ARAP3 protects from excessive formylated peptide‐induced microvascular leakage by acting on endothelial cells and neutrophils

Vascular permeability is temporarily heightened during inflammation, but excessive inflammation‐associated microvascular leakage can be detrimental, as evidenced in the inflamed lung. Formylated peptides regulate vascular leakage indirectly via formylated peptide receptor‐1 (FPR1)‐mediated recruitment and activation of neutrophils. Here we identify how the GTPase‐activating protein ARAP3 protects against formylated peptide‐induced microvascular permeability via endothelial cells and neutrophils. In vitro, Arap3−/− endothelial monolayers were characterised by enhanced formylated peptide‐induced permeability due to upregulated endothelial FPR1 and enhanced vascular endothelial cadherin internalisation. In vivo, enhanced inflammation‐associated microvascular leakage was observed in Arap3−/− mice. Leakage of plasma protein into the lungs of Arap3−/− mice increased within hours of formylated peptide administration. Adoptive transfer experiments indicated this was dependent upon ARAP3 deficiency in both immune and non‐immune cells. Bronchoalveolar lavages of formylated peptide‐challenged Arap3−/− mice contained neutrophil extracellular traps (NETs). Pharmacological inhibition of NET formation abrogated excessive microvascular leakage, indicating a critical function of NETs in this context. The observation that Arap3−/− mice developed more severe influenza suggests these findings are pertinent to pathological situations characterised by abundant formylated peptides. © 2024 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.


Introduction
Vascular permeability is temporarily heightened during the normal inflammatory response to permit protein-rich plasma to enter the interstitial space and promote the timely removal of threats.Many pathological situations are characterised by excessive inflammation-associated vascular leakage.For example, in acute lung inflammation, the capillaries of the lung become excessively permeable as a consequence of severe inflammation, allowing protein-rich plasma to enter the alveoli of the lung.In acute respiratory distress syndrome (ARDS), a dangerous complication of influenza and other pre-existing conditions, excessive oedema formation interferes with gas exchange.
Many soluble agents act directly on ECs to promote permeability, but neutrophil interactions with the endothelium are also deeply involved in regulating inflammation-associated vascular leakage, e.g. in response to stimulation with formylated peptides [6,7], although the underlying mechanism remains largely obscure.Formylated peptides are danger/ pathogen-associated molecular patterns (DAMPs/ PAMPs) of bacterial or mitochondrial origin, liberated for example by necrotic cells.Formylated peptides are powerful activators of diverse neutrophil functions and signal via formylated peptide receptors (FPRs), in particular FPR1 [8].
Neutrophils are abundant circulating leukocytes that act early in immune defence against infections and that are rapidly recruited to sites of infections or sterile injuries [9,10].If the tight control of neutrophils in inflammation goes awry, they can promote serious host damage.In particular, neutrophil extracellular traps (NETs), strands of decondensed chromatin decorated with cytotoxic proteins that are released by neutrophils under certain situations, are important contributors of neutrophil-derived host damage in the lung and elsewhere.
In this study, we addressed ARAP3 function in endothelial permeability, focusing in particular on formylated peptide-induced microvascular leakage.We show that ARAP3 provides protection from excessive inflammationassociated microvascular leakage without being involved in regulating basal leakage of protein-rich plasma.Mechanistically, ARAP3 in both neutrophils and ECs contributes to limiting excessive microvascular leakage.This is because (i) ARAP3 protects endothelial VE-cadherin from excessive formylated peptide-induced internalisation and subsequent trafficking to lysosomes, (ii) ARAP3 downregulates the endothelial FPR1 receptor to protect ECs from formylated peptide-induced endothelial permeability, and (iii) ARAP3 prevents excessive formylated peptide-induced neutrophilic inflammation via the formation of NETs.In an influenza infection model, ARAP3 deficiency promotes more severe outcomes, suggesting that our findings are of physiological consequence.

Materials and methods
Unless stated otherwise, materials were of the lowest available endotoxin level and obtained from Sigma-Aldrich/Merck (Gillingham, UK).Tissue culture reagents were from Gibco (Fisher Scientific, Loughborough, UK) and tissue culture plastics from Corning (Fisher Scientific).

Study approval
Animal work was approved by the University of Edinburgh Animal Welfare Committee and conducted under the control of a UK Home Office project license (PFFB 42579 and PP4667029).

