked state (blue background). (A) RANTES and MIP-1α bind only to Gi/o-precoupled CCR5, which then leads to activation of Gi/o. After activation of Gi/o, CCR5 can then recruit and activate Gq. 6P4 and PSC bind to both Gi/o-precoupled CCR5 and naked CCR5. Thus, activation of Gq protein occurs independently of Gi/o proteins. (B) 5P12 binds to both Gi/o-precoupled CCR5 and naked CCR5, but binding leads to activation of Gi/o only.

This study demonstrates a type of ligand bias that is G protein subtype specific. Another peptide receptor, pituitary adenylate cyclase activating polypeptide (PACAP) receptor, responds to the two ligands PACAP-38 and PACAP-27 with similar EC50 values for the Gs signaling pathway. However, the two PACAP receptor ligands have EC50 values that are 100-fold different for a Gq signaling pathway (46). The free fatty acid receptors (FFARs) also signal through multiple G protein subtypes, but depending on the activating ligand, the FFARs can selectively signal through a single G protein subtype. For example, synthetic agonists targeting FFAR1 can signal through both Gq and Gs, whereas the endogenous agonists signal only through Gq (47). In addition, an allosteric modulator targeting FFAR2 selectively activates Gi/o signaling but not Gq signaling (48). In our study, we described chemokines with similar capacity to evoke CCR5-mediated Gi/o signaling, but with both diminished and enhanced Gq signaling.

How the modification of the N-terminal region of RANTES leads to such pronounced changes in CCR5-mediated Gq signaling, but not Gi/o signaling, is unclear. Nonetheless, some insight can be garnered from crystal structures of CCR5, including CCR5 in complex with 5P7, another RANTES analog (49, 50). 5P7 differs from 5P12 by one amino acid, Thr instead of Leu at position 7, and lacks agonist activity for Ca2+ flux (9). When comparing the crystal structure of 5P7 to a model of CCR5 with RANTES, several different interactions are apparent that could be key for Gq protein activation. Whereas RANTES has polar residues in positions 4 to 7 that can engage a minor pocket polar network, 5P7 has bulky hydrophobic residues at these positions. Similarly, the Gq-inactive RANTES analogs 5P12 and 5P14 have bulky hydrophobic residues at positions 4 to 7, whereas the superagonist PSC has polar residues. However, the superagonist 6P4 has mostly less-bulky hydrophobic amino acids at these positions. A crystal structure of CCR5 engaged with a superagonist will be necessary to understand precisely how Gq signaling is modulated by changes in the N-terminal sequences of RANTES.

The RANTES analogs 5P12 and 5P14, which fail to activate Gq, may be safer drugs because they discriminate between Gi/o and Gq protein activation. In bypassing Gq signaling, several physiological effects of CCR5 activation that are PTX resistant or downstream of Ca2+ flux would be avoided, including CCR5 recruitment to the immunological synapse during T cell stimulation (16). In addition, an in vitro predictor of an inflammatory response, chemotaxis, requires additional GPCR signaling because Gi/o signaling is necessary, but not sufficient, to induce chemotaxis (21). Because Ca2+ flux is necessary for actin rearrangement, an important step in chemotaxis, it is plausible that this additional signaling component is Gq activation (51). Although the RANTES chemokine analogs have not been compared in their ability to induce chemotaxis, PSC does cause chemotaxis of CD4 T cells (6). The contribution of these two signaling arms to the physiological effects of CCR5 activation remains unknown. Thus, these RANTES analogs could serve as tools to determine the effects of Gi/o signaling alone compared with Gq and Gi/o protein signaling in combination. Future studies should use these chemokine analogs to determine how CCR5-mediated Gi/o and Gq signaling affect events such as chemotaxis and T cell activation using physiologically relevant cell types.

