Regulation of the chemokine receptors CXCR4 and ACKR3 by receptor activity-modifying proteins

Abstract

The chemokine CXCL12 and its two cognate receptors-CXCR4 and ACKR3-are key players in various homeostatic and pathophysiological processes, including embryonic development, autoimmune diseases, tissue repair, and cancer. Recent reports identified an interaction of CXCR4 and ACKR3 with receptor activity-modifying proteins (RAMPs), and RAMP3 has been shown to facilitate ACKR3's recycling properties. Yet, the functional effects of RAMPs on the CXCL12 signaling axis remain largely elusive. Here, we characterize the effects of RAMPs on CXCR4 and ACKR3 function. We show that, in the absence of a ligand, RAMPs do not affect the cell membrane localization or constitutive internalization of the two receptors. RAMP3 inhibits ligand-stimulated internalization of ACKR3, which retains the receptor at the membrane and inhibits its ability to scavenge CXCL12. In addition, while cAMP inhibition by CXCR4 is unaffected by RAMPs, basal and ligand-stimulated β-arrestin recruitment to both CXCR4 and ACKR3 is reduced in the presence of RAMP3 due to complex formation at the cell surface. The effects on ACKR3 are observed for chemokine, small molecule, and peptide agonists as well as for a N-terminal truncated receptor variant, suggesting that RAMP regulation involves contacts with the transmembrane domain of the receptor. Taken together, our results show that RAMPs regulate the CXCL12 signaling axis by directly interfering with receptor function. These findings could have direct implications for the interplay between receptors in vivo as well as future drug design in the therapeutic targeting of the CXCL12 signaling axis.

Keywords: ACKR3; CXCL12; CXCR4; G protein-coupled receptor (GPCR); arrestin recruitment; chemokine receptor; protein-protein interaction; receptor activity-modifying proteins (RAMPs); receptor internalization; signal transduction.

Figure 1

CXCR4 and ACKR3 expression.A, surface expression of both CXCR4 and ACKR3 was quantified by SNAP-labeling in HEK293A cells transfected with either receptor −/+ RAMP1-3 (n = 4 for CXCR4 and n = 7 for ACKR3). B, total expression of CXCR4 and ACKR3 was quantified by Western blot analysis of transfected cells. (n = 4 for CXCR4 and n = 5 for ACKR3). (theoretical MWs: SNAP-ACKR3 62.355 kDa, SNAP-CXCR4 60.608 kDa). Each data point represents one experiment shown as a percentage of the control column (receptor alone) and plotted as mean ± SD (bar charts) (A & B). C, Western blot of ACKR3 (100–200ng transfected ACKR3 DNA + RAMP3 in 1:3 receptor:RAMP ratio) from transiently transfected, solubilized HEK293A cells. Statistical analysis was performed on the non-normalized data using ordinary two-way ANOVA of main effects with Dunnett’s test for multiple comparisons (A & B) (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗p < 0.05).

Figure 2

Constitutive and CXCL12-induced receptor internalization in the presence of RAMPs.A, schematic representation of the internalization assay. SNAP-receptors are labeled at the cell surface and internalization is stimulated with ligand (CXCL12). When receptors internalize, the quenching effects of the extracellularly surrounding Fluorescein are lost, resulting in a measurable increase of 615nm emission, consistent with receptor endocytosis. B, constitutive and CXCL12-internalization of CXCR4 and ACKR3), shown as mean ± SEM of independent experiment (n = 4 for CXCR4 and n = 7–10 for ACKR3). The bar plot illustrates the percentage of constitutive internalization as compared to CXCL12-induced internalization for either receptor, quantified as the AUC over 90 min (mean ± SEM). C and D, CXCL12-induced internalization of CXCR4 (C) or ACKR3 (D) in the presence of RAMPs. Internalization curves (left) are presented as mean ± SEM of n = 4 and n = 5 to 10 experiments for CXCR4 and ACKR3, respectively. For the quantification of internalization (bar plots, right), each data point represents the AUC (90 min) of one experiment shown as a percentage of the control column (receptor alone) and plotted as mean ± SEM of biological replicates (n = 4 and n = 5–10 experiments for CXCR4 and ACKR3, respectively). E, remaining SNAP-ACKR3 at the cell surface after 30-min incubation with CXCL12, plotted as a percentage of the remaining receptor after treatment with buffer (basal conditions) (n = 3–4). All experiments were performed in technical triplicates. Statistical analysis was performed on the non-normalized data using ordinary two-way ANOVA of main effects with Dunnett’s test for multiple comparisons and presented as percentage of receptor alone (mean ± SEM) (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, - = ns).

Figure 3

CXCL12 scavenging determined by ELISA.A, schematic representation of the scavenging assay: Cells were transfected with receptor, receptor + RAMPs or pcDNA and plated in a 96-well plate. Following overnight incubation with CXCL12, CXCL12 concentration in supernatants was determined by ELISA. B, remaining CXCL12 in cell supernatants in the absence or presence of either ACKR3 or CXCR4, presented as percentage of pcDNA (n = 4) C, CXCL12 scavenged by ACKR3 in the presence or absence of either RAMP1 or RAMP3 (n = 4–5). All experiments were performed in technical triplicates. Statistical analysis was performed on the non-normalized data using ordinary two-way ANOVA of main effects with Dunnett’s test for multiple comparisons and presented as percentage of the control column (pcDNA (B) or ACKR3 (C)) (mean ± SEM) (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, - = n.s.).