Mouse models
Inducible Arap3 fl/fl mice crossed with Rosa-ERT2Cre mice and matched Arap3 +/+ Rosa-ERT2Cre controls were induced by five successive oral gavages containing 1.5 mg tamoxifen each, followed by a period of rest of at least 10 days as described elsewhere [18].For ease of reading, tamoxifen-induced Arap3 fl/fl Rosa-ERT2Cre + mice are referred to as Arap3 À/À and tamoxifen-induced Arap3 +/+ Rosa-ERT2Cre + controls are referred to as wild-type (wt) controls.Fpr1 À/À mice [19] were used in some experiments, with sex-and age-matched wt littermates serving as controls.Mice were housed in a specific pathogen-free small-animal barrier unit at the University of Edinburgh in individually ventilated cages.

Vascular leakage in the absence of challenge
Mice were injected with 100 μl 1% Evans blue into the tail vein and culled 30 min later by i.p. pentobarbital.Following transcardial perfusion with 20 ml saline supplemented with 1 mM EDTA, lung, liver, heart spleen, and brain were collected and weighed.Evans blue was extracted in formamide overnight, and contents were established according to a standard curve and plotted as μg/mg wet tissue.

Immune cell depletions
To deplete neutrophils, mice were injected i.p. with 0.25 mg low endotoxin, azide-free anti-GR1 as described elsewhere [20].To specifically deplete alveolar macrophages as described [22], 0.0625 mg clodronate vesicles (Liposoma, Amsterdam, The Netherlands) in 50 μl saline were given to mice i.t.96 h prior to the experiment.

Influenza A virus (IAV) infection model
Seven-week-old, female tamoxifen-induced Arap3 fl/fl Rosa-ERT2Cre + and Arap3 +/+ Rosa-ERT2Cre + mice were infected i.t. with 20 PFU of H1N1 (PR8) virus.Mice were provided with a nutritional gel supplement to counteract excessive weight loss during IAV infection.Mice were weighed and scored for clinical signs (posture, piloerection, spontaneous activity, response to touch, loss of body temperature, respiratory distress) daily and, once weight loss reached 20% of starting weight, twice daily.

Flow cytometry
In some experiments, single-cell lung digests were generated for analysis by flow cytometry [18].Lavage or lung digest neutrophils were identified as CD45 + , Ly6G high singlets.Flow cytometry was performed using an Attune NxT flow cytometer (Fisher Scientific), and data were analysed using FCS Express 7 (De Novo Software, Pasadena, CA, USA).

HUVEC culture and functional experiments
Human umbilical vein endothelial cells (HUVECs) from pooled donors (Lonza Bioscience, Visp, Switzerland) were cultured in EGM-2 complete endothelial growth medium (Lonza Bioscience) and used until passage 5.For siRNA-mediated knock-down, HUVECs were transfected with siRNA oligonucleotide duplexes targeting the gene of interest (ARAP3: SR312043; FPR1: SR301651) or control siRNA (SR30004) obtained from Origene (Herford, Germany).Transfections were performed with GeneFECTOR reagent (Venn Nova, Pompano Beach, FL, USA) in OptiMEM as described [23].Transfected cells were used 48 h after transfection.To analyse HUVECs by flow cytometry, Enzyme-free Cell Dissociation Buffer (Fisher Scientific) was used to dissociate them from the culture vessel.HUVEC RNA was isolated using a kit (Zymo Research, Freiburg, Germany), reverse transcribed, and amplified using a GoTaq real-time quantitative polymerase chain reaction (RT-qPCR) kit (Promega).Primers (supplementary material, Table S1) were designed using PrimerBank [24] (https://pga.mgh.harvard.edu/primerbank/),and qPCR was performed on a Quantstudio 5 system (Fisher Scientific).To measure endothelial permeability, 3 Â 10 5 HUVECs were plated into fibronectin-coated Transwell inserts with a 0.4-μm pore size (Sarstedt, Nümbrecht, Germany), cultured until confluent, and washed in EGM-2 basal medium. 1 μg/ml FITC-dextran (40 kDa) in the presence of 100 nM fMLF or 100 ng/ml vascular endothelial growth factor A (VEGF-A) (Invitrogen, Paisley, UK) in basal medium was added to the top chamber, and medium was collected from the lower chamber for analysis of fluorescence on a Cytation plate reader (Biotek, Agilent, Santa Clara, CA, USA) at indicated times, essentially as described [23].For antibody feeding experiments, HUVECs were plated onto coverslips that had been coated with 20 μg/ml fibronectin.Confluent monolayers were held on ice, labelled with AF647-conjugated anti-VE-cadherin for 30 min, washed, and moved to warm growth medium supplemented with 100 nM fMLF for 2 h in a humidified, controlled CO 2 incubator at 37 C, before being put on ice for labelling with AF595-conjugated anti-VE-cadherin, washed, fixed in 4% paraformaldehyde, and mounted in Prolong Gold (Fisher Scientific).Image stacks were obtained on a Leica SP8 confocal microscope.Raw confocal image stacks were deconvolved and, using the co-localisation of fluorescent signal module, analysed using Huygens Professional software (Scientific Volume Imaging).The global intersection coefficient i was plotted.Flattened optical stacks were generated and pseudocolouring applied to representative examples using Fiji freeware (NIH).

