ACKR4 Recruits GRK3 Prior to β-Arrestins but Can Scavenge Chemokines in the Absence of β-Arrestins

Chemokines are essential for guiding cell migration. Atypical chemokine receptors (ACKRs) contribute to the cell migration process by binding, internalizing and degrading local chemokines, which enables the formation of confined gradients. ACKRs are heptahelical membrane spanning molecules structurally related to G-protein coupled receptors (GPCRs), but seem to be unable to signal through G-proteins upon ligand binding. ACKR4 internalizes the chemokines CCL19, CCL21, and CCL25 and is best known for shaping functional CCL21 gradients. Ligand binding to ACKR4 has been shown to recruit β-arrestins that has led to the assumption that chemokine scavenging relies on β-arrestin-mediated ACKR4 trafficking, a common internalization route taken by class A GPCRs. Here, we show that CCL19, CCL21, and CCL25 readily recruited β-arrestin1 and β-arrestin2 to human ACKR4, but found no evidence for β-arrestin-dependent or independent ACKR4-mediated activation of the kinases Erk1/2, Akt, or Src. However, we demonstrate that β-arrestins interacted with ACKR4 in the steady-state and contributed to the spontaneous trafficking of the receptor in the absence of chemokines. Deleting the C-terminus of ACKR4 not only interfered with the interaction of β-arrestins, but also with the uptake of fluorescently labeled cognate chemokines. We identify the GPCR kinase GRK3, and to a lesser extent GRK2, but not GRK4, GRK5, and GRK6, to be recruited to chemokine-stimulated ACKR4. We show that GRK3 recruitment proceded the recruitment of β-arrestins upon ACKR4 engagement and that GRK2/3 inhibition partially interfered with steady-state interaction and chemokine-driven recruitment of β-arrestins to ACKR4. Overexpressing β-arrestin2 accelerated the uptake of fluorescently labeled CCL19, indicating that β-arrestins contribute to the chemokine scavenging activity of ACKR4. By contrast, cells lacking β-arrestins were still capable to take up fluorescently labeled CCL19 demonstrating that β-arrestins are dispensable for chemokine scavenging by ACKR4.


INTRODUCTION
Chemokines, a group of about 50 chemotactic cytokines, have fundamental roles in regulating immune responses, primarily by orchestrating leukocyte migration and controlling their localization (1,2). The biological functions of chemokines are typically mediated by signaling through seven-transmembrane spanning, G-protein coupled receptors (GPCRs) (3,4). The chemokines CCL19 and CCL21 are essential for guiding dendritic cells and subsets of T cells to lymph nodes by signaling through the cognate chemokine receptor CCR7 (5,6) and thereby initiate adaptive immune responses. Notably, canonical CCR7 signaling by CCL19 and CCL21 is controlled by the G i subfamily of Gproteins (7)(8)(9). In addition, chemokines bind to a small family of atypical chemokine receptors (ACKRs), which are structurally related to GPCRs but seem unable to elicit canonical, G-proteindependent signal transduction pathways upon ligand binding (10,11). However, ACKRs are emerging as crucial regulators for the availability of chemokines. Namely, ACKRs function as "decoy" or "scavenger" receptors that progressively internalize chemokines and sort them for lysosomal degradation to limit local and systemic chemokine concentrations (10,12). The atypical chemokine receptor ACKR4, formerly also known as CCRL1 and CCX-CKR, is the scavenging receptor for CCL19, CCL21, and CCL25 (13)(14)(15)(16). Notably, mice lacking ACKR4 systemically have a 5-fold increase in the level of CCL21 in the blood and a 2-to 3-fold increase in CCL19 and CCL21 in peripheral lymph nodes (17). Despite its expression on thymic epithelial cells where ACKR4 is supposed to scavenge the CCR7 and CCR9 ligands CCL19/CCL21 and CCL25, respectively, mice lacking ACKR4 seem to have a fairly normal thymic T cell lymphopoiesis (18). By contrast, ACKR4 expression by skin keratinocytes and a subset of dermal endothelial cells is critical for shaping functional CCL19/CCL21 gradients under steady-state and inflammatory conditions (17). These local CCL19/CCL21 gradients are essential for allowing dendritic cells to egress the skin and enter lymphatic vessels (19,20). In addition, ACKR4 is present on lymphatic endothelial cells lining the ceiling of the subcapsular sinus, but not on those lining the floor, forming local CCL21 gradients in lymph nodes to guide dendritic cell homing in a CCR7-dependent manner (21). Consequently, the frequency of dendritic cells in the skin of ACKR4 deficient mice increases and dendritic cells fail to efficiently egress and migrate to draining lymph nodes (20,21).
