Frontline Science: Antagonism between regular and atypical Cxcr3 receptors regulates macrophage migration during infection and injury in zebrafish

Abstract The CXCR3‐CXCL11 chemokine‐signaling axis plays an essential role in infection and inflammation by orchestrating leukocyte trafficking in human and animal models, including zebrafish. Atypical chemokine receptors (ACKRs) play a fundamental regulatory function in signaling networks by shaping chemokine gradients through their ligand scavenging function, while being unable to signal in the classic G‐protein‐dependent manner. Two copies of the CXCR3 gene in zebrafish, cxcr3.2 and cxcr3.3, are expressed on macrophages and share a highly conserved ligand‐binding site. However, Cxcr3.3 has structural characteristics of ACKRs indicative of a ligand‐scavenging role. In contrast, we previously showed that Cxcr3.2 is an active CXCR3 receptor because it is required for macrophage motility and recruitment to sites of mycobacterial infection. In this study, we generated a cxcr3.3 CRISPR‐mutant to functionally dissect the antagonistic interplay among the cxcr3 paralogs in the immune response. We observed that cxcr3.3 mutants are more susceptible to mycobacterial infection, whereas cxcr3.2 mutants are more resistant. Furthermore, macrophages in the cxcr3.3 mutant are more motile, show higher activation status, and are recruited more efficiently to sites of infection or injury. Our results suggest that Cxcr3.3 is an ACKR that regulates the activity of Cxcr3.2 by scavenging common ligands and that silencing the scavenging function of Cxcr3.3 results in an exacerbated Cxcr3.2 signaling. In human, splice variants of CXCR3 have antagonistic functions and CXCR3 ligands also interact with ACKRs. Therefore, in zebrafish, an analogous regulatory mechanism appears to have evolved after the cxcr3 gene duplication event, through diversification of conventional and atypical receptor variants.