b.End5 cells
b.End5 cells were obtained from the European Tissue Culture Collection, tested for mycoplasma upon arrival, and then assumed to remain mycoplasma free.Arap3 À/À b.End5 cells were generated with CRISPR Cas9.Candidate gRNAs (supplementary material, Table S2) were designed using E-CRISP [25] and cloned into lentiCRISPRv2 [26,27] following the lentiCRISPRv2 cloning instructions available via Addgene.Transduced clones were screened and the relevant Arap3 exon sequenced (not shown).For rescue experiments, Arap3 À/À b.End5 cells were transduced with lentiviruses encoding full-length GFP-ARAP3 constructs [12].To analyse permeability of b.End5 monolayers, cells were plated into wells of 20 μg/ml laminin-coated real-time cell analysis (RTCA) E-plates (Agilent), allowed to reach confluence, and conductance of monolayers that were stimulated with 1 μM fMLF, 100 ng/ml VEGF-A, or vehicle was analysed by xCELLigence RTCA (S16; Agilent) following the manufacturer's instructions.Arf6/RhoA activity was analysed with confluent b.End5 monolayers that had or had not been stimulated with 1 μM fMLF or 100 ng/ml VEGF-A using G-LISA kits (cytoskeleton) following the manufacturer's instructions.

Statistical analyses
Where data met assumptions for parametric tests, for pairwise comparisons two-tailed t-tests were used, and ARAP3 protects from formylpeptide-induced microvascular leakage for multiple comparisons ANOVA with suitable post hoc test was used.Pairwise comparisons of nonparametric data were performed using Mann-Whitney rank sum tests and multiple comparison by Kruskal-Wallis tests with suitable post hoc test, as detailed in figure legends.For kinetic experiments area under the curve (AUC) was calculated.p values <0.05 were considered statistically significant.

Endothelial ARAP3 regulates inflammationassociated vascular leakage
We previously showed that knocking out Arap3 caused embryonic lethality due to a severe developmental angiogenesis defect [13], whereas loss of Arap3 is well tolerated in adult mice in which Arap3 has been ubiquitously knocked out in an inducible fashion in the absence of challenge [18] (see Materials and methods for details on genetic model and induction).In keeping with this, we did not observe enhanced endothelial permeability in unchallenged ARAP3-deficient mice (Figure 1A).However, in thioglycollate peritonitischallenged mice, we observed increased lavage protein in ARAP3 deficiency as compared to controls (supplementary material, Figure S1A,B), suggesting that ARAP3 is a regulator of inflammation-associated vascular leakage.
ARAP3 expression in mice is restricted to a few cell types [11,13,16] and is highest in the lung (supplementary material, Figure S1E), leading us to further analyse its function in microvascular leakage at this site.Administration of the synthetic formylated peptide fMLF either i.t. or i.v.caused significant leakage of