In the present study, we provide evidence that ACKR4 neither interacts with nor activates heterotrimeric G-proteins. Chemokine binding to ACKR4 does also not activate canonical chemokine receptor kinases, such as Erk1/2, Akt, or Src. By contrast, chemokine triggering recruited β-arrestin1 and β-arrestin2 to ACKR4. Moreover, we identify GRK3, and to a lesser extent GRK2, as interaction partner of chemokine engaged ACKR4 and show that GRK3 recruitment precedes the recruitment of β-arrestins upon receptor triggering. We further demonstrate that the C-terminus of ACKR4 is critical for spontaneous receptor trafficking, β-arrestin recruitment and chemokine scavenging. Strikingly, overexpression of βarrestins increased chemokine uptake by ACKR4, whereas in the absence of β-arrestins ACKR4 was still able to take up cognate chemokines although to a lesser extent, thus providing clear evidence that β-arrestins are dispensable for chemokine scavenging.

Generation of Expression Plasmids
Reagents for molecular biology were purchased from Thermo Fisher Scientific and custom-designed primers from Microsynth. An overview of chemokine, receptor and β-arrestin constructs with corresponding primer sequences used for cloning are listed in Table 1. Briefly, pcDNA3 β-arrestin2i1-NLuc was generated by amplification of human β-arrestin2 and NLuc and subsequent ligation of the two PCR products over a common ClaI restriction site, followed by subcloning the DNA conjointly into the HindIII and XbaI sites of pcDNA3. Chemokines were amplified by PCR and further cloned into the XhoI and BsaI restriction sites of pET-His 6 -SUMO (41). SUMO-hCCL19-S6 was amplified by PCR and cloned into the XhoI and XbaI restriction sites of pET-His 6 -SUMO (41).
Nluc-GRK2 was generated by amplifying human GRK2 and Nluc separately, ligating the PCR products over a common ClaI restriction site and cloning it conjointly into the HindIII and XbaI sites of pcDNA3. The other GRK constructs were prepared  by replacing GRK2 with GRK3, GRK4, GRK5, and GRK6 using HindIII and ClaI or ClaI and XbaI, as listed in Table 2.
Site directed mutations of putative ACKR4 phosphorylation sites are listed in Table 3. Multiple site directed PCR were performed in consecutive cloning rounds to get ACKR4 TT , ACKR4 SSS, and ACKR4 TTSSS mutants.
To generate the BRET constructs for G proteins, all redundant HindIII, ClaI, BamHI, XhoI, or XbaI sites were removed by introducing silent mutations as listed in Table 4. Then, a BamHI site encoded in a SGGGGS linker was introduced (Table 5, Supplementary Figure 3). Further, the modified Gα-subunits were amplified with adjacent HindIII and XbaI sites and cloned into pcDNA3. PCR amplified Nluc was introduced into the BamHI sites ( Table 5). An exception is Gα q , were the RLuc8 in Gα q -RLuc8 was replaced via BamHI cutting and insertion of Nluc. To generate pIRES Gβ-2A-cpV-Gγ2 Gα-Nluc, a redundant HindIII was removed and a new one added after the IRES sequence using site directed mutagenesis. Then the mutated IRES sequence was amplified using the forward primer hybridizing at a SalI site and the reverse primer with a HindIII site, hybridizing to the one introduced beforehand, followed by an XbaI site at its end. The PCR-product was ligated into the original IRES plasmid, removing the Gαi2-mTurquoise2 sequence, after digesting both with SalI and XbaI. Gα i -Nluc was cut out from pcDNA3 Gα i -Nluc utilizing HindIII and XbaI and ligated into the modified IRES vector.