that they bind (CCL, CXCL, XCL, and CX3CL). 6,7 A distinctive feature of chemokine signaling is its pleiotropic nature. Most chemokine receptors can bind multiple chemokines, and chemokines can also bind to numerous receptors. 2,5 The redundancy of the interactions and the diversity of processes involving chemokine receptors require tightly regulated mechanisms to confer specificity to the response resulting from a receptor-ligand interaction. 6,8,9 Therefore, chemokine signaling-axes regulation and signal integration occur at different levels (genetic, functional, spatial, and temporal) and engage a wide variety of mechanisms to evoke specific responses. [10][11][12] One kind of mechanism for regulating chemokine receptor activities involves atypical chemokine receptors (ACKRs), a heterogeneous group of proteins. 13,14 Despite their structural diversity and distant evolutionary relationships, all ACKRs are unified by their inability to signal in the classic G protein-dependent fashion and by their shared capacity to shape chemokine gradients. 13,15 These receptors display characteristic features such as amino acid substitutions within the central activation E/DRY-motif (aspartic/glutamic acidarginine-tyrosine-motif), 13,16 which is crucial for G-protein coupling and further downstream signaling. 16 The central arginine (R) of the E/DRY-motif is highly conserved (96%) among functional GPCRs as it is critical for locking and unlocking the receptor and substitutions of this residue usually result in loss of function. 16,17 In addition, ACKRs show alterations in amino acid residues within the TM domains that function as microswitches by stabilizing the active conformation of a GPCR. ACKRs have been shown to exert their function by scavenging or sequestering chemokines or by altering the activity or membrane expression of conventional chemokine receptors. 10,13 The functional read-out of ACKRs is that they fail to induce cell migration, contrary to the well-characterized chemotactic function of conventional chemokine receptors. 13,18 The zebrafish model has been successfully used to functionally unravel mechanistic processes underlying chemokine networks involving ACKRs. 19,20 The optical transparency of larvae facilitates live visualization of immunological processes and provides a reasonably simplified in vivo model for chemokine signaling if used before adaptive immunity arises. [21][22][23][24] Besides, due to the extensive duplication of chemokine receptor genes in teleost fish, the zebrafish provides a useful experimental system to address sub-functionalization or loss of function events. The sub-functionalization of 2 CXCR4 genes, cxcr4a and cxcr4b, was determined using the zebrafish model. In several studies, cxcr4a was associated primarily with cell proliferation, 11,19 whereas cxcr4b was related to the retention of hematopoietic stem cells in hematopoietic tissue, recruitment of leukocytes to sites of infection and damage, modulation of inflammation, neutrophil migration, primordial cell and tissue migration, and tissue regeneration. 25 Cxcr4b interacts with Cxcl12a and it was shown that this chemokine is also a ligand for the scavenger receptor Cxcr7 (ACKR3). 26,27 Interacting with both receptors, Cxcl12a has been shown to control the migration of a tissue primordium, in which expression of cxcr4b and cxcr7 is spatially restricted to the leading and trailing edge, respectively. 11,19 The scavenging role of CXCR7 (ACKR3) in the regulation of the CXCL12-CXCR4 axis was later confirmed in human cells. 26 Moreover, the zebrafish model allowed to visualize the contribution of endogenous chemokine receptors in shaping self-generated gradients of migrating cells, 20 and revealed how the cell-type expressing a given chemokine receptor is the major determinant for the functional specificity of a chemokine receptor-ligand interaction, and not the receptor-ligand pair itself. 28 The human CXCR3 chemokine receptor and its ligands (CXCL9 -11) have been proven instrumental for T-cell functioning as well as for macrophage recruitment to sites of infection and injury, and are therefore implicated in several infectious and pathological conditions, including tuberculosis. 29,30 CXCR3 ligands have been proposed as clinical markers for the diagnosis of this infectious disease and the response to treatment. 31,32 In a previous study, we assessed the role of CXCR3 in mycobacterial infection using the zebrafish-Mycobacterium marinum model and observed that CXCR3 ligands were induced upon infection in this model, such as in human patients. 29,33 Mycobacterium marinum is a close relative of Mycobacterium tuberculosis and a natural pathogen of various ectotherms, such as zebrafish, which has become widely used to unravel early innate immune responses against mycobacterial infections. 21,33,34 In zebrafish there are 3 copies of the CXCR3 gene: cxcr3.1, cxcr3.2, and cxc3. 3. We determined that the latter 2 are expressed on macrophages at early developmental stages as well as at 5 and 6 days postfertilization (dpf) 35 and that cxcr3.2 is a functional homolog of human CXCR3. 29 Macrophages play a pivotal role in mycobacterial infections because they are motile and phagocytic cells as well as a constituent cell type of the characteristic granulomas that represent inflammatory infection foci. 30,33 The efferocytosis of infected macrophages in granulomas contributes to the amplification of the infection and is a crucial process to consider to design new therapeutic strategies. 21,29 In a previous study, we showed that Cxcr3.2 is required for the proper migration of macrophages to infectious foci. 29 However, in agreement with studies in cxcr3 mutant mice, mutation of cxcr3.2 is beneficial to the host in the context of mycobacterial infection. 30 We showed that cxcr3.2 mutation favors bacterial contention, because it results in a reduced macrophage motility, thereby preventing macrophage-mediated dissemination of bacteria and limiting the expansion of granulomas.
Although Cxcr3.2 is required for macrophage migration in zebrafish, the function of its paralog, Cxcr3.3, which is also expressed on macrophages, remains unknown. In the present study, we investigated the regulatory interplay between Cxcr3. 2  receptor can bind the same ligands as Cxcr3.2 because of the high conservation of the ligand-binding sites, but also that it cannot signal using classic G protein-dependent pathways. Taking both our structural and functional data together, we posit that the 2 CXCR3 zebrafish paralogs cxcr3.2 and cxcr3.3 function antagonistically. We propose that Cxcr3.3 is an ACKR that functionally regulates the activity of Cxcr3.2 by scavenging common ligands and that knocking out cxcr3.3 results in an exacerbated Cxcr3.2 signaling due to an excess of available chemokines.

Phylogenetic analysis and protein-ligand binding site prediction
Amino acid sequences of CXCR3 genes and ACKRs from 13 species (Supplementary Table 1) were aligned and trimmed using the freeaccess server gBlocks 40 and the protein evolution analysis method was fitted using ProtTest3. 41 Evolutionary analyses were conducted in MEGA7. 42

Systemic infection with M. marinum and determination of bacterial burden
Mycobacterium marinum M-strain, expressing the fluorescent marker wasabi, was grown and prepared freshly for injection as described by Benard et al., 48 and embryos were systemically infected with 300 CFU of M. marinum-wasabi by microinjection at 28 hpf in the blood island (BI). 48,49 Infected larvae were imaged under a Leica M165C stereoflorescence microscope and the bacterial burden was determined using a dedicated pixel counting program at 4 days postinfection (4 dpi). 50 Data were analyzed using a two-tailed t-test and a one-way ANOVA when more than 2 groups were compared. Results are shown as mean ± SEM (ns P > 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, and ****P ≤ 0.0001) and combine data of 3 independent replicates of 20-30 larvae each. A Mann-Whitney test was used to analyze the overall bacterial burden of the pooled data of 3 independent replicates of 9 fish each, where data are shown as mean ± SEM. A Kolmogorov-Smirnov test was used to analyze the distribution of bacterial cluster sizes (ns P > 0.05).