ARAP3 regulates fMLF-induced permeability of endothelial monolayers
In keeping with the fact that fMLF is a ligand for the formylated peptide receptor FPR1, Fpr1 À/À mice [19] displayed no microvascular leakage in response to challenge with fMLF (supplementary material, Figure S2A).Formylated peptides were previously shown to cause microvascular leakage in a neutrophil-dependent fashion [6], a cell type in which FPR1 function is particularly well characterised [28].To our surprise, however, analysis of wt > wt and wt > Fpr1 À/À bone marrow chimeras suggested that FPR1 in a radiation-insensitive compartment mediated sensitivity to fMLF-induced microvascular leakage in the lung (Figure 2A).We derived Arap3 À/À b.End5 cells (supplementary material, Figure S2B) and analysed endothelial permeability in monolayers of control or Arap3 À/À ECs in response to treatment with fMLF and, as a control, VEGF-A.We characterised conductance with ECs plated onto sensor chips by RTCA, where Arap3 À/À monolayers displayed more fMLF (and also VEGF) induced permeability than controls (Figure 2B,C and supplementary material, Figure S2C,D).The same trend was observed when analysing fluorescein-dextran leakiness of such monolayers in Transwell assays (Figure 2D and supplementary material, Figure S2E).ARAP3 siRNA-transfected HUVEC monolayers were also characterised by higher endothelial fMLF-and VEGF-induced permeability than controls (Figure 2E and supplementary material, Figure S2F-H).Taken together, these observations suggest that ARAP3 protects ECs from formylated peptideinduced leakage.

ARAP3 downregulates endothelial FPR1
Although best characterised in leukocytes, FPR1 is also expressed by some other cell types.In the absence of antibodies capable of convincingly detecting FPR1 by western blotting, we tested for FPR1 expression in ECs by flow cytometry, making use of an antibody that detects non-denatured human FPR1.Knocking down FPR1 in HUVECs resulted in reduced signal with this antibody (Figure 3A,B), suggesting that the antibody was specific and that FPR1 was indeed expressed by ECs.We examined FPR1 in HUVECs in which ARAP3 had or had not been knocked down, observing increased cell surface FPR1, but not FPR1 mRNA expression, in ARAP3 knock-down HUVECs (Figure 3C-E).Knock down of ARAP3 but not FPR1 and ARAP3 together caused elevated fMLF-induced dextran permeability of HUVEC monolayers (Figure 3F).Altogether, this suggests that ARAP3 functions to Red asterisks in (B) refer to significant differences between fMLF-stimulated and vehicle-treated Arap3 À/À cells; red number symbols refer to significant differences between fMLF-stimulated Arap3 À/À and control cells.
ARAP3 protects from formylpeptide-induced microvascular leakage 351 downregulate endothelial FPR1 cell surface expression, providing an explanation for the increased susceptibility of ARAP3-deficient ECs to formylated peptide-induced permeability.

ARAP3 regulates VE-cadherin trafficking
Availability of VE-cadherin, the major determinant of endothelial adherens junctions, is regulated by endosomal trafficking.We performed antibody feeding pulse-chase experiments to obtain insight into VE-cadherin internalisation in formylated peptidestimulated ECs.This involved labelling VE-cadherin in control-and ARAP3 siRNA-transfected HUVEC monolayers at two time points using two different fluorescent conjugates that detect different epitopes on the extracellular domain of VE-cadherin while stimulating the cells with fMLF (Figure 3G).We observed reduced co-localisation of the two VE-cadherin signals, suggesting that junctional VE-cadherin was internalised more efficiently, in ARAP3 knock-down HUVECs (Figure 3H,I and supplementary material, Figure S3).When labelling a lysosomal marker (LAMP-1) and a recycling endosomal marker (Rab11A) alongside total VE-cadherin in fMLF-stimulated HUVECs in which ARAP3 had or had not been knocked down, we observed enhanced overlap of the lysosomal and VE-cadherin signals in ARAP3-deficient cells (Figure 3J).