Figure 4

ACKR3 internalization (−/+ RAMPs) in response to different agonists.A, ACKR3 agonists are grouped into chemokines (red), small molecule (blue), and peptides (green) (PDB: 7SK6). B, AUC (90 min) analysis of ACKR3 internalization curves with the different ACKR3 agonists, presented as percentage of constutitive internalization (n = 3–10). C–H, internalization of ACKR3 (−/+ RAMP1&3) in response to the chemokine CXCL11 (n = 4) (C), the small molecule VUF11207 (n = 5) (D), and the peptides adrenomedullin (n = 3) (E), PAMP-12 (n = 3) (F), and TC14012 (n = 3) (G). Internalization curves (left in each panel) are mean ± SEM of biological replicates and quantified as the AUC over 90 min (barplots, right in each panel), where each datapoint represents the mean of a single experiment. All experiments were performed in technical triplicates. Statistical analysis was performed on the non-normalized data by ordinary two-way ANOVA of main effects with Dunnett’s correction for multiple testing. (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, - = ns).

Figure 5

Internalization of SNAP-RAMPs.A, schematic representation of the internalization assay: SNAP-RAMPs are co-expressed with CLR (grey) or ACKR3 (red) and RAMP internalization is monitored over time, as described in Fig. 2. B–D, internalization curves (left in each panel) for RAMP alone (open circles) and co-expressed with CLR (gray circles) or ACKR3 (red circles). In the presence of a receptor, RAMP internalization was stimulated with receptor-specific ligands (adrenomedullin for CLR, CXCL12 for ACKR3). Internalization curves are shown as mean ± SEM of 3 to 4 independent experiments (n = 3–4). AUC of internalization curves (200 min) analyzed for RAMP1 (B), RAMP2 (C), and RAMP3 (D) in bar plots (right), both in the presence and in the absence of receptor and buffer or respective ligand adrenomedullin (CLR) or CXCL12 (ACKR3). All experiments were performed in technical triplicates. Statistical analysis was performed on the non-normalized data by ordinary two-way ANOVA of main effects with Dunnett’s correction for multiple testing. (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01,- = ns).

Figure 6

β-Arrestin recruitment to CXCR4 and ACKR3 −/+ RAMPs.A, schematic representation of the assay. Upon recruitment of GFP- β-arrestin to the receptor-Rluc, energy is transferred (BRET) from the Rluc (donor) to the GFP (acceptor), resulting in a measurable increase of the BRET ratio. B, constitutive β-Arrestin recruitment to CXCR4 (top, n = 3) and ACKR3 (bottom, n = 14), presented as the decrease in baseline of BRET-ratios compared to receptor alone (Δ Baseline change) C and D, CXCL12-induced recruitment of β-Arrestin to CXCR4 (n = 4) (C) and ACKR3 (n = 6) (D) (−/+ RAMPs) (left), with potency differences (ΔpEC50, pEC₅₀(_+RAMP) – pEC_{50(only receptor)}) compared to receptor alone (bar graph, middle) and efficacy changes (% Emax change) (bar graph, right) in absence and presence of RAMPs. Dose-response curves (left) are presented as mean ± SEM of independent experiments and potency and efficacy changes from each experiment plotted as a single data point (mean ± SEM). All experiments were performed in technical triplicates. Statistical analysis was performed on the non-normalized data by ordinary two-way ANOVA of main effects with Dunnett’s correction for multiple testing. (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, - = ns).

Figure 7

β-Arrestin recruitment to ACKR3 (−/+ RAMPs) with different ligands.A–C, Arrestin recruitment was induced by CXCL11 (n = 4) (A), TC14012 (n = 3) (B) and VUF11207 (n = 4) (C). Efficacy changes are presented as % change of Emax of ACKR3 alone (bargraph, middle in each panel) and potency differences as ΔpEC50 (pEC₅₀(_+RAMP) – pEC_{50(only receptor)}) compared to ACKR3 alone (bargraph, right in each panel). D, β-Arrestin recruitment to an N-terminal cleavage mutant of ACKR3 (−/+ RAMP3), where 28 residues have been removed from ACKR3’s N-terminus (n = 3). Dose-response curves are presented as mean ± SEM of independent experiments preformed in technical triplicates. Statistical analysis was performed on the non-normalized data by ordinary two-way ANOVA of main effects with Dunnett’s correction for multiple testing. (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, - = ns).

Figure 8

β-Arrestin recruitment to surface ACKR3.A, ACKR3 was co-expressed with the dominant negative Dynamin mutant (Dyn-K44A) to retain the receptor at the cell surface, and β-arrestin2 recruitment was quantified at different CXCL12 concentrations (n = 4). B, CXCL12-induced β-arrestin recruitment to surface-retained ACKR3 in the presence of RAMP3 (n = 4). C, β-arrestin2 recruitment to ACKR3 in the presence or absence of SNAP-RAMPs and induced with CXCL12 (n = 3). All dose responses are shown as mean ± SEM of independent experiments performed in technical duplicates.