In conclusion, we investigated G protein subtype–specific biased signaling of the chemokine receptor CCR5 using a series of chemokine analogs developed as HIV-1 microbicides. Using a specific inhibitor, we found that Gq/11 is responsible for Ca2+ flux downstream of CCR5 in HEK293T cells, which earlier had been attributed to Gβγ subunit released from Gi/o. RANTES analogs had a wide range of efficacy for CCR5-mediated Gq signaling, but similar Gi/o signaling efficacy, opening the possibility for selectively tuning chemokine signaling. Noncanonical Gq/11 signaling bias in the chemokine receptor network to our knowledge has not been previously reported. Our findings also provide an experimental strategy in which a Gq-specific inhibitor is used to study Gq signaling bias relevant to Ca2+ flux signals from GPCRs thought to couple to Gi/o.

MATERIALS AND METHODS

Materials

Coelenterazine 400A for BRET2 experiments was from Biotium. The IP-One HTRF assay kit was from Cisbio. Forskolin, PTX, and poly-d-lysine were from Sigma-Aldrich. YM-254890 was from Wako Pure Chemical Industries. Dulbecco’s modified Eagle’s medium Glutamax (DMEM-Q), Hanks’ balanced salt solution (HBSS), and Lipofectamine 2000 were from Life Technologies. Fetal bovine serum (FBS) was from Atlanta Biologicals. Bovine serum albumin (BSA) fraction V, fatty acid free was from Roche, and 96-well white, clear bottom microplates and 384-well black, clear bottom microplates were from Corning. RANTES and MIP-1α were from PeproTech. The RANTES analogs 5P12, 5P14, 6P4, and PSC were a gift from O. Hartley (Université de Genève).

Transfection constructs

The human CCR5 complementary DNA (cDNA) in pcDNA3.1(+) encoded a C-terminal 1D4 epitope tag (TETSQVAPA). The exchange protein directly activated by cAMP 2 (EPAC2) reporter was a gift from M. Bouvier (Université de Montréal). Gq and Gi2 α subunit plasmids are wild-type human cDNA cloned in pcDNA3.1(+) (cDNA Resource Center). Gqi5 refers to a construct in which the α subunit of Gq is modified so that its C-terminal tail amino acid residues are derived from the α subunit of Gi2 (such that EYNLV-COOH in Gq becomes DCGLF-COOH in Gqi5). Gqi5 was cloned into pcDNA1 (Addgene) (22).

Cell culture and transfection

HEK293T cells were maintained in DMEM-Q with 10% FBS. Transient transfections were performed using Lipofectamine 2000 according to the manufacturer’s instructions with some modifications as previously described (52). Cells were transfected in different plate formats with the following amounts of total DNA: 2 μg per well in 6-well plates, 100 ng per well in 96-well plates, and 20 ng per well in 384-well plates. Total transfected plasmid DNA was kept constant by adding empty vector pcDNA3.1(+) when necessary.

Adenylyl cyclase activity assay

Inhibition of forskolin-stimulated cAMP production in response to each chemokine was monitored in HEK293T cells cotransfected with CCR5 and an EPAC reporter protein that shows a decrease in BRET upon cAMP binding (53). For each well of a 96-well plate, HEK293T cells (100,000 in 0.1 ml of DMEM) were transfected with 12 ng of RLuc3-EPAC-GFP (green fluorescent protein), a BRET2 cAMP sensor, with or without cotransfection of 24 ng of CCR5 wild type. Cells were then plated at 100 μl per well into 96-well white, clear bottom microplates previously coated with 0.01% (w/v) poly-d-lysine. Twenty-four hours after transfection, medium was replaced with BRET buffer [phosphate-buffered saline (PBS) containing 0.5 mM MgCl2 and 0.1% BSA]. Coelenterazine 400A was added to a final concentration of 5 μM followed by a 5-min incubation at room temperature. Cells were then stimulated with chemokine in the presence or absence of 5 μM forskolin at room temperature for 5 min before BRET2 reading to yield dose-response curves. For kinetic studies, HEK293T cells were transfected as previously described with 15 mM Hepes added to the culture media. Coelenterazine 400A was added in a single column of the 96-well plate, and after 5 min, wells were treated with chemokine in the presence of 5 μM forskolin or 200 μM 3-isobutyl-1-methylxanthine to assess the inhibition of forskolin-stimulated cAMP production or cAMP production, respectively. Then, the plate was immediately read repetitively for 12 min. Cells were treated and wells were read one single column of the 96-well plate at a time, producing one replicate at a time, and repeated six times per plate to produce six replicates per experiment. Luminescence and fluorescence readings were collected using the Synergy NEO2 plate reader (BioTek Instruments) and Gen5 software. BRET2 readings between RLuc3 and GFP10 were collected by sequential integration of the signals detected in the 365- to 435-nm (RLuc3) and 505- to 525-nm (GFP10) windows. BRET2 ratios were calculated as previously described (52) and are shown as a percentage of the forskolin-stimulated response. For experiments involving PTX, cells were treated with PTX (100 ng/ml) for 16 hours at 37°C and 5% CO2 before stimulation with forskolin and chemokines. All experimental manipulations were performed while cells remained attached to the 96-well plates.