BRET Measurements
Transfected HeLa cells were grown in 6 well plates, washed with PBS, and detached using PBS based Gibco TM cell dissociation buffer (#13151014, Thermo Fisher Scientific) for a minimum of 3 min. Cells were collected in twice the volume of dissociation buffer with DMEM containing 10% FCS before being centrifuged for 2 min at 200 g. Cells were washed and resuspended in PBS containing 5% (w/v) glucose (PBS-G  (48). For short term observations (−1 to 3 min), the integration time of both signals was decreased to 250 ms and an injector used for chemokine addition. To calculate NetBRET, BRET ratio of control wells containing luciferase and HA-tagged receptor instead of EYFP-tagged receptor was subtracted from the sample BRET ratio. For Gα i activation, the control wells contained cells transfected with pIRES Gα-Nluc Gβγ-cpVenus alone. Area under the curve analysis (AUC) was performed using the measurements before stimulation as baseline and integrating the peak starting from 0 min until the end of measurement. For data representation of GRK and G protein activation, baseline reduction was performed using the measurements before addition of ligands, which is referred to as "corrected NetBRET."

Data Analysis
Data analysis and presentation was performed using GraphPad Prism V.7 and V8. For statistics with one variable, RM oneway ANOVA or mixed-effects analysis, both with Dunnett's multiple comparisons test with a single pooled variance was performed (Figure 4). For statistics of Western blot ratio, mixedeffect model with Tukey's multiple comparisons test with a single pooled variance was performed (Figure 1). For experiments using two variables, ordinary two-way ANOVA with Tukey's multiple comparisons test, with individual variances computed for each comparison was performed (Figures 2, 5). EC50 values were calculated fitting a three parameter [agonist] vs. response curve. * p < 0.05, * * p < 0.005, * * * p < 0.0005, * * * * p < 0.0001.

ACKR4 Does Not Elicit Canonical Chemokine-Mediated Signal Transduction Pathways
ACKRs, including ACKR4, were reported not to signal through heterotrimeric G i -proteins manifested by the failure to induce cell migration or calcium mobilization (16,26). However, it has been speculated that the G i -protein might associate with ACKR4, hence sterically block G s activation unless it dissociates from the receptor (26). To address this possibility, we established bioluminescence resonance energy transfer (BRET)based assays to measure G-protein activation and interaction with the receptor. We engineered human Gα-proteins (Gα i , Gα q , Gα s , Gα o , or Gα 13 ) where we introduced the Nano luciferase (Nluc) as luminescence donor into the unstructured region  after the second helix (51). As luminescence-acceptor, we used human ACKR4, or human CCR7, fused to EYFP to measure steady-state association, chemokine-driven recruitment, as well as activation-dependent dissociation of the heterotrimeric Gprotein from the receptor (Figure 1A). Co-expressing Gα-Nluc variants together with ACKR4-EYFP in HeLa cells revealed that none of these tested Gα-proteins associated with ACKR4 under steady-state conditions ( Figure 1B). Stimulating cells with 1 µg/ml of human chemokine CCL19, known to elicit strong CCR7 responses, neither recruited one of the Gαproteins to ACKR4, nor resulted in the dissociation of one of the G-proteins from the receptor ( Figure 1B). By contrast, Gα i was found to pre-associate with the canonical chemokine receptor CCR7-EYFP and dissociated from the receptor upon CCL19 stimulation (Figure 1B), which is in line with the preassociation of the G s -protein with the adrenergic receptor and its subsequent ligand-induced dissociation (51). Next, we used Gα i -Nluc and Gβγ fused to cpVenus as luminescence-acceptor to monitor activation-induced dissociation of Gβγ from the Gα i -subunit ( Figure 1C). Whereas, CCL19 stimulation led to the dissociation of the Gα i from the Gβγ-subunit upon CCR7 triggering, this was not observed for ACKR4 ( Figure 1D). Consistent with these findings, CCL19 triggering of CCR7, but not of ACKR4, lead to the phosphorylation and activation of the MAP kinase Erk1/2 ( Figure 1E, Supplementary Figure 1) and protein kinase B/Akt (Figure 1F, Supplementary Figure 1) through the canonical G i -signaling pathway (7-9). Moreover, CCL19 stimulation of CCR7 caused the phosphorylation of the kinase Src ( Figure 1G, Supplementary Figure 1), which occurs through G-protein-independent signaling (9). Again, no CCL19-mediated Src phosphorylation was observed upon ACKR4 triggering (Figure 1G, Supplementary Figure 1). These data provide comprehensive evidence that ACKR4 does neither associate with and activate G-proteins, nor elicits canonical chemokine receptor signaling pathways involving Erk1/2, Akt or Src kinases.

