Orphan GPR84 facilitates uropod de-adhesion at multiple steps during leukocyte extravasation

The GPCR Tre1 is required for effective hemocyte extravasation from Drosophila pupal veins

We recently demonstrated that Drosophila pupal wings, in which hemocytes circulate through vessels (the wing “veins”), offer the opportunity to model immune cell extravasation in response to an inflammatory stimulus in the form of a small epithelial wound adjacent to a wing vein (Fig. 1A–D) (Thuma et al, 2018). Within this model, we can observe hemocytes circulating within the hemolymph pulsing through wing vessels, with vessel LV3 (that extends in one focal plane) being particularly amenable for imaging (Fig. 1B,C). Within 10 min of making a laser wound in the epithelium adjacent to this vessel, hemocytes are seen to probe the vessel wall in the vicinity of the injury before extravasating and subsequently migrating towards the wound (Fig. 1D).

Harnessing the live imaging opportunities of the translucent pupa, together with Drosophila’s genetic tractability, we sought novel genes that might drive immune cell extravasation. Here, extending our previous works (Thuma et al, 2018), and utilizing both a hemocyte-specific Tre1-RNAi knockdown (Tre1^KD) approach and a Tre1 null (Tre1^KO) mutant (in which the endogenous Tre1 locus has been removed) (Kim et al, 2021), our live imaging studies show Tre1 to be fundamental for complete hemocyte extravasation (Figs. 1C–H and  EV1A–C). An additional genetic recovery approach in which we re-express Tre1 specifically in hemocytes (using UAS-Tre1 driven by srp-Gal4) (Chen et al, 2024) within a Tre1 null background confirmed a role for hemocyte Tre1 during extravasation (Figs. 1G,H and  EV1D). Intriguingly, within the wing veins, hemocytes lacking Tre1 showed an intravenous morphological phenotype even in unwounded conditions; Tre1-deficient hemocytes exhibited cytoplasmic extensions that appeared stuck to the luminal side of the vein cells, resulting in reduced circularity and luminal adherence (Figs. 1C,F and  EV1E), a sustained phenomenon that was evident for the duration of the imaging period. Given that hemocyte-specific Tre1 expression recovered the circularity and extravasation defects of the Tre1 null, our data suggest an autonomous role for Tre1 within hemocytes.

Following wounding, Tre1 null hemocytes showed an extreme phenotype, with fewer hemocytes fully extravasating and those that did fully extravasate, taking longer than that of controls or RNAi-based Tre1 knockdowns (Figs. 1D–H and EV1B–D; Movie EV1). These striking extravasation defects occurred despite no initial differences in hemocyte recruitment (Fig. EV1F). The tails of Tre1-deficient hemocytes remained adherent to the vein walls, suggesting that Tre1 may regulate the perivascular detachment phase of hemocyte extravasation (Fig. 1E). We next undertook correlative light and transmission electron microscope imaging (CLEM) studies to compare hemocyte morphologies post wounding in control and Tre1-deficient pupae (Fig. 1I–N). Time-lapse movies enabled the capture of hemocytes in the act of extravasating in vivo (Fig. 1I,J), and these samples were immediately prepared to visualize this process at the ultrastructural level via transmission electron microscopy (Fig. 1K–N). In control unwounded pupae, we observed hemocytes within the lumen of the wing vein (Fig. 1L); following wounding, hemocytes were observed outside of the vessel exhibiting relatively short lamellipodia en route to the wound (Fig. 1M). Tre1-deficient hemocytes, however, showed long leading extensions and retained multiple lagging tethers linking them to the vein wall cells and to the interior of the vein (Fig. 1Nⁱ,Nⁱⁱ). Consistent with our live-imaging studies, these CLEM data suggest that hemocytes are disrupted in their capacity to extravasate completely through wing vessels after a sterile injury, potentially due to an incapacity to de-adhere from vessel cells.

Tre1 shares structural homology with its conserved mammalian ortholog, GPR84

To assess whether our Drosophila model could identify extravasation players with conserved roles in mammals, we explored whether the human and mouse genomes possess Tre1 orthologs. Phylogenetic analysis of amino acid sequences revealed several Drosophila paralogues and mammalian orthologs of Tre1 (Fig. 2A), including GPR84 with 34% amino acid sequence homology. Neither Tre1 nor its mammalian orthologs have experimentally derived structures, but AlphaFold enabled us to compare their predicted structures (Fig. 2B; Movie EV2) (Jumper et al, 2021). This in silico comparison suggested high structural homology between Tre1 and human GPR84 with significant identical elements (yellow, Fig. 2Bⁱⁱ), suggesting likely shared functionality; less structural homology was observed between Tre1 and an unrelated Class A GPCR, Drosophila Rhodopsin (yellow, Fig. 2Bⁱ).

GPR84 mediates neutrophil extravasation in a murine model of acute inflammation

Extending the Drosophila works, we sought to investigate the functional role of the predicted mammalian ortholog of Tre1, GPR84. In initial studies, aiming to unravel the stage of immune cell migration mediated by GPR84, we analyzed immune cell behavior of WT and Gpr84^−/− mice in control (PBS) or acutely- inflamed mouse cremaster muscles by brightfield intravital microscopy (IVM) (Fig. 3A–D; Movie EV3). Here, local TNF (4 h) induced a significant and comparable level of intraluminal leukocyte adhesion in both WT and Gpr84^−/− mice, as compared to PBS-treated tissues (Fig. 3C). In contrast, TNF-induced leukocyte extravasation was significantly reduced in Gpr84^−/− mice as compared to WT littermates (Fig. 3D). Furthermore, chimeric animals with selective myeloid GPR84 deficiency (generated through transfer of bone marrow cells to irradiated recipients; Fig. 3E), showed defective neutrophil extravasation (Fig. 3F,G). Of note, we detected no differences in leukocyte adhesion or extravasation, nor any distinct morphological differences in PBS-treated WT or KO tissues. Furthermore, we saw no difference in the total or differential leukocyte counts in the circulating blood of WT and KO mice under basal conditions Fig. EV2A), and the chimeric mouse lines showed equal reconstitution efficiency (Fig. EV2B,C). Collectively, the present findings illustrate a specific and cell-autonomous role for GPR84 in regulating the passage of leukocytes through inflamed venular walls.