Ca2+ flux assay

Ca2+ mobilization was monitored by loading CCR5-transfected HEK293T cells with a Ca2+-sensitive fluorescent dye and measuring changes in cytosolic Ca2+ concentrations immediately after chemokine addition. For each well of a 384-well plate, HEK293T cells (20,000 in 0.02 ml of DMEM) were transfected with 10 ng of CCR5 or 20 ng of empty vector. Where applicable, 5 ng of Gq, Gi2, or Gqi5 vector constructs was cotransfected. Cells were then plated at 20 μl per well into 384-well black, clear bottom microplates previously coated with 0.01% (w/v) poly-d-lysine hydrobromide. Twenty-four hours after transfection, FLIPR Calcium 6 dye (20 μl per well; Molecular Devices) was added to the cells and incubated for 1.5 hours at 37°C with 5% CO2. The dye was dissolved in HBSS-H [HBSS with 20 mM Hepes (pH 7.4)] with 0.4% BSA. Before measurement, the plate was incubated at 37°C for an additional 30 min in a prewarmed FlexStation II 384 Plate Reader (Molecular Devices). Chemokine at a 5× final concentration was diluted in HBSS-H with 0.2% BSA. Fluorescence readings were collected using the FlexStation plate reader with excitation at 485 nm, emission at 535 nm, and the dichroic mirror at 525 nm. The FlexStation took measurements over a 100-s time course, with 10 μl of ligand added to the cells 20 s after the start of measurement. RFU are reported as the peak magnitude signal minus the basal signal in each well. For experiments involving PTX, cells were treated with PTX (100 ng/ml) for 16 hours at 37°C and 5% CO2 before chemokine stimulation. For experiments involving YM-254890, 10 μM YM-254890 [dissolved as a 10 mM stock in dimethyl sulfoxide (DMSO)] or equivalent DMSO vehicle was added to the cells and incubated for 30 min at 37°C.

IP1 accumulation assay

IP1 concentrations in CCR5-transfected cells were measured using a competitive homogeneous time-resolved fluorescence assay after incubation with each chemokine for 3 hours. For each well of a low-volume, 384-well plate, HEK293T cells (5000 in 7 μl of DMEM) were transfected with 11 ng of CCR5 vector or 11 ng of empty vector. Where applicable, 5.5 ng of Gq, Gi2, or Gqi5 vector constructs was cotransfected. Twenty-four hours after transfection, cells were stimulated by chemokine diluted in 7 μl of 1× stimulation buffer [10 mM Hepes, 1 mM CaCl2, 0.5 mM MgCl2, 4.2 mM KCl, 146 mM NaCl, 5.5 mM glucose, and 50 mM LiCl (pH 7.4)] with 0.2% BSA and 50 mM LiCl (to prevent IP1 degradation). The chemokine was incubated at 37°C for 3 hours. After incubation, cells were lysed by the addition of d2-fluorophore–labeled IP1 analog (3 μl per well) as the fluorescence acceptor and the terbium cryptate–labeled anti-IP1 monoclonal antibody as the fluorescence donor. Both fluorescent donor and acceptor were diluted in the kit-supplied lysis buffer. The plates were incubated overnight at 4°C, and time-resolved fluorescence signals were read using the BioTek Synergy NEO plate reader (BioTek Instruments, Winooski, VT) at 620 and 655 nm. Results were calculated as a 665-nm/620-nm signal ratio, and IP1 concentrations were interpolated from a standard curve prepared using the supplied IP1 calibrator. Results are shown as picomoles of IP1 formed per well. For experiments involving PTX, cells were treated with PTX (100 ng/ml) for 16 hours at 37°C and 5% CO2 before stimulation. For experiments involving YM-254890, 10 μM YM-254890 or equivalent DMSO vehicle was added to the cells and incubated at 37°C for 30 min.