RNA extraction, cDNA synthesis, and qPCR analysis
For every qPCR assay, a total of 3 biological samples ( A one-way ANOVA was used to test for significance and data are plotted as mean ± SEM (ns P > 0.05, *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001). an area of 500 µm from the cut toward the trunk were counted as recruited cells (Fig. 5F). For both the hindbrain injection and the tailamputation assays, a Kruskal-Wallis test was conducted to assess significance (*P ≤ 0.05, ***P ≤ 0.001, and ****P ≤ 0.0001) and data are shown as mean ± SEM.

Tracking of migrating macrophages
Time-lapse images of migrating macrophages from 2 independent replicates (5 larvae per genotype each) near the caudal hematopoietic tissue were acquired every 2 min for 3 h under basal conditions The tracking setting used were as follows: Log detector, estimated blob diameter = 20 microns, threshold diameter = 15 microns, and no further initial thresholding method was applied. The chosen view was hyperstack displayer and the tracking algorithm chosen was the simple LAP tracker, keeping the default settings. Tracks were later filtered according to the numbers of spots on track (>40 spots/track) and spots, links, and track statistics were used to estimate the mean speed of moving macrophages and the total displacement. The total displacement was manually calculated in Excel by adding all the links of a given track and a filter was applied to classify tracks with a maximum displacement <20 microns as static cells (mean speed = 0 and total displacement = 0). Data were analyzed with a one-way ANOVA (ns P > 0.05, *P ≤ 0.05, and **P ≤ 0.01) and are shown as mean ± SEM.

Macrophage circularity assessment
The cell circularity indexes (CIs) were calculated using the "analyze particle" option in the Fiji/ImageJ software. 52 The maximum projection images of migrating macrophages of the 3 genotypes were processed in Fiji/ImageJ by using the "despeckle" filter and by generating a binary image. In total, 30 macrophages per larvae were manually selected and the circularity of the cell in every frame was determined using the "analyze particle" option. A frequency histogram (%) for each group was plotted using cell CI bins as follows: 0-0.2, 0.2-0.4, 0.4-0.6, 0.6-0.8, and 0.8-1. The frequency distributions were analyzed using a Kolmogorov-Smirnov test taking the WT distribution as reference distribution (**P ≤ 0.01 and ****P ≤ 0.0001).
Whole larvae and tail areas were imaged with a Leica M165C stereofluorescence microscope and visualized with the LAS AF lite software.
Images were cropped in such way that the area encompassing the tail was always the same size (10.16 cm × 27.94 cm). The number and size of distal granulomas were analyzed with the "analyze particle" function in Fiji/ImageJ. 52 Particles with a total area >0.002 were considered as granulomas; smaller particles were excluded from our analysis. The percentage of infected larvae that developed distal granulomas was manually calculated and a 2 test was used to assess significance.
Larvae were put back into 2 ml egg water containing either DMSO or NBI 74330 after the amputation for 4 h followed by fixation with 4% PFA. Imaging of the tail region and quantification of macrophage recruitment to the tail-amputation area was done as described above.
For the bacterial burden assay, approximately 30 larvae of each group were pre-incubated either with 25 µM NBI74330 or 0.01% DMSO for 3 h before infection (24-27 hpf). Larvae were infected with 300 CFU M. marinum-wasabi at 28 hpf in the BI and NBI74330 and DMSO treatments were refreshed at 48 hpi. Imaging and bacterial pixel quantification were performed as described above.

Cxcr3.3 has features of both conventional Cxcr3 receptors and ACKRs
We have previously shown that zebrafish Cxcr3.   Table 2). (C and D) The whole predicted structure of the Cxcr3.2 and Cxcr3.3 receptors (a), the ligand binding site of both proteins (b) and the binding of one of the shared predicted ligands (0NN) by each receptor (c) among the groups. We also quantified macrophages in the head and tail because these were relevant areas for our experimental setups.
We observed that at day 4, cxcr3.2-/-larvae had transiently fewer cells in the head area (Fig. 2F). On the other hand, cxcr3.3 mutant embryos had more macrophages during the first 2 days but leveled off after this time point (Fig. 2G). Neutrophils were quantified in the same fashion as macrophages, using a Tg (mpx: eGFP) reporter line, but we did not detect any difference between the groups at any time point (Supplementary Fig. 1). Taking these observations into account, we performed all our experiments avoiding the time points at which macrophage development was inconsistent to prevent biased observations.