GPR84-deficient neutrophils display excessive adhesive interactions within inflamed vessels

Having identified defective immune cell extravasation in both Tre1-deficient Drosophila and GPR84-deficient mice, we next sought to gain more insight into the mechanism of action of GPR84 in real-time and at the single-cell level using high-resolution 4D confocal IVM. For this purpose, we interbred Gpr84^−/− mice with the compound mouse reporter strain Lyz2-EGFP-ki; Acta2-RFPcherry-Tg that exhibits endogenous GFP^high neutrophils and RFP⁺ pericytes/smooth muscle cells. For imaging ECs, or more specifically EC junctions, the mice received a local injection of non-blocking fluorescently labeled anti-PECAM-1 mAb, as previously described (Woodfin et al, 2011). The application of this platform to exteriorized TNF-stimulated cremaster muscles revealed aberrant neutrophil behaviors in GPR84 KO mice at multiple stages of neutrophil extravasation, all reminiscent of defective de-adhesion. Quantification of luminal responses (Fig. 4A) revealed significantly slower crawling of KO neutrophils within inflamed postcapillary venules, as compared to the motility of WT neutrophils (Fig. 4B). This was associated with the detection of an increased number of neutrophil fragments in close apposition to sites of neutrophil adhesion in KO mice, as compared to WTs (Fig.  4C,Dⁱ–Dⁱⁱ). This response may allude to excessive neutrophil adhesion and difficulties with detachment from the endothelium, as further indicated by excessive tethering of Gpr84^−/− neutrophils to the endothelium (Fig. 4Dⁱⁱⁱ). With respect to the latter, despite no significant difference in the absolute number of neutrophil tethers detected between WT and KO mice (0.186 ± 0.057 and 0.166 ± 0.047 tethers/intraluminal neutrophils over 1 h imaging period, respectively), the profiles detected between the two mouse lines were different. Specifically, 35% of WT neutrophils, as compared to 43% of Gpr84^−/− neutrophils, displayed tethers that were elongated by at least one cell body (≥ ~10 µm) in length. These observations are similar to the aberrant intraluminal structures detected in Tre1 null Drosophila hemocytes (Fig. 1C,F).

In analyzing the frequency and dynamics of neutrophil transendothelial cell migration (TEM) (Fig. 4E), GPR84-deficient neutrophils displayed a significantly reduced number of TEM events (Fig. 4F) and consistently showed an extended duration to fully breach the endothelium (Fig. 4G). Post breaching of the endothelium, KO mice subsequently presented multiple aberrant interactions between neutrophils and pericytes (Fig. 4H). Specifically, despite no overall differences in the number of neutrophils breaching the pericyte layer (termed trans-pericyte migration; TPM) in WT and KO mice (Fig. 4I), we noted an increase in the number of neutrophils that probed but ultimately “aborted” their migration across the pericyte layer (Fig. 4J). Furthermore, KO neutrophils exhibited a distinct hesitancy and increased oscillatory behavior, characterized by a repetitive probing and retracting of the neutrophil cell body, prior to eventual completion of their migration across the pericyte sheath (Fig. 4K). In many instances, this wavering neutrophil-pericyte interaction was observed throughout the whole duration of the imaging period in GPR84 KO mice, and this step ultimately failed (see Movie EV4). Interestingly, similar to the defective intraluminal responses, post breaching the pericyte layer, a subset of Gpr84^−/− neutrophils again presented elongated tail-like structures within the sub-EC space (Fig. 4L). The latter were commonly associated with evidence of neutrophil fragment deposition, supporting the notion of a generally impaired leukocyte de-adhesion within inflamed vessel walls of KO mice.

GPR84-deficient neutrophils exhibit defective detachment from inflamed vessels

Guided by the striking phenotype of elongated tails observed in Tre1-deficient hemocytes (Fig. 1), we extended our murine confocal IVM studies to analyze neutrophil detachment from inflamed venules in the perivascular space (Fig. 5A). As a quantitative assessment of neutrophil morphology, we measured the maximum length of perivascular neutrophils during the imaging period (Fig. 5B). This analysis revealed that in GPR84-deficient mice, perivascular neutrophils present two principal profiles, (i) a population exhibiting a “spherical” morphology that shows almost total perivascular arrest, and (ii) a population with an extended and persistent uropod formation, reaching >50 µm in length in some cells. The resultant effect was a delayed detachment of KO neutrophils from inflamed venular walls. Hence, in entering the perivascular space, while WT neutrophils displayed brief attachments (∼12 min) to the exit site of the venular wall, in Gpr84^−/− mice, ∼80% of the perivascular neutrophils required >30 min to fully detach and enter the surrounding interstitial space (Fig. 5C; Movie EV5). The aberrant GPR84 KO neutrophil tails (Fig. 5D) were probed further at the ultrastructural level by transmission electron microscopy (Fig. 5E,F). Whereas, in WT mice we saw fully detached neutrophils with short uropods (as if recently detached from adjacent pericytes) (Fig. 5E), in Gpr84^−/− mice we observed extensive lagging tails that appeared tethered to the pericyte wall (Fig. 5Fⁱ). The latter were often associated with fibrillar matrix (Fig. 5Fⁱⁱ–Fⁱⁱ). Collectively, the findings suggest exaggerated and prolonged leukocyte adhesion at multiple sites within and along the venular wall.

Transcriptomic analysis of GPR84-deficient murine neutrophils indicates defective actin cytoskeletal regulation and cell adhesion

With our live imaging studies strongly suggesting a role for Tre1/GPR84 in leukocyte detachment during extravasation, we next wished to gain insight into the likely signaling machinery involved. For this, we undertook an unbiased transcriptomic analysis of circulating WT and GPR84-deficient neutrophils. Principal Component Analysis (PCA) of data acquired from sorted blood neutrophils (isolated from untreated mice) showed a separation between WT and KO neutrophils (Fig. 6A). This revealed 713 differentially expressed genes (DEGs), with 367 significantly upregulated and 346 significantly downregulated in GPR84-deficient neutrophils relative to WT cells (Fig. 6B). Gene set enrichment analysis indicated dysregulated cytoskeletal remodeling and cell adhesion properties in GPR84-deficient neutrophils (Fig. 6C,Dⁱ–Dⁱⁱ), consistent with the observed phenotype of impaired detachment from the venular wall. Specifically, numerous candidates related to cytoskeletal remodeling were downregulated in GPR84-deficient neutrophils (Fig. 6Dⁱ), including Pak1, Arhgap5, Ccdc125—all considered essential effectors of Rho-GTPases (Shin et al, 2013; Zhao and Manser, 2005). Interestingly, GPR84-deficient neutrophils also upregulated Rho regulators such as Pip5k1, a kinase implicated in RhoA polarization and uropod formation in neutrophils (Xu et al, 2010).