ARAP3 protection of VE-cadherin proteolysis occurs in a GAP-independent fashion
Several small GTPases were previously shown to act as regulators of endothelial permeability [3][4][5].Since ARAP3 is a dual GAP for RhoA and Arf6 [11,12], we analysed RhoA and Arf6 activities in confluent control and Arap3 À/À b.End5 cell monolayers.fMLF stimulation induced only very modest activation of Arf6 and RhoA irrespective of ARAP3 (Figure 4A-D).RhoA activation was reduced in ARAP3-deficient, VEGF-stimulated monolayers (supplementary material, Figure S4).GAPs are large multidomain proteins that fulfil many functions, frequently acting as signalling scaffolds in addition to their catalytic function.To test whether ARAP3-dependent regulation of formylated peptideinduced endothelial permeability was dependent on its catalytic activity, we rescued Arap3 À/À b.End5 cells by introducing a GFP-tagged wt ARAP3 or a GFP-tagged ARAP3 protects from formylpeptide-induced microvascular leakage 353 GAP-dead point mutation construct (C504A, R982A [12]; Figure 4E).Expressing either ARAP3 construct in Arap3 À/À b.End5 cells resulted in monolayers that behaved like wt control cells in terms of endothelial permeability in response to fMLF-stimulation (Figure 4F,G).Overall, these observations suggest that ARAP3 protects ECs from excessive formylated peptide-induced endothelial permeability by (i) reducing surface expression of endothelial FPR1 and (ii) protecting VE-cadherin from internalisation and subsequent trafficking to the lysosome.Our data suggest, moreover, that ARAP3 protects endothelial monolayers from formylated peptideinduced permeability in a GAP-independent fashion.

Neutrophils make an important contribution to formylated peptide-induced microvascular leakage of ARAP3-deficient mice
Having established that fMLF stimulation causes enhanced permeability of ARAP3-deficient EC monolayers in vitro, we set out to explore the role of neutrophils in fMLFinduced microvascular leakage in vivo.We challenged control or Arap3 À/À mice in which neutrophils had or had not been depleted with i.t.fMLF.Neutrophil depletion significantly reduced leakage of plasma protein into the alveolar space of ARAP3-deficient mice (Figure 5A,B).In contrast, depletion of alveolar macrophages, important sentinels that highly express FPR1 [29], but not detectably ARAP3, did not affect formylpeptide-induced microvascular leakage (supplementary material, Figure S5A-C).This identified neutrophils as being crucial for fMLF-induced capillary leakage in the lung.We next asked about the role of ARAP3 in radiosensitive and radioresistant compartments.We generated ARAP3 criss-cross bone marrow chimeras and challenged the chimeras with fMLF.Simultaneous ARAP3 deficiency in both compartments was required for the heightened fMLF-induced leakage observed with ARAP3-deficient animals (Figure 5C-E).This contrasts with thioglycollate-induced leakage in the peritoneum of criss-cross chimeras, which was entirely dependent upon ARAP3 deficiency in the radioresistant compartment (supplementary material, Figure S1C,D).In summary, neutrophils were required for the induction of fMLF-induced microvascular leakage in the lungs of ARAP3-deficient animals, but ARAP3 deficiency in both radioresistant and radiosensitive compartments was required for this leakage to develop.

NETs drive fMLF-induced neutrophilic inflammation in ARAP3-deficient mice
We analysed neutrophils within bronchoalveolar lavage fluid (BALf) of fMLF-challenged mice, observing surprisingly low numbers of BALf neutrophils (Figure S5D).This contrasted with significantly increased neutrophil numbers observed in lung digests from ARAP3-deficient mice that had been challenged with fMLF (Figure 5F).In keeping with the latter observation, lung sections of ARAP3-deficient mice that had been challenged with fMLF were characterised by areas of dense inflammatory infiltrate that contained numerous neutrophils (Figure 5G,H).
We hypothesised that the neutrophils contained in ARAP3-deficient lungs would be difficult to lavage due to being 'stuck' and that chromatin derived from neutrophil extracellular traps might be responsible.Analysis of lavage cells in samples from ARAP3-deficient mice that had been challenged with fMLF did indeed identify the presence of NETs (supplementary material, Figure S6A for an example).We performed ELISAs for a quantitative read-out, testing for two NET markers, citrullinated histone H3 and neutrophil elastase.Both were significantly elevated in BALf samples from fMLF-challenged Arap3 À/À mice that were characterised by elevated BALf protein (Figure 6A-C).Administering either DNAse I or the PAD-4 inhibitor GSK-484 just prior to challenge with fMLF protected ARAP3-deficient mice from developing fMLF-induced elevated microvascular leakage and interfered with NET generation according to the two markers (Figure 6A-C).This suggests that formylated peptides promoted NET generation by neutrophils in the lungs of ARAP3-deficient mice, in turn driving microvascular leakage.We did not observe elevated NET generation in peritoneal lavages of ARAP3-deficient mice after induction of zymosan peritonitis or with Arap3 À/À neutrophils in vitro (supplementary material, Figure S6B,C), suggesting that ARAP3 deficiency does not promote NET generation per se.