The C-Terminus of ACKR4 Controls Interaction and Recruitment of β-Arrestins and Is Essential for Chemokine Uptake
As the role of β-arrestins in chemokine scavenging by ACKR4 is debated (16,26), we determined β-arrestin1 and β-arrestin2 recruitment to ACKR4 by BRET (Figure 2A). We co-expressed EYFP-tagged human ACKR4 together with either Nluc-tagged human β-arrestin1 or β-arrestin2 in Hela cells and stimulated the cells with graded concentrations of the human ACKR4 ligands CCL19, CCL21, and CCL25.
Chemokine Triggering Recruits GRK3, and to a Lesser Extent GRK2, to ACKR4 To identify which GRK promotes putative receptor phosphorylation and subsequent β-arrestin recruitment we established BRET assays to measure recruitment of individual GRKs to engaged ACKR4 ( Figure 3A). Therefore, we fused Nluc to all ubiquitously expressed human GRKs and coexpressed them individually with ACKR4-EYFP in HeLa  cells. As internal control we also determined β-arrestin2-Nluc recruitment to ACKR4-EYFP. Cells were stimulated with the three ACKR4 ligands at a concentration representing about 3-times the EC50 value of β-arrestin2 recruitment.
To determine chemokine-mediated recruitment of signaling molecules, basal NetBRET values were subtracted for each condition and the area under the curve (AUC) for the first 3 min of stimulation were calculated as depicted in Figure 3B. A comprehensive analysis revealed that CCL19, CCL21, and CCL25 selectively and efficiently recruited GRK3 to ACKR4, whereas no interaction of ACKR4 with GRK4, GRK5, or GRK6 was observed ( Figure 3C). Dose-response kinetic analysis revealed similar EC50 values for the recruitment of GRK3 ( Figure 3D) by CCL19 (EC50∼42 nM), CCL21 (EC50∼142 nM), and CCL25 (EC50∼147 nM), as determined for the recruitment of β-arrestin2 by these ACKR4 agonists. Chemokine-mediated GRK3 recruitment to ACKR4 was fast, reaching its maximum within a minute (Figure 3E), and preceded the recruitment of β-arrestin1 and β-arrestin2 ( Figure 3F). A less pronounced chemokine-mediated BRET signal was also observed between GRK2 and ACKR4 (Figures 3C,H). Subsequent dose-response kinetic analysis for GRK2 ( Figure 3G) revealed EC50 values for CCL19 (EC50∼35 nM), CCL21 (EC50∼129 nM), and CCL25 (EC50∼87 nM) that are comparable to those for GRK3. The chemokine-mediated interaction between GRK2/3 and ACKR4 was not as transient as one could expect, which can be explained by the spontaneous trafficking of ACKR4 that continuously deliver receptor molecules to the plasma membrane that can interact with GRK2/3 over time and upon chemokine triggering.
Notably, steady-state interaction of GRK2/3 with the tailless variant ACKR4t was abrogated and no chemokine-mediated recruitment of GRK2 or GRK3 to ACKR4t was observed ( Figure 3I). To investigate the role of GRK2/3 in the recruitment of β-arrestin to ACKR4, we treated cells with cpmd101, a known GRK2/3 inhibitor (31). Treating cells with cpmd101 reduced both basal interaction of ACKR4 with β-arrestin2 (Figure 3J), as well as chemokine-mediated β-arrestin2 recruitment to the receptor (Figures 3K,L).