Several DEGs indicative of altered cell adhesion, most notably integrins involved in cell-cell attachment and cell-ECM adhesion, were also identified (Fig. 6Dⁱⁱ). Specific examples that were downregulated in KO neutrophils include Itga5 that encodes for the α subunit of the integrin fibronectin receptor α5β1 (VLA-5) and Itgb3 that encodes for the β3 integrin subunit. The latter is a component of multiple cell surface receptors that play roles in cell adhesion, migration, and signaling, and participates in RGD-dependent binding to ECM ligands (Chastney et al, 2025). Whilst suggesting defective adhesion of KO neutrophils to venular cells and basement membrane structures, we found no significant difference in the cell surface protein expression of key integrin subunits in KO vs WT neutrophils (Fig. EV3). The latter may be due to potential regulation of transcripts by negative feedback to maintain appropriate protein expression levels. KO neutrophils also showed altered expression of other genes implicated in cell migration, such as Ly6g, a downregulated candidate implicated in neutrophil infiltration and considered to act as a modulator of integrin functions (Song et al, 2025; Wang et al, 2012). Interestingly, KO cells displayed upregulation of Frem1 that encodes for the generation of extracellular matrix proteins involved in the formation of basement membranes (Smyth et al, 2004). Though not previously linked to neutrophils, its upregulation may be a cellular compensatory mechanism aimed at normalizing impaired integrin-ECM interactions. Collectively, these data suggest that Tre1/GPR84 might regulate leukocyte extravasation via Rho-dependent cytoskeletal rearrangement and regulation of integrin adhesion dynamics.

Drosophila Tre1 supports localized Rho1 activity during hemocyte extravasation

Based on our live in vivo imaging and transcriptomic data, we speculated that Tre1/GPR84 may regulate detachment of hemocytes/neutrophils from the vessel wall by Rho-mediated events. Tre1 contains two highly conserved sequences (the E/N/DRY and NPxxY motifs), known to act via distinct signaling pathways (G-protein signaling and Rho1/E-cadherin, respectively) for different elements of the spatio-temporal guidance of germ cells (LeBlanc and Lehmann, 2017). A sequence comparison of these motifs between Drosophila, mouse and human highlighted that the NPxxY (so called Rho binding domain), is fully conserved between Tre1 and mammalian GPR84, whereas the E/N/DRY is not (Fig. 7A). Drosophila studies of migrating germ cells have shown the NPxxY motif is required for downstream Rho1 activation and that Tre1^NPxxY mutant germ cells demonstrate aberrant migration (Kim et al, 2021; LeBlanc and Lehmann, 2017). Our in silico AlphaFold3-mediated analysis (Abramson et al, 2024) of the predicted interface between Drosophila Rho1 and Tre1 suggested that Rho guanine nucleotide exchange factor 2 (RhoGEF2) and G protein alpha subunit o (Gαo) might act as intermediate accessory proteins to the stable structure (Fig. 7B). Consistent with this, Gαo has been reported to directly bind to and act downstream of Tre1 during Drosophila neuroblast cell polarization (Yoshiura et al, 2012); Gαo also acts downstream of Tre1’s closely-related paralogue Moody to help stabilize the extended morphology of the surface glia in the Drosophila blood brain barrier (Schwabe et al, 2005). RhoGEF2 is an established activator of Rho1 that acts by facilitating the switch of GDP to GTP (Mosaddeghzadeh and Ahmadian, 2021); previous work has also suggested a connection between Tre1 and RhoGEF2 during the migration of the Drosophila primordial germ cells (Lin et al, 2022). Since we found hemocytes and neutrophils lacking Tre1/GPR84 to exhibit similar morphological phenotypes of lagging tails and transcriptomic alteration of Rho family members, we hypothesized that these GPCRs may signal through their conserved NPxxY motifs. Consequently, we considered that the latter might then activate Rho1 via RhoGEF2 and Gαo to facilitate tail retraction/detachment during the latter stages of extravasation (Fig. 7C).

To address this proposed mechanism, we utilized high-resolution in vivo imaging of extravasating Drosophila hemocytes in response to a laser wound. Independent analysis of Gαo-RNAi and RhoGEF2-RNAi knockdown hemocytes revealed both proteins to be essential for timely and proper hemocyte extravasation from the vasculature. Indeed, Gαo-RNAi and RhoGEF2-RNAi knockdown phenocopied Tre1 mutants with hemocytes displaying de-adhesion defects (Figs. 7D–G and EV4A,B; Movie EV6). Expression analyses confirmed Gαo-RNAi caused a significant reduction in Gαo transcripts within hemocytes (Fig. EV4C–E). Utilizing a transgenic fluorescent biosensor for in vivo Rho1 activity that consists of GFP tagged to the Rho1 GTP-binding domain of Anillin (Munjal et al, 2015), we visualized the dynamics of Rho1 activity in vivo during extravasation (Fig. 7H–N; Movie EV7). As control hemocytes extravasated from vessels, we observed clear punctae of active Rho1 localized to the hemocyte tail (preceding tail retraction) that dissolved after the tail had detached (Fig. 7I,J). In Tre1-deficient hemocytes, active Rho1 was mostly absent from the tail, which remained adherent for longer and often failed to fully detach from the vein (Fig. 7K,L). Quantification of Rho1 activity (duration and area of “active Rho1” fluorescence) confirmed that Tre1-null hemocytes have smaller and more transient flashes of localized Rho1 activity, even on the rare occasions when they occurred (Fig. 7M,N). Similarly, using a RhoA G-LISA (G-protein Linked Immunosorbent Assay), GPR84 KO bone-marrow-derived mouse neutrophils exhibited reduced GTP-bound RhoA following stimulation with the potent GPR84 agonist L237, as compared to WT cells (Fig. 7O). Notably, within this assay, both WT and KO neutrophils responded equally to the synthetic FPR activator, WKYMVm, showing that GPR84-deficient mouse neutrophils have no intrinsic abnormalities in RhoA activation. Together, our findings demonstrate defective Rho1/RhoA activity in Tre1/GPR84-deficient immune cells in Drosophila and mice (Fig. 7).