ARAP3 deficiency increases severity of influenza infection
Influenza infection causes the release of DAMPs including mitochondrial formylated peptides, triggering an inflammatory response by inducing respiratory epithelial cell death [30][31][32].We infected control and Arap3 À/À mice with IVA.Both genotypes experienced very similar weight loss in response to the infection (Figure 6D), suggesting the systemic inflammation in both groups caused a similar extent of anorexia.However, Arap3 À/À mice developed more severe disease according to clinical scoring during this infection (Figure 6E), displaying lethargy, hunching, piloerection, loss of resistance to handling, and laboured breathing.Lung histology of ARAP3-deficient mice identified denser inflammatory infiltrates as well as pulmonary oedema (Figure 6F for a representative example), and their BALf was characterised by elevated serum protein (supplementary material, Figure S6D,E).This suggests that ARAP3-dependent protection from formylated peptide-induced microvascular leakage is pertinent to physiological situations.

Discussion
This work interrogated the role of ARAP3 in inflammation-associated endothelial permeability, analysing contributions of both ECs and neutrophils.We focused on the lung, where ARAP3 is most highly

354
JY Chu et al expressed, and made intriguing observations with fMLF-induced capillary leakage at this site.In summary, our findings are that formylated peptides promote microvascular leakage by acting directly on ARAP3-deficient ECs and on neutrophils.In keeping with this, IAV infection resulted in more severe disease in ARAP3-deficient mice.
Our in vivo data suggest an important function of NETs in mediating formylated peptide-induced microvascular leakage of ARAP3-deficient mice, while our in vitro experiments demonstrated that ARAP3 protected cultured EC monolayers from permeability induced by physiological concentrations of synthetic fMLF.The mechanism by which ARAP3 ARAP3 protects from formylpeptide-induced microvascular leakage 355 downregulates surface endothelial FPR1 remains to be established; however, neutrophil FPR1 availability is known to be regulated by trafficking between an intracellular pool and the plasma membrane.Specifically, FPR1 internalisation is dependent upon serine/ threonine phosphorylation of residues in its C-terminus, while its recycling depends upon arrestin binding [33,34].Together these mechanisms determine the availability of neutrophil FPR1 for activation by fMLF.Intriguingly, however, ARAP3 deficiency does not enhance neutrophil responsiveness to fMLF [16][17][18], contrasting with the situation in ECs described here.This suggests that FPR1 is likely to be regulated in a cell typedependent fashion.Both Arf6 and RhoA were previously shown to modulate adherens junctions.MYD88-ARNO-Arf6 signalling controls VE-cadherin internalisation in response to IL-1β and LPS, both of which induce Arf6 activation [5,35], while Slit-Robo4 signalling was shown to inhibit VEGF-mediated Arf6 activation via the GAP protein GIT1 [36].RhoA also functions at adherens junctions, regulating actin tension and VE-cadherin internalisation and reducing endothelial barrier function.This was previously shown to occur in a p190RhoGAP-A and Rho GEF-H1 controlled fashion [37][38][39], including in response to challenge with thrombin or LPS.We did not observe fMLF-dependent activation of Arf6 in our experiments, suggesting Arf6 is an unlikely regulator of endothelial permeability in this context.fMLF stimulation caused mild activation of endothelial RhoA.ARAP3 deficiency did not significantly affect this, again not directly implying RhoA in fMLF-induced endothelial permeability.Like many other GAP proteins, ARAP3 is a large multidomain protein.As such, it is likely involved in regulating signalling not only via its catalytic activity but also by providing a scaffold.Intriguingly, the data presented here suggest that ARAP3-mediated regulation of VE-cadherin is likely to occur due to a scaffold function rather than ARAP3 catalytic activity.
Formylated peptides are DAMPs/PAMPs derived from bacterial or mitochondrial proteins that bind to formylated peptide receptors, specifically FPR1.Although they were originally assumed to be restricted to leukocytes, mapping of protein and mRNA expression has since also demonstrated some FPR1 expression in a range of non-immune cell types, including endothelial and epithelial cells (e.g.https://www.proteinatlas.org/ENSG00000171051-FPR1).
A previous study identified formylated peptides in the BALf of ARDS patients [40].This and several other reports demonstrated already that administration of formylated peptides to mice could induce acute lung injury and microvascular leakage in the lung [40][41][42][43].
We described here excessive microvascular leakage in ARAP3-deficiency that occurs at very early time points under conditions in which wt animals are not affected.Our work singles out Arap3 as a susceptibility gene for formylated peptide-induced microvascular leakage.It established that ARAP3 in either radiation-sensitive or -insensitive compartments is sufficient to provide protection from the excessive leakage we report.Specifically, our in vivo data show that in ARAP3-deficiency formylated peptides induced important NET formation in the lung.Mechanistically, NET release may be triggered due to extensive interactions between ARAP3-deficient neutrophils, which are hyperadhesive due to their activated integrins, and fMLF-activated ECs.Due to their activated integrins, ARAP3-deficient neutrophils not only adhere better in response to stimulation but also produce more ROS [16,18], a prerequisite for NET generation.NETs in turn activate ECs, promoting the expression of adhesion proteins ICAM-1 and VCAM-1, promoting the adhesion of more neutrophils [44].Meanwhile, activated ECs promote NET generation by neutrophils [45], creating optimal conditions for a vicious cycle.NETs and the neutrophil proteins that decorate them are highly pro-inflammatory structures that mediate heightened endothelial permeability [46,47] and promote endothelial damage in pneumonia and sterile acute lung injury [48][49][50][51][52].Given that both infections and trauma drive the release of formylated peptides in the lung and in light of the more severe IAV infections observed with ARAP3-deficient mice, our observations are likely to be relevant to lung inflammation under a range of conditions.