Notably, although cpmd101 treatment interfered with the recruitment of β-arrestin to ACKR4, the interaction was not completely abolished, suggesting that other kinase(s) contribute to potential ACKR4 phosphorylation and subsequent β-arrestin recruitment. To address this, we searched for putative serine/threonine phosphorylation sites of ACKR4. In silico studies using NetPhos 2.0 server (http://www.cbs. dtu.dk/services/NetPhos/) and NetPhosK 1.0 server (www. cbs.dtu.dk/services/NetPhosK) predicted several putative phosphorylation sites or protein binding motifs for ACKR4 ( Figure 4A). In order to validate these putative phosphorylation and kinase binding sites of ACKR4, we performed site-directed mutagenesis to exchange tyrosine residues for phenylalanine and serine/threonine residues for alanine and determined steady-state interaction and CCL19-driven recruitment of β-arrestin2 by BRET as depicted in Figure 4B. Whereas none of the tyrosine mutants affected steady-state interaction or stimulationdependent recruitment of β-arrestin2 to ACKR4, a number of serine/threonine single point-mutants significantly reduced the interaction between β-arrestin2 and ACKR4 ( Figure 4C). Most prominently, ACKR4 T142A showed severely impaired steady-state interaction with β-arrestin2 without affecting the chemokine-driven β-arrestin2 recruitment. Importantly, none of these ACKR4 mutants showed significantly impaired surface expression or CCL19 Dy649P1 binding capabilities ( Figure 4D). Additional sites affecting the steady-state interaction with β-arrestin2 include ACKR4 T226A , ACKR4 S309A , ACKR4 S323A , and ACKR4 S330A , which, together with ACKR4 T142A , are predicted as putative PKC phosphorylation sites (Figures 4A,C). Thus, we generated additional mutants, where the two threonine residues (ACKR4 T142T226 ), the three serine residues (ACKR4 S309S323S330 ) or the combination thereof (ACKR4 TTSSSmut ) were replaced by alanines. Notably, steady-state interaction of ACKR4 TTSSSmut with β-arrestin2 was profoundly reduced, whereas the other two mutants showed an intermediate phenotype ( Figure 4E). Similarly, ACKR4 TTSSSmut showed a significantly decreased ability to recruit β-arrestin2 upon CCL19 stimulation (Figure 4E), while retaining their surface expression and chemokine binding abilities ( Figure 4F).
Taken together, these data demonstrate that chemokine triggering selectively recruits GRK3, and to a lesser extent GRK2, to ACKR4 and suggest that GRK2/3 and other serine/threonine kinases contribute to the recruitment of β-arrestins to the receptor. Although, mutating selected serine and threonine residues is not a direct proof that these residues are indeed phosphorylated by GRK2/3, our data provide evidence that these residues are critical for β-arrestin recruitment.
β-Arrestins Contribute to, but Are Dispensable for Chemokine Uptake To assess the role of β-arrestins in steady-state trafficking and chemokine scavenging by ACKR4, we exploited wild-type HeLa (HeLa wt) and β-arrestin1/β-arrestin2-double deficient HeLa (HeLa KO) cells expressing mTurquoise2-tagged ACKR4 (ACKR4-mTq2). ACKR4-mTq2 showed the expected surface and mainly vesicular localization in HeLa wt cells (Figure 5A). By contrast, ACKR4-mTq2 predominantly associated with the plasma membrane in HeLa KO cells and was less present in vesicular structures (Figure 5A), similarly to ACKR4t-EYFP ( Figure 2F). Reconstituting HeLa KO cells by Where indicated, cells were exposed to a short acidic wash to remove surface bound, but not internalized CCL19 Dy649P1 . n = 4.
reintroducing β-arrestin2-YFP, promoted the re-localization to predominantly vesicular and surface localization of ACKR4-mTq2 (Figure 5A), supporting the notion that β-arrestins control the steady-state trafficking of ACKR4. To asses ACKR4mediated chemokine scavenging, we incubated transfected HeLa cells with fluorescently labeled CCL19 Dy649P1 for various time points at 37 • C (Figures 5B,C). CCL19 Dy649P1 was steadily taken up over time by ACKR4 in HeLa wt cells ( Figure 5D). Exposing cells to a short acidic wash hardly reduced chemokine-derived fluorescence, indicated that CCL19 Dy649P1 was indeed internalized (Figure 5E). Remarkably, uptake of CCL19 Dy649P1 was significantly reduced by roughly ∼40-50% in HeLa KO cells, but was not completely abolished (Figures 5D,E). Moreover, overexpression of β-arrestin2 in either HeLa wt or HeLa KO cells significantly enhanced CCL19 Dy649P1 uptake (Figures 5D,E).
In summary, we show that β-arrestins interact with ACKR4 in the steady-state and contribute to the spontaneous trafficking of the receptor. Furthermore, we demonstrate that β-arrestins enhance the scavenging activity of ACKR4, but are dispensable for chemokine uptake.