Leukocyte Tre-1/GPR84 modulate integrin-mediated adhesion

A key finding of the present study is that hemocytes and neutrophils lacking Tre1/GPR84 have long lagging tails that can persist for hours after they have attempted to leave the circulation. In control mammalian leukocytes, uropod adhesion and then retraction result from the accumulation of integrin-to-ECM focal adhesions that dissolve through a combination of contractile and biochemical forces (Hind et al, 2016). We speculated that Tre1/GPR84 facilitates tail retraction through the regulation of cytoskeletal contraction and dissolution of integrin focal adhesions mediated by Rho1/A activation. This notion is supported by previous in vitro studies showing that pharmacological blockade of RhoA leads to abnormally elongated tails in monocytes adherent to extracellular matrix (Worthylake et al, 2001), an observation that resembles Tre1/GPR84 null immune cells attempting to extravasate in vivo.

In hypothesizing that activated Tre1 dissolves integrin adhesions in the hemocyte tail, we predicted that elevating hemocyte integrin levels would phenocopy Tre1-deficient hemocytes. Indeed, we observed that overexpression of myospheroid (mys; the predominant Drosophila β-integrin) specifically in hemocytes led to a strikingly similar defective extravasation phenotype (Fig. 8A–D) despite not impacting the number of extravascular hemocytes pre-wound (Fig. EV5A–C). To further address whether Tre1/GPR84 regulates focal adhesion dissolution in lagging tails as leukocytes extravasate, we examined whether mys levels were perturbed in Drosophila lacking Tre1. We found that in Tre1 null pupae, mys mRNA levels were significantly reduced, as were transcripts for focal adhesion kinase (Fak), a known regulator of integrins (Fig. 8E,F). These data are consistent with the transcriptomic analysis of GPR84 KO mouse neutrophils, suggesting a conserved role for Tre1/GPR84 in modulating integrin-based adhesion (Fig. 6C), and our observations of defective adhesion of GPR84 mouse KO neutrophils to intercellular adhesion molecule 1 (ICAM-1)-coated plates (Fig. 8G–I). ICAM-1 is highly expressed on TNF-stimulated endothelial cells and pericytes (Proebstl et al, 2012; Cooper et al, 2024; Recio et al, 2018; Yousefi et al, 2001; Bouchard et al, 2007) and was thus used as an in vitro adhesive substrate to functionally mimic neutrophil interactions with vascular cells. Following stimulation with the chemokine CXCL1, whilst the average surface area of WT and KO neutrophils was not significantly different (64.42 µm² ± 7.42 and 53.88 µm² ± 6.74, WT and KO neutrophils, respectively; n = 283 WT and n = 186 KO neutrophils, from n = 4 mice, P = 0.2582), the frequency of Gpr84^−/− neutrophils that displayed a “spreading” morphology was significantly reduced (Fig. 8G,H). This was associated with an overall defective motility in vitro (Fig. 8I) and observed reduced membrane ruffling and polarization, compared to controls (Fig. 8G), suggesting an impairment in integrin-induced cytoskeletal remodeling in response to activation.

Collectively, the data acquired from our Drosophila and mouse studies, using both in vitro and in vivo assays, strongly support the concept that Tre1/GPR84 exhibits conserved functionalities in supporting Rho-mediated regulation of integrin activity.

Pharmacological blockade of GPR84 suppresses neutrophil extravasation in acute inflammation

Given the conserved role for GPR84/Tre1 in supporting effective immune cell extravasation, in a final series of experiments, we explored the effects of pharmacologically modulating murine GPR84 during acute inflammation. For this purpose, we tested the efficacy of a selective and orally active functional GPR84 antagonist (GLPG1205) in multiple murine models of acute inflammation. Initially, using our TNF-inflamed cremaster muscle model, a reaction in which the GPR84 KO mice exhibited ~70% inhibition of leukocyte migration (Fig. 3D), we found that treatment of mice with the antagonist led to a dose-dependent suppression of neutrophil extravasation (Fig. 9A,B). Using the higher dose of the antagonist (30 mg/kg), these findings were extended to three other inflammatory settings, namely a zymosan-induced peritonitis model, a cutaneous ischemia–reperfusion (IR) injury model, and finally a skin wounding assay. In the peritonitis model, a well-characterized self-resolving acute inflammatory reaction, 24 h post injection of zymosan, GLPG1205 pre-treated mice exhibited reduced peritoneal cavity infiltration of both neutrophils and monocytes (Fig. 9C). These results suggest that GPR84 can mediate appropriate migration of multiple populations of myeloid cells. The cutaneous IR injury model, employed as a model of a pressure ulcer, similarly showed an effective neutrophil infiltration response in control mice that was significantly suppressed in mice treated with the GPR84 antagonist (Fig. 9D). The latter studies included both male and female mice, demonstrating that the efficacy of GPR84 blockade was not sex specific.

GPR84 expression increases in murine myeloid cells (neutrophils, macrophages, and monocytes) following skin wounding (Cooper et al, 2024). To investigate how wound inflammation might be impacted by GPR84 signaling, we generated punch biopsy skin wounds in GLPG1205-treated versus control adult male and female mice (Fig. 9E–H). Standardly, the initial influx of neutrophils to such a skin wound begins from 6 h post wounding, with numbers peaking at 24 h and diminishing back to basal levels after 48 h, as macrophage numbers begin to predominate in the wound. To capture potential differences in neutrophil influx in treated versus control mice, we harvested wounds at 18 and 36 h and analyzed wound sections stained with Hematoxylin and Eosin (Fig. 9E,F); here, the automated software QuPath (Bankhead et al, 2017) enabled efficient quantification of Hematoxylin-stained neutrophil nuclei within Eosin-stained cytoplasms (inset, Fig. 9E). Our data show a clear dampening of neutrophil influx into the wound bed at both timepoints in female (Fig. 9G) and male (Fig. 9H) mice.

Collectively, using a pharmacological approach, these results demonstrate the importance of GPR84 in regulating leukocyte extravasation in several inflammatory contexts in different organs, in response to different stimuli, and in both male and female animals.