M. marinum infection burden, whereas overexpressing the gene lowers bacterial burden
We previously showed that mutation of cxcr3.2 enabled zebrafish larvae to better control M. marinum infection, a phenotype that could be explained by a reduction of macrophage migration in the absence of Cxcr3.2, which limits the dissemination of infection. 29 To investigate our hypothesis that Cxcr3.  The bacterial burden data were analyzed using a two-tailed t-test (A-C) and a one-way ANOVA (E). Results are shown as mean ± SEM (ns P > 0.05, * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001) and combine data of 3 independent replicates of 20-30 larvae each. The qPCR data were analyzed with the 2 -∆∆Ct method and a one-way ANOVA. Results are plotted as mean ± SEM (ns P > 0.05, * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001)  (Fig. 3G). On the other hand, cxcr3.3 expression was lower in cxcr3.2 mutant larvae and it was moderately induced upon infection (Fig. 3H). We also assessed the expression of the Cxcl11aa ligand, as it is the most up-regulated one out of the 7 Cxcl11-like ligands during M. marinum infection, in both cxcr3 mutants. 29,31 The gene was induced upon infection independently of the expression on cxcr3.2 and cxcr3.3 (Fig. 3I) Figure 4A shows no difference between WT and mutants regarding the total number of bacterial clusters in the tail area. We divided bacterial clusters into 3 groups according to the number of bacteria they contained: 1-5 bacteria (small cluster), 6-10 bacteria (medium-sized cluster), and >10 (large cluster) as shown in the representative image illustrating the cluster size categories in Fig. 4B. The frequency distributions of the 3 different cluster sizes in each genotype were compared and no significant difference was found (Fig. 4C). Mycobacterial clearance remained unaffected in the absence of Cxcr3.3, suggesting that the poor control of the infection in cxcr3.3 mutants is not due to a deficient bacterial clearance. As a positive control, we also ran this assay using DNA-damage regulated autophagy modulator 1 (dram1) mutant larvae, and their WT siblings, because dram1 mutants cannot efficiently clear M. marinum. 37 The total number of clusters was higher in dram1 mutants and large bacterial clusters were more frequent ( Supplementary Fig. 3). Therefore, we conclude that macrophages in cxcr3.3 mutants, contrary to dram1 mutants, are not affected in their microbicidal capacity.

Cxcr3.3-deficient macrophages show enhanced recruitment to sites of infection, toward Cxcl11aa, and to sites of injury
Several studies have shown that macrophage recruitment is essential for bacterial clearance and containment during mycobacterial pathogenesis but supports bacterial dissemination and granuloma formation at early stages of the infection. 55, 56 We previously found that cxcr3.2 mutant larvae showed an attenuated recruitment of macrophages to sites of infection and toward Cxcl11aa ligand. This study suggested that macrophage-mediated dissemination of bacteria was reduced due to this recruitment deficiency in cxcr3.2 mutants because fewer cells would become infected with M. marinum. 29 We addressed cell recruitment to examine whether the process was

Cxcr3.3 depletion has no significant effect on neutrophil recruitment to sites of infection or injury
Although macrophages are the first responders toward mycobacterial infection and the main components of granulomas, neutrophils are also recruited to infectious foci and participate in the early immune response. 57,58 Besides, both Cxcr3.2 and Cxcr3.3 are also expressed on this cell type. 29 Therefore, we assessed the effect of the cxcr3.2 and cxcr3.3 mutations on neutrophil recruitment to local M. marinum infection and upon injury similar as for macrophages in the previous section (Fig. 6). When M. marinum was locally injected into the hindbrain, fewer neutrophils were recruited to the cavity in cxcr3.2 mutants at 3 hpi, whereas there was no difference between WT and cxcr3.3 mutants (Figs. 6A and B). Using the tail-amputation model to assess cell recruitment, we observed the same pattern: the lack of cxcr3.2 reduced neutrophil recruitment to injury, whereas recruitment