Figure 1 .
Figure 1.ARAP3 provides protection from formylated peptide-induced microvascular leakage.Tamoxifen-induced ERT2Cre + controls (wt) or Arap3 fl/fl ERT2Cre + (À/À) mice (A) were injected i.v. with Evans blue dye without further challenge as detailed in Materials and methods.Evans blue leakiness is plotted.(B-G) Animals were challenged with fMLF i.t. or i.v. and BALf sampled for analysis.(B and E) Experimental scheme.(C, D, F, and G) Analysis of BALf proteins.Experiments performed on at least two separate occasions are pooled in the graphs presented.Each symbol represents one experimental animal.Data were analysed by two-way ANOVA with Šidák's multiple comparison test (A) or Tukey's multiple comparison test (C, D, F, and G).** or ## , p < 0.01; *** or ### , p < 0.001; **** or #### , p < 0.0001; ns, not significant.
350JY Chu et al plasma protein into the alveolar space of inducible ARAP3-deficient but not in inducible control mice (Figure1B-G).In contrast, i.t.LPS or MIP2 (CXCL2) did not cause elevated microvascular leakage into the alveolar space of ARAP3-deficient mice (supplementary material, FigureS1F,G).

Figure 2 .
Figure 2. ARAP3 protects from formylated peptide-induced endothelial permeability.(A) Bone marrow chimeras as indicated were challenged with fMLF i.t.BALf was harvested 6 h later for analysis of total protein.(B-E) Monolayers of ARAP3-deficient b.End5 or HUVECs and matched controls were stimulated with fMLF or vehicle.(B and C) RTCA of electrical impedance measurements was performed to characterise b.End5 monolayer permeability.(B) Time course.(C) AUC.(D and E) Transwell FITC-dextran permeability of vehicle or fMLF-stimulated (D) bEnd5 and (E) HUVEC monolayers that had been transiently transfected with scrambled control or ARAP3 targeting siRNA.(A) Each symbol represents one experimental animal.Experiments performed on two separate occasions are pooled.(B-E).Graphs show mean ± SEM and combine results obtained in four separately conducted experiments.Transwell experiments are expressed as fMLF-stimulated experimental condition over its vehicle-stimulated control.Analysis was by two-way ANOVA with Tukey's multiple comparison test or (B) unprotected Fisher's least significant difference multiple comparison test.*, # p < 0.05; **, ## p < 0.01; ***p < 0.001, ****p < 0.0001; ns, not significant.Red asterisks in (B) refer to significant differences between fMLF-stimulated and vehicle-treated Arap3 À/À cells; red number symbols refer to significant differences between fMLF-stimulated Arap3 À/À and control cells.