Drosophila stocks and husbandry

Fly stocks were maintained according to standard protocols. All crosses were performed at 25 °C unless otherwise stated. The following Drosophila stocks were used: ubi-GMA, srp-gal4 (Brückner et al, 2004), UAS-mcherry, srp-mcherry (gift from Daria Siekhaus), Act-Gal4, UAS-GMA, w;;Tre1-RNAi P{TRiP.HMS00599} (Bloomington Drosophila Stock Center #33718), Oregon R, AniRBD (a gift from T. Lecuit (Munjal et al, 2015)), w;Tre1^KO (a gift from D.J. Andrew (Kim et al, 2021)), w;;UAS-mys (Bloomington Drosophila Stock Center #68158), w;Gαo-RNAi P{TRiP.hms01129} (Bloomington Drosophila Stock Center #34653). Of note, in all studies conducted in Drosophila¸ all driver and reporter lines were out-crossed to Oregon R (wild-type) and the offspring used as a comparative baseline (control) against mutants.

Drosophila dissections, wounding, and imaging

Larvae were sexed, and male pre-pupae were collected 75 h prior to wounding and kept at 25 °C to ensure consistent staging. 75 h after puparium formation (APF), pupae were then placed on double-sided tape ventral side down, and forceps were used to remove the pupa. For imaging, dissected pupae were placed left side down in a glass-bottomed microscopy dish (Greiner Bio-One GmbH, 627861). Wounds were induced using a Micro Point ablation laser (Andor Technology Oxford Instruments) either attached to an Olympus BX53 upright microscope or a Leica SP8 confocal microscope.

Identifying Tre1 orthologs, multiple sequence alignment, and phylogenetic tree construction

Tre1 paralogues and orthologs in Drosophila, mice, and humans were identified using DIOPT, with those sequences having a weighted score >5 selected for further analysis (Hu et al, 2011). Drosophila protein sequences were retrieved from FlyBase, and human and mouse protein sequences from the NCBI protein database. The longest isoform was selected if there were multiple isoforms per gene. Multiple sequence alignment was performed with Clustal Omega, and a phylogenetic tree was constructed with JalView. Structural alignment and visualization of orthologous and non-homologous protein structures were performed using UCSF ChimeraX (Pettersen et al, 2021).

RNA isolation, reverse transcription, and real-time quantitative PCR

RNA was isolated from WT and Tre1^KO;; 75 h APF pupae by crushing in TRIzol (Life Technologies) and RNA purified using an RNeasy Mini Kit (Qiagen). Equal quantities of RNA were then reverse transcribed using a Thermo Scientific Maxima First Strand cDNA Synthesis Kit, and genomic DNA was eliminated using double-strand-specific DNase (Thermo Scientific). Relative quantification of gene expression was performed using SYBR Green Supermix with a real-time PCR machine. Target gene expression was normalized to the expression of the housekeeping reference gene Gapdh1 using the ΔΔCt analysis method. The following primers were used in this study: Tre1 F-primer 5′-ATTAGTGCCTGTGTCTTTGTGAC-3′ and R-primer 5′-GGAGATGCTTAGCGAAATGACG-3′, mys F-primer 5′-CGCGGTGCTACCAAAACAC-3′ and R-primer 5′-GAATCTGCTCAACTGTTATCGGA-3′, Fak F-primer 5′-GCTGACCGATGATTATGC-3′ and R-primer 5′-CGAACGGTGGGCGTAGAGTAG-3′, and Gapdh1 F-primer 5′-TAAATTCGACTCGACTCACGGT-3′ and R-primer 5′-CTCCACCACATACTCGGCTC-3′.

Hybridization chain reaction (HCR) on pupae

HCR fluorescence in situ hybridization (HCR-FISH) was performed on whole Drosophila 75 h APF pupae using the HCR v3.0 system (Molecular Instruments, USA) and following the protocol of Rose et al (Rose et al, 2022) with minor modifications. Briefly, whole pupae were pierced near the wing hinge to improve permeability of tissue and fixed in 4% formaldehyde for 30 min at RT, followed by methanol series to dehydrate and rehydrate samples. Pupae were then incubated in proteinase K (30 µg/ml) for 30 min at RT, followed by another fixation with 4% formaldehyde to help preserve tissue integrity before proceeding to hybridization. Samples were incubated overnight at 37 °C with HCR panel targeting Gαo (X3) and srp (X1) (1:50 dilution). Then, the samples were treated for 24 h at RT with HCR hairpin amplifiers X1-488 and X3-647 (1:50 dilution). Finally, samples were mounted in antifade mounting medium containing DAPI (Vector Laboratories, USA) and imaged on a Leica SP8 confocal microscope.

Correlative light electron microscopy on pupae

For CLEM studies, confocal images were acquired as described, and wing hinges were punctured with forceps before fixation in freshly prepared 2.5% glutaraldehyde (Thermo Fisher Scientific) in 0.1 M cacodylate buffer (Thermo Fisher Scientific) for 24 h at 4 °C. Pupae were then washed in 0.1 M sodium cacodylate (three times for 10 min) before being placed in a secondary fixative containing 1% OsO₄, 1.5% potassium ferrocyanide, and 0.1 M sodium cacodylate for 2 h at 4 °C. They were again washed in 0.1 M sodium cacodylate (three times for 10 min) and wings dissected and embedded in a solution of 13.5% BSA, 0.1 M cacodylate, and 5% glutaraldehyde for 10 min. Wings were then placed in an automatic tissue processor (Leica EM TP). In all, 70 nm sections were cut with a UC6 ultramicrotome (Leica) with Diamond Knife (Diatome), and were collected onto pioloform-coated copper slot grids (Agar Scientific). These sections were then imaged using a Tecnai 12 BioTwin Spirit TEM (tungsten filament, 120 kV) and a FEI Eagle 4k×4k CCD camera.

Murine husbandry

All transgenic mice used were on a C57BL/6 N background. Gpr84^−/− mice (and WT littermate controls) were derived from cryopreserved sperm (provided by GlaxoSmithKline) at MRC Harwell. GPR84 (GenBank accession number NM_030720, Ensembl identification number ENSMUSG00000063234) KO mice were provided by Deltagen under a GSK license agreement. As summarized in (Nicol et al, 2015), the insertion of an IRES–LacZ–poly(A) expression cassette replaced 257 bp of coding sequence to generate a premature stop codon, thus deleting in the first three transmembrane domains. Heterozygote breeding pairs were used to maintain the colony at Queen Mary University of London.