Macrophages lacking Cxcr3.3 move faster than WT cells under basal conditions and upon mechanical damage, and have an elongated and branched morphology
We previously reported that macrophage recruitment to sites infection was attenuated in cxcr3.2 mutant macrophages because cells were less motile. 29  and D). Our data show that cxcr3.3 mutant larvae more frequently developed distal granulomas (22%) than the other 2 groups (Fig. 8A). In addition, the average number of the distal granulomas per fish within this group was higher (Fig. 8C) and the quantified structures were also larger (Fig. 8D). Consistent with previous work, 29 A and B). The tail fin of WT larvae and cxcr3.2 and cxcr3.3 mutants was amputated and neutrophil recruitment was assessed at 4 h postamputation. There were fewer recruited neutrophils in the cxcr3.2 mutants, whereas there was no difference between cxcr3.3 mutants and WT. The PBS-injected control group PBS combines WT, cxcr3.2, and cxcr3.3 mutants and shows no cell recruitment at 3 hpi. In all cases, statistical analyses were done with pooled data of 3 independent replicates (20-30 larvae per group each). A Kruskal-Wallis test was used to assess significance (ns P > 0.05, *** P ≤ 0.001, **** P ≤ 0.0001) and data are shown as mean ± SEM granulomas compared with the wild type (12.7%). Our data suggest that cxcr3.3 mutant macrophages favor bacterial dissemination and the seeding of secondary granulomas due to their enhanced recruitment to sites of infection and their higher speed.

Chemical inhibition of both Cxcr3 receptors affects only macrophages expressing Cxcr3.2 and phenocopies cxcr3.2 mutants regarding bacterial burden and macrophage recruitment efficiency
To further inquire into the roles and interactions of Cxcr3.

F I G U R E 7 Cxcr3.3-depleted macrophages move faster than WT cells under basal conditions and upon mechanical damage and have a lower CI.
Panel A shows representative images of tracks of macrophages of 3-day-old larvae from the 3 genotypes under unchallenged conditions (random patrolling). Macrophages were tracked for 3 h and images were taken every 2 min. Graphs in B show the total displacement of all cells tracked shortly after amputation in each group throughout 3 h (B-2) and the average speed of each cell (B-2). In this case, macrophages were tracked for 1.5 h and images were acquired every 1 min. There was no significant difference between the groups in terms of total cell displacement (B-1.), however cxcr3.3-/-macrophages did move faster than the remaining groups as indicated by the dot-plots in (B-2.). Panel C shows representative images of macrophage tracks of the 2 groups directly after a tail amputation. The tracks of cxcr3.2-/-macrophages were shorter than those of the remaining groups (D-1.) and cxcr3.3-/-macrophages moved faster than the other 2 groups when mechanical damage was inflicted (D-2.). Data of unchallenged larvae were collected from 2 independent replicates (5 larvae per group each) and for the tail-amputation model data from 3 independent replicates (4 larvae per group each) were pooled together for analysis (Continues) F I G U R E 7 (Continued) One-way ANOVA was performed to test for significance (ns P > 0.05, * P ≤ 0.05, ** P ≤ 0.01) and data are shown as mean ± 3 mutants developed more distal granulomas (22%) than WT (12.7%) and cxcr3.2 mutants (5%), whereas the latter developed fewer than the other 2 groups (A). Embryos from the 3 genotypes were infected at 28 hpf and imaged under the stereo fluorescence microscope (whole body and zoom to the tail) at 4 dpi. Panel B illustrates the imaging process of a representative cxcr3.3 mutant larvae. Cxcr3.3-depleted larvae developed more distal granulomas per fish (C) and these granulomas were also larger in cxcr3.3 mutants than the other 2 groups, whereas cxcr3.2 mutants showed an opposite trend (D). Graphs show pooled data from 4 independent replicates, each of 12-15 infected larvae per group. The number and size of distal granulomas were determined using the "analyze particle" function in Fiji. A 2 test was conducted to assess differences in the proportion of larvae that developed distal granulomas within the 3 groups and a one-way ANOVA to compare the number and size of distal granulomas (ns P > 0.05, * P ≤ 0.05, *** P ≤ 0.001 and **** P ≤ 0.001). Data are shown as mean ± SEM These results support our hypothesis that Cxcr3.2, an active GPCR, is essential for macrophage motility and recruitment to different stimuli, whereas Cxcr3.3, an ACKR with predicted scavenger function, does not play a direct role on these processes but indirectly regulates them by competing with active receptors for shared ligands.