DISCUSSION
ACKR4 plays an important role in the regulation of immune cell migration by shaping local chemokine gradients (20,21). The molecular mechanism how ACKR4 scavenges its cognate ligands remains poorly understood. Initially, biotinylated CCL19 was detected in vesicular structures of ACKR4 transfected MEFs derived from β-arrestin1/β-arrestin2 double-deficient mice, suggesting that chemokine uptake by ACKR4 is not critically dependent on β-arrestins (16). By contrast, CCL19 stimulation of an ACKR4 transfected osteosarcoma cell line was shown to result in the translocation of β-arrestin2-GFP to vesicular structures (26). In addition, chemokine stimulation recruited β-arrestin1 and β-arrestin2 to ACKR4 using slit-galactosidase and BRET assays in CHO cell transfectants (26). This later study is in line with the common concept of a β-arrestin-dependent receptor trafficking route taken by class A GPCRs (27). In the present study we show that CCL19, CCL21, and CCL25 effectively recruit β-arrestin1 and β-arrestin2 to engaged ACKR4, which confirms the study by Watts and colleagues (26). In addition to that study, we found that β-arrestin already interacts with ACKR4 prior to chemokine stimulation and that this steady-state interaction occurs at vesicular structures. Notably, an ACKR4 mutant lacking its C-terminus not only failed to interact with and recruit βarrestins, it also lost its vesicular localization and showed an impaired capacity to take up chemokines. Interestingly, a Cterminally truncated variant of ACKR2 also fails to recruit βarrestins, but was still able to scavenge chemokines (29), whereas a C-terminal deletion variant of ACKR3 (31,53) showed a similar absence of chemokine scavenging behavior as ACKR4t. Together with the finding that overexpression of β-arrestin2 enhanced chemokine uptake, our data indicate that β-arrestins control steady-state trafficking of ACKR4 and contributes to an enhanced chemokine scavenging activity. However, we also provide experimental evidence that β-arrestins are dispensable for chemokine uptake by ACKR4, as β-arrestin1/β-arrestin2double deficient HeLa cells are still able to internalize chemokines although less efficient that wild-type cells. Notably, CCL19 uptake by ACKR4 was shown to be partially reduced in HEK293 cells treated with methyl-β-cyclodextrin to deplete cholesterol or in cells overexpressing caveolin-1 or a dominant-negative form of dynamin, but not in cells overexpressing a dominant-negative form of Eps15 or Rab5 (16). These data conjointly suggest, that ACKR4 and likely other ACKRs utilize additional routes of endocytosis compared with canonical chemokine receptors.
Due to the lack of canonical G protein-dependent signaling, ACKRs were initially considered to be silent receptors. More recently, ACKR3 was described to execute a signaling bias toward β-arrestins leading to MAP kinase activation (24), an alternative signaling pathway for canonical class A GPCRs (27). β-arrestin signaling usually relies on GPCR kinase recruitment and subsequent receptor phosphorylation. Consistent with this concept, GRK2 (and partially GRK5) recruitment was shown to induce ACKR3 phosphorylation upon chemokine stimulation (31). Here, we identified that GRK3 and GRK2, but no other GRK, are selectively recruited to chemokine engaged ACKR4 and that GRK2/3 recruitment precedes the recruitment of β-arrestins, pointing to a remarkable specificity of GRKs for different ACKRs. Inhibiting GRK2/3 by cpmd101 partially, but significantly reduced steady-state interaction as well as chemokine-driven recruitment of β-arrestins to ACKR4. However, we did not find any experimental evidence for a βarrestin-dependent or independent phosphorylation of Erk1/2 and Akt upon chemokine triggering of ACKR4. Our data are thus in line with a previous study on ACKR4 showing no Erk1/2 activation (26) and one on the adrenergic receptor showing that β-arrestins are dispensable for Erk1/2 phosphorylation (54).
In conclusion, it emerges that distinct GRKs are recruited to ACKRs (GRK2/5 for ACKR3; GRK2/3 for ACKR4) upon ligand stimulation, which phosphorylate C-terminal serine/threonine residues of the receptor (31) and thereby recruit β-arrestins. We herein further provide evidence that β-arrestins control steadystate trafficking of ACKR4 and promote chemokine uptake. However, it is becoming clear that β-arrestins are dispensable for chemokine scavenging by ACKR2 (29), ACKR3 (30,31) and ACKR4 (16). The fact that GRKs are recruited to and phosphorylate the receptors strongly indicates that ACKRs are not silent receptors, but are able to elicit alternative, yet unknown signaling pathways. This is most convincingly supported by the fact that mice lacking ACKR3 die at birth with ventricular septal defects and semilunar heart valve malformation (55), while mice expressing a chemokine scavenging deficient ACKR3 are vital (31).