For confocal intravital microscopy studies, myeloid cells and pericytes were visualized by generating a Lyz2-EGFP-ki; Acta2-RFPcherry-Tg colony. The Lyz2-EGFP-ki mice, summarized in (Faust et al, 2000), were crossed with Acta2-RFPcherry-Tg mice in which insertion of the RFP variant mCherry under control of the Acta2 promoter to generate RFP+ pericytes and smooth muscle cells (Proebstl et al, 2012). Lyz2-EGFP-ki; Acta2-RFPcherry-Tg mice were crossed with Gpr84^+/− (heterozygote) mice across several generations to generate WT and Gpr84^−/− mice on the Lyz2-EGFP-ki; Acta2-RFPcherry-Tg background.

For the chimeric studies detailed below, Lyz2-EGFP-ki mice were used as bone-marrow donors to C57Bl6/J recipient mice with a CD45.1 allelic variant, which were a gift from Dr Suchita Nadkarni (William Harvey Institute, Queen Mary University London).

All animals were used between 8 and 20 weeks old, and all experimental groups were age-matched. For all pharmacological blockade studies, wild-type C57/Bl6J mice were purchased from Charles River. Male mice were used for all cremaster and peritonitis studies; both males and females were used for ear IR and cutaneous wound healing experiments.

All animals were group housed in individually ventilated cages under specific pathogen-free (SPF) conditions and a 12 h light–dark cycle. Animals were humanely sacrificed via cervical dislocation or exsanguination at the end of experiments in accordance with UK Home Office regulations. All in vivo experiments were conducted at the William Harvey Research Institute, Queen Mary University of London, UK, under the UK legislation for animal experimentation and in agreement with the UK Home Office Animals Scientific Procedure Act 1986 (ASPA).

Unless constrained by evident phenotypic alterations (e.g., leukocyte recruitment solely observed in stimulated animals but not in control subjects), all experiments involved steps in which experimenters were blinded. All sample sizes were determined based on previous studies, and the order of GLPG1205 treatment was randomized across batches.

Brightfield intravital microscopy and analysis

Cremaster muscles were inflamed with an intrascrotal (i.s.) injection of TNF (300 ng) or vehicle control in a 400 µl bolus of PBS. The mice were administered an i.p. injection of ketamine (100 mg/kg) and xylazine (10 mg/kg) 1.5 h prior to exteriorization of the cremaster muscle onto an imaging stage. The cremaster muscles were continuously superfused in warm Tyrode’s solution, and the animals were maintained at 37 °C throughout image acquisition.

Brightfield intravital microscopy was used as a preliminary assessment of real-time leukocyte dynamics. Firmly adhered leukocytes were considered those that remained stationary on the luminal endothelium for ≥30 s. The number of extravasated leukocytes within 50 µm of either side of a 100 µm post-capillary venule segment was quantified at the 4 h experimental end point. Post-capillary venules were observed and movies captured using a ×63 water dipping objective on a transmitted light upright fixed stage microscope (Axioskop FS,Carl Zeiss) with a digital CMOS camera (Hamamatsu).

Generation of bone marrow chimeric animals

To generate chimeric mice, C57Bl6/J mice (on a CD45.1 allelic background) were irradiated with 4 Gy (3 min 25 s) two times, 4 h apart. The day after irradiation, ∼1.5 × 10⁶ bone marrow-derived cells were administered to each mouse by i.v. injection. Four weeks after irradiation, myeloid cell reconstitution efficiency was assessed by whole blood collection and flow cytometry to calculate absolute leukocyte counts prior to induction of inflammation in the cremaster muscles (all mice showed an effective reconstitution efficiency, as shown in Fig. EV2). After 4 h, the animals were sacrificed, the cremaster muscles were exteriorized and fixed prior to permeabilization and immunofluorescence staining.

Inflammatory response in the cremaster muscles

Mice were briefly anesthetized with 3% isofluorane and injected i.s. with TNF (300 ng) (R&D Systems) or phosphate-buffered saline (PBS) as a vehicle control, together with fluorescently labeled non-blocking anti-CD31 mAb (4 µg, clone 390, Thermo Fischer Scientific) to label endothelial cell junctions, in a 400 µl bolus using a 2–5 h stimulation period.

Whole-mount immunofluorescence staining

Exteriorized cremaster or ear tissues were fixed for 1 h in 4% paraformaldehyde (PFA) washed in PBS and permeabilised for 3 or 5 h, respectively, in PBS containing 25% fetal calf serum (FCS) and 0.5% Triton-X. Tissues were stained with fluorescently labeled monoclonal antibodies, MRP-14 (a gift from Professor Nancy Hogg) or PECAM-1, overnight in PBS containing 10% FCS at 4 °C. Antibody conjugation to AF488, 647, or 405 was performed using kits according to the manufacturer’s instructions (ThermoFisher Scientific).

Confocal microscopy and image analysis

Immunostained whole-mount cremaster muscles or ears were imaged with an upright Leica TCS SP8 (Leica) (with a white light laser (tuneable 470–670 nm)) or an inverted LSM800 (Carl Zeiss) confocal laser scanning microscope equipped with solid-state laser diodes (405, 488, 561, and 640 nm excitation wavelengths), respectively. For high-resolution confocal imaging of neutrophil-vessel wall structures, either the LSM800 (Carl Zeiss) or a CSU-W1 SoRa Spinning Disk microscope (Nikon) was used prior to deconvolution with either Huygens 25.10 (Scientific Volume Imaging) or NIS version 6.10.02 (Nikon), respectively.

For all quantified images, Z-stacks of half post-capillary venules, in a longitudinal orientation, were collected. Post-capillary venules 20–45 µm in diameter were selected for image acquisition, and images were collected at 1024 × 1024 pixels. In cremaster tissues, the number of extravasated neutrophils was quantified per field of view (typically 330 × 160 × 45 µm). For IR injury experiments in the ear (described below), the number of extravasated neutrophils within 50 µm from the vessel wall was quantified.