Measurements of cell surface CCR5

The amount of CCR5 on the cell surface was determined by flow cytometry. For one well of a six-well plate, HEK293T cells (1,000,000 in 2 ml of DMEM) were transfected with 2 μg of empty vector, 1 μg of CCR5 plus 1 μg of empty vector, or 1 μg of CCR5 plus 0.5 μg of Gi2, Gq or Gqi5 and 0.5 μg of empty vector. Twenty-four hours after transfection, cells were detached in ice-cold PBS and transferred to a microcentrifuge tube. Cells were spun down and then incubated with 100 μl of fluorescence-activated cell sorting buffer (PBS with 0.5 mM MgCl2 + 0.1% BSA) containing 8 μl of anti-CCR5 T21/8 phycoerythrin-conjugated antibody (eBioscience) for 60 min on ice. Antibody incubation was followed by three washes with ice-cold PBS. The cell surface receptor amount was quantified by flow cytometry using the Accuri C6 flow cytometer (BD Biosciences) within 1 hour after preparation.

Data analysis

Data were graphed and analyzed using GraphPad Prism 6.0 software (San Diego, CA). Dose-response curves were fitted by a three-parameter logistic equation.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/552/eaao6152/DC1

Fig. S1. Response of cells transfected with EPAC reporter and vector to chemokines and RANTES analogs.

Fig. S2. Dose-response curves for CCR5-mediated Gi/o protein activation by chemokines and RANTES analogs using cells expressing higher amounts of CCR5.

Fig. S3. The effect of chemokines and RANTES analogs on cAMP accumulation in the presence of PTX.

Fig. S4. Time courses of CCR5-mediated inhibition of cAMP accumulation induced by chemokines and RANTES analogs.

Fig. S5. Response of vector-transfected cells to chemokines and RANTES analogs through stimulation of Ca2+ flux.

Fig. S6. The effect of G protein subunit cotransfection on the amount of CCR5 molecules on the cell surface.

Fig. S7. ATP-induced Ca2+ flux in HEK293T cells expressing CCR5 and G protein subunits.

Fig. S8. The effect of G protein subunit cotransfection on CCR5-mediated Ca2+ flux and IP1 accumulation in response to chemokines and RANTES analogs.

Fig. S9. IP1 accumulation in vector-transfected cells in response to chemokines and RANTES analogs.

Fig. S10. Dose-response curves for CCR5-mediated Ca2+ flux in response to chemokines and RANTES analogs using cells expressing higher amounts of CCR5.

REFERENCES AND NOTES

Acknowledgments: We thank M. Bouvier (Université de Montréal) for providing the EPAC reporter cAMP reporter plasmid, B. Conklin (University of California, San Francisco) for the gift of the Gqi5 transfection vector, and O. Hartley (Université de Genève) for the gift of 5P12, 5P14, 6P4, and PSC. We thank R. Vaughn for help with statistical analysis and M. Kazmi and O. Hartley for critical reading of the manuscript. Funding: E.L. was funded by the David Rockefeller Fellowship. E.C. was funded by the François Wallace Monahan Fellowship. We also acknowledge grant support from the Robertson Foundation, the Crowley Family Fund, and the Danica Foundation to T.H. Author contributions: E.L., Y.A.B., E.C., C.A.R., A.F., T.H., and T.P.S. designed the experiments. E.L., Y.A.B., E.C., and C.A.R. performed analyses. E.L., T.H., and T.P.S. wrote the paper. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article