DISCUSSION
Our At present, 5 ACKRs have been described in vertebrates, namely, ACKR1 (DARK), ACKR2, ACKR3 (CXCR7), ACKR4, and ACKR5. 12,18 The identification of ACKRs and the subsequent classification of these receptors within the subfamily is complex given their structural heterogeneity and the limited phylogenetic homology. 15,17,18 However, as in all GPCRs, the E/DRY motif and microswitch elements are indicative of the activation status and function of a receptor. 16 Microswitches stabilize the active conformation of GPCRs and are highly conserved residues, which are unchanged in Cxcr3.2 but not in Cxcr3.3. 13,16 The Asp (D) and the Arg (R) of the E/DRY-motif are key residues to stabilize the inactive conformation of GPCRs by forming a salt bridge between the third IC loop and TM6 that blocks G-protein coupling.
This so-called "iconic-lock" breaks upon binding of an agonist and triggers structural rearrangements that expose the G-protein docking site and enables canonical (G protein-dependent) downstream signaling. 16 Substitutions in the E/D and Y within the E/DRY-motif are commonly associated with the permanent activation of the receptor and gain of function events, whereas substitutions of the R, as found in Cxcr3.3 (DCY motif), have been shown to result in the permanent "locking" of the receptor and a consequent loss of function. 16,54,62 The E/DRY motif also interacts with the IC COOH terminus that is critical for GPCR activation and with G subunits. It is noteworthy to mention that chemokine receptors can also signal in a G protein-independent manner through -arrestin in the context of chemotaxis, and that this pathway is associated with the internalization and subsequent IC degradation of ligands. 16,62 Altogether, this information led us to propose that Cxcr3.3 is an ACKR.
The zebrafish genome encodes a family of 7 cxcl11-like paralogous genes, which are thought to share common ancestry with CXCL9-10-11, the ligands of human CXCR3. 29 We have previously shown that one of the cxcl11-like genes, cxcl11aa, is strongly inducible by mycobacterial infection and by mechanical damage. 29 Interestingly, although neutrophil recruitment was reduced in the cxcr3.2 mutant, it remained unaltered in cxcr3.3 mutants, suggesting that Cxcr3.3 has no effect on neutrophil migratory properties.
To examine whether altered cell motility was the underlying reason for enhanced recruitment in cxcr3.3 mutant macrophages, we used a tail-amputation assay to assess migration in terms of total cell displacement and average speed. We showed that cxcr3. CXCR3-A mediates chemotaxis and proliferation, whereas CXCR3-B inhibits cell migration and proliferation, and induces apoptosis. 67,68 Both splice forms can bind to CXCL9-11 chemokines but mediate opposite functions. While there are no splice variants of cxcr3.2 and cxcr3.3 in zebrafish, 69 the regulatory antagonism between the 2 paralogs resembles the interaction between the 2 human CXCR3 splice variants, which might suggest a form of convergent evolution. However, this functional diversification of CXCR3 variants is not conserved in the murine model, where CXCR3 is a single copy gene and no splice variants have been identified so far. 30,67 In conclusion, our work illustrates the antagonistic interaction between the 2 CXCR3 paralogs Cxcr3.2 and Cxcr3.3 in zebrafish. The structural analysis of Cxcr3.3 supports that this receptor is unable to signal in the conventional G protein-dependent way, but that it can still bind ligands and shape chemokine gradients, thereby regulating active receptors with shared ligands. Our experimental data show that the absence of the scavenging function of Cxcr3.3 is detrimental in the context of mycobacterial infection due to an exacerbated Cxcr3.2 signaling and a consequently enhanced macrophage motility that facilitates bacterial dissemination. However, we propose that enhanced macrophage motility could be benign in other contexts, such as tissue repair. Our findings suggest an extensive crosstalk among several chemokine signaling axes such as CXCR3-CXCL11 and CXCR4-ACKR3 (CXCR7), because ACKR3 also binds CXCL11 besides CXCL12. 18,26 Furthermore, ACKR1 a silent receptor that does not scavenge chemokines but redistributes them to mediate leukocyte extravasation and shares the CXCL11 and CXCL4 ligands with CXCR3. 70,71 The complexity of signaling axis integration further emphasizes the relevance of unraveling the fundamental mechanistic principles underlying intricate chemokine networks. Our findings contribute to understanding one such mechanistic interaction and suggest that a more comprehensive ACKR classification needs to be developed to aid the understanding of complex chemokine signaling regulation.