Confocal intravital microscopy and analysis

Confocal IVM was used to assess the migration dynamics of neutrophils across the endothelium and interactions with pericytes. WT and Gpr84^−/− mice on a Lyz2-EGFP-ki; Acta2-RFPcherry-Tg mice, mice were subject to an injection of TNF (300 ng). Movies were acquired on a Leica SP8 laser scanning microscope with a ×20 water-dripping objective (NA 1.0) and 8 kHz resonant scanner. Postcapillary venules between 20 and 45 µm were imaged for 1 h with two venules imaged per mouse 2–5 h post TNF-injection. Half vessels in a longitudinal orientation were recorded to achieve dimensions of typically 300 × 130 × 35 µm with a resulting voxel size of approximately 0.29 (x) × 0.29 (y) × 0.69 (z) µm, and serial z-stacks of 0.7 µm optical sections were acquired every 60 s. The time-lapse serial z stacks were assembled and processed into 3D videos and rendered using IMARIS software^TM (Bitplane, Zurich, Switzerland) for analysis.

Neutrophil dynamics were determined by manual tracking of individual neutrophils using IMARIS software. When assessing intraluminal behaviors, neutrophil “crawling” on endothelial cells was defined as a directional movement of ~1 cell radius (~10 µm). Individual cells, which were clearly visible in terms of dynamics and location, were tracked over 3 or more frames (≥ 3 min) to generate the speed of crawling (µm/min). Neutrophil tethers were defined as elongated structures of 5–10 µm in length with a rear contact point on the luminal endothelium. The number of neutrophil-derived fragments within a post-capillary venule, over each 1 h imaging period, was manually counted. Neutrophil fragments were considered as spherical, GFP⁺ structures that were notably smaller than neutrophils with an average radius of approximately ~2 µm.

To assess neutrophil dynamics across the vessel wall, neutrophil TEM was defined as an event where neutrophils fully migrated through EC junctions in a luminal-to-abluminal direction. Trans- pericyte migration (TPM) was defined as the point at which a neutrophil displayed an initial protrusion through the pericyte layer before completing their migration into the perivascular space. Neutrophils were categorized as displaying “aborted” TPM when they displayed a minimum of 1 oscillation during the 1 h imaging period—a protrusion of the leading edge through the pericyte border followed by a retraction into the sub-endothelial space. In Fig. 5B, the maximum length of a perivascular neutrophil was measured within a 1 h imaging period; measurements were taken from the leading edge of the neutrophil to the point of attachment between the posterior protrusion and the venular wall. In Fig. 5C, the time attached to the vessel wall was considered the period between the end of TPM and the completion of tail detachment from the venular wall.

Transmission electron microscopy

For mouse transmission electron microscopy studies, cremaster muscles were excised from mice subjected to 4 h TNF-induced inflammation, pinned to dental wax before fixation in a freshly prepared mix of 2.5% glutaraldehyde (Thermo Fisher Scientific) in 0.05 M cacodylate buffer (Thermo Fisher Scientific) supplemented with 0.18 mM calcium chloride, for 16 h at 4 °C. Samples were then secondarily fixed in 1% OsO₄ for 1 h, and taken through serial dehydrations in an automatic tissue processor (Leica EM TP) prior to embedding in resin. 70 nm sections were cut with a UC6 Leica ultramicrotome and Diamond Knife (Diatome) and were collected onto pioloform-coated copper slot grids (Agar Scientific) and imaged using a Tecnai 12 BioTwin Spirit TEM (tungsten filament, 120 kV) and a FEI Eagle 4k×4k CCD camera.

Whole blood collection

Mice were either terminally anesthetized with 3% isoflurane and the abdomen exposed to collect whole blood from the caudal vena cava, followed by cervical dislocation, or sacrificed prior to a cardiac puncture with collection of whole blood directly into 50 mM EDTA.

Flow cytometry

Differential leukocyte counts in whole blood and peritoneal lavage were assessed by flow cytometry. Blood samples were incubated with ACK erythrocyte lysis buffer (150 mM NH₃CL, 1 mM KHCO₃, and 1 mM EDTA) for 5 min. Samples were then washed once with staining buffer (2 mM EDTA and 0.5% BSA in PBS). Samples were then incubated with Fc-block (anti-CD16/32) antibodies (5 µg/ml) diluted in staining buffer for 15 min at 4 °C. Subsequently samples were stained for surface cells markers diluted in staining buffer (and their appropriate isotypes controls); anti-CD45 PB (1 µg/ml; 30-F11; 103126; BioLegend), CD3 PE (1 µg/ml; 17A2; 100205; BioLegend), CD19 AF488 (1 µg/ml; 6D5; 115524; BioLegend), CD11b PE/Cy7 (0.2 µg/ml; M1/70; 101216; BioLegend), Ly6C PerCP-Cy5.5 (0.2 µg/ml; HK1.4; 128012; BioLegend), Ly6G AF647 (0.5 µg/ml; 1A8; 127610; BioLegend), F4/80 PE (1 µg/ml; BM8; 123110; BioLegend), CD115 APC-Cy7 (1 µg/ml; AFS98; 135531; BioLegend), CD62L PE (0.2 µg/ml; MEL-14;104407; BioLegend), CD11a PE (0.1 µg/ml; M17/4; 101107; BioLegend), CD11b PE (0.1 µg/ml; M17/0; 101207; BioLegend), CD61 PE (0.4 µg/ml; 2C9.G2 (HMβ3-1); 101107; BioLegend), CD51 PE (0.4 µg/ml; RMV-7; 104105; BioLegend), CD29 PE (0.4 µg/ml; HMβ1-1; 102207; BioLegend), CD18 PE (2 µg/ml; M18/2; 101407; BioLegend), CD49 PE (0.4 µg/ml; 5H10-27 (MFR5); 103805; BioLegend), CD11c PE (1 µg/ml; N418; 117307; BioLegend). Gating strategies were based on isotype and FMO controls. Accurate cell counts were validated using counting beads and analyzed using a LSR Fortessa flow cytometer (BD Biosciences) and FloJo software (TreeStar).

Bulk-RNA sequencing

Neutrophils from the whole blood of male WT versus GPR84 KO mice were FACs sorted and gated using Zombie Yellow™ fixable Viability kit as a live/dead marker, single cells, CD45⁺, CD3^-, CD19^-, CD11b⁺, Ly6C^int, Ly6G⁺ using a FACS Aria™ II (BD Biosciences). Exactly 10,000 neutrophils were collected in Trizol and sent to GENEWIZ (Azenta Life Sciences, US) for RNA isolation, quality control, and ultra-low input RNA sequencing amplification. Sequencing was performed on an Illumina HiSeq instrument with a 2×150-base-pair (BP) paired-end configuration. In total, 35 million reads were mapped to the reference mouse transcriptome (Gencode release v28, GRCm39), and transcripts were quantified by the STAR aligner v2.7.3a in 2-pass mapping mode. Protein-coding and immunoglobulin genes were used for analysis. Counts were subjected to variance-stabilizing transformation (VST) using DESeq2 v.1.42.1. Principal component analysis (PCA) was undertaken with the prcomp function in R using the VST data, and its plots were generated using ggplot2 v3.4.4. Differential expression analysis was performed on count data using DESeq2 v.1..42 using an unadjusted P value cut-off of 0.05 to identify transcripts differentially expressed between genotypes, where the volcano plot was visualized using easylabel v0.3.3. Pathway analysis was performed on DE transcripts using the WikiPathways 2024 by EnrichR database (as described by (Kuleshov et al, 2016) with enriched pathways considered significant with an adjusted P value of P < 0.05. Heatmaps were generated using ComplexHeatmap v2.21.1.

In vitro adhesion assay

Ibidi chamber 8-well slides were coated overnight with 3 µg/ml rmICAM-1 (Abcam) at 4 °C, blocked with 1% Casein (in PBS) for 1 h at room temperature (RT), and washed with PBS. Murine neutrophils were isolated from bone marrow cells (harvested from tibiae and femora) by negative magnetic cell sorting using the Neutrophil Isolation Kit (Miltenyi Biotec) according to the manufacturer’s instructions. The purity of isolated neutrophils (CD45⁺/Ly6G⁺) expression profile) was consistently > 95% as determined by flow cytometry. Isolated neutrophils were stained with Cell Tracker™ Green CMFDA or Cell Tracker™ Orange CMRA (Thermofisher Scientific) at 0.5 µM for 10 min at RT. Potential effects of Cell Tracker™ colors were controlled for by alternating between genotypes for experimental replicates. Isolated neutrophils were seeded at 100,000 of each genotype and left to settle for ~20 min at RT prior to stimulation with 10 nM CXCL1 (Peprotech) for 15 min at 37 °C and imaged on a Nikon Spinning Disk SoRa at 60x magnification prior to assessment of the proportion of “spread” or motile cells by manual counting on IMARIS software. Neutrophils were classified as “spread” when displaying a flattened and non-spherical-like structure and considered as “motile” when displaying membrane ruffling and migration on the ICAM-1 coated surface. To assess cell surface area, images at the 15 min endpoint were analyzed using ImageJ (NIH).

RhoA G-LISA GTPase activation assay

Murine bone marrow-derived neutrophils from WT and Gpr84^−/− mice were isolated as detailed above. At least 3 million cells (pooled from 2 to 3 mice) per sample were stimulated with either the GPR84 agonist LY237 (10 µM) (MedChem Express), or as a positive control, with the FPR agonist WKYMVm (1 µM), for 4 min at 37 °C. Cells were lysed at 4°, the protein content collected, and immediately snap-frozen. Samples were normalized for protein content and processed using RhoA G-LISA GTPase Activation Assay Kit (Cytoskeleton) according to the manufacturer’s instructions. Biological replicates were performed, and all samples were run in technical triplicates with absorbance read at 490 nm using a Sunrise microplate reader (Tecan^TM).

Zymosan-induced peritonitis

C57Bl6/J mice (Charles River) received 10 or 30 mg/kg GLPG1205 (Galapagos Pharmaceuticals) in (40% PEG 200, 60% sterile water) or vehicle control by oral gavage 24 h prior to induction of inflammation in the TNF-inflamed cremaster model or zymosan-induced peritonitis. Peritonitis was induced by an i.p. injection of zymosan (1 mg) in 500 µl of PBS or PBS alone as a control. Subsequently, 24 h later, the peritoneal lavage was collected in 5 ml PBS containing 5 mM EDTA (Sigma-Aldrich) and 0.25% BSA. The peritoneal exudates were labeled for myeloid cell markers of interest for flow cytometry and gated as follows for neutrophils (single cells, live, CD45⁺ CD115⁻ F4/80⁻ Ly6C^int Ly6G⁺) and monocytes (single cells, live, CD45⁺ CD115⁺ F4/80^low-int Ly6C^hi-int Ly6G^-). Absolute cell counts were calculated with the application of counting beads and assessed by flow cytometry.

Ischemia–reperfusion injury in the ear

Male and female C57Bl6/J mice (Charles River) received 30 mg/kg GLPG1205 or vehicle control as above. Mice were briefly anesthetized with 3% isoflurane prior to application of magnets covering the lower region of the ear; anesthesia was reduced to 1.5% during the 1.5 h ischemia period, and a heating pad was used to maintain body temperature at 37 °C. The magnets (2 mm thick 45H Neodymium with 0.8 kg pull) were removed prior to a 3 h reperfusion period. Mice were sacrificed, whole blood collected (to assess differential leukocyte counts), and ear tissues harvested and processed for staining and analysis.

Cutaneous wound injury

Male and female C57Bl6/J mice (Charles River) received 30 mg/kg GLPG1205 or vehicle control as above. Briefly, 8-week-old mice were anesthetized with 3% isofluorane, prior to shaving their back skin and generating paired excisional wounds with a 4 mm biopsy punch (Kai Medical). Animals were sacrificed at 0, 18, and 36 h post-wounding, and skin samples were processed for histological analysis.

Histological staining and image analysis

Tissue was fixed in 4% paraformaldehyde overnight prior to transfer through serial dehydration steps in preparation for embedding in paraffin using an automatic tissue processor (Expredia Exelsior AS). In all, 7-μm sections were cut with a Leica 2125 microtome and subsequently deparaffinized and stained with hematoxylin and eosin. Stained sections were then imaged in an Olympus VS200 slide scanner, followed by analysis in QuPath (Bankhead et al, 2017). Quantification of the total number of neutrophils was performed using the built-in Cell Detection algorithm that detects hematoxylin-stained nuclei with minor adjustments to intensity parameters for neutrophil nuclear morphology: threshold (0.15) and cell expansion (2 µm). This was validated by comparing manual and automated counts.

Quantification and statistics

Statistical analyses were performed using Prism 10.1 software (GraphPad), and all datasets are expressed as mean ± SEM. Details for exact n numbers are reported in the legends for each Figure. Sample sizes were chosen based on published work in which similar results have been reported.

The authors declare no competing interests.