Figure 3 .
Figure 3. ARAP3 regulates endothelial FPR1 and VE-cadherin.HUVECs were transfected with scrambled (Ctrl) siRNA or siRNA targeting FPR1 or ARAP3 as indicated.(A-D) FPR1 was detected using flow cytometry.(A and C) Histogram of representative experiments with isotype control in black, control siRNA-transfected cells in grey and (A) FPR1 or (C) ARAP3 siRNA transfected cells in red.(B and D) Mean fluorescence intensity (MFI) of six separately conducted experiments.(E) Expression of FPR1 mRNA performed with three independent transfections.(F) Transwell FITC-dextran permeability combining a minimum of four separately conducted experiments.In the legend of this graph, ARAP3 is shortened to A3. (G-J) HUVECs transfected with siRNA as indicated were plated onto glass coverslips, allowed to form confluent monolayers, and stimulated with 100 nM fMLF.(G) Schematic diagram of antibody feeding experiments.(H) VE-cadherin analysis by confocal microscopy.(I) Automated analysis of co-localisation was performed with Huygens Software as detailed in Materials and methods.Co-localisation according to efficient i is plotted.Each symbol represents a separate raw image; images were obtained from a minimum of two separate transfections.(J) Fixed, permeabilised cells were labelled for VE-cadherin, LAMP-1, and Rab11A.(H and J) Examples of representative flattened confocal image stacks; scale bars, 10 μm (H) and 20 μm (J).Data were analysed by (B, D, and E) one-way ANOVA with Dunnet's multiple comparison test, comparing all experimental conditions to control siRNA transfected cells, (F) two-way ANOVA with Tukey's multiple comparison test with asterisk referring to activation and number symbol to differences between conditions tested and (I) unpaired two-tailed t-test.*, # p < 0.05; **p < 0.01.

Figure 4 .
Figure 4. ARAP3 regulates fMLF-induced endothelial permeability in a GAP-independent fashion.(A-D) Wild-type (wt) control and Arap3 À/À b.End5 monolayers were stimulated for the indicated times with fMLF prior to generation of lysates for analysis of Arf6 (A and B) and RhoA (C and D) activity by G-LISA.(A and C) Graphs combine a minimum of four separately performed experiments; representative western blots show Arf6/RhoA contents of the cell lysates analysed.(B and D) AUC of graphs plotted in (A) and (C).(E-G) Arap3 À/À b.End5 cells were lentivirally transduced to express GFP-ARAP3 (GFP-WT) or a GAP dead (GFP-GD) point mutation construct (GFP-ARAP3 C504A, R982A) [11,12].(E) Representative example of a western blot showing expression of endogenous or GFP-ARAP3 expression in b.End5 cells and a loading control.Note that two different exposures of the same blot are spliced in the ARAP3 panel to enable seeing endogenous ARAP3 without excessive background in lanes expressing the GFP-fusion proteins.Endogenous and GFP-ARAP3 constructs are indicated by arrowheads.(F and G) Endothelial permeability of vehicle or fMLF-stimulated wt control and 'rescue' Arap3 À/À b.End5 monolayers by RTCA.(F) Time course.(G) AUC.Data analysis was (B and D) by two-tailed t-test or by two-way ANOVA with (A, C, and G) Tukey's multiple comparison test or with (F) unprotected Fisher's least significant difference multiple comparison test.Comparisons within genotypes to unstimulated control in (A) and (C) are indicated by colour-matched symbols above bars.*p < 0.05; **p < 0.01; ns, not significant.Differences between genotypes did not reach significance.

Figure 5 .
Figure 5. Neutrophils promote formylated peptide-induced vascular leakage in vivo in ARAP3-deficiency.Tamoxifen-induced ERT2Cre + controls (wt) or Arap3 fl/fl ERT2Cre + (À/À) mice (A and B) were subjected to neutrophil depletion prior to being challenged with fMLF; BALf was collected 6 h later.Analysis of (A) total BALf protein and (B) BALf IgM.(C-E) Analysis of criss-cross chimeras.(C) Schematic representation, (D) total BALf protein, and (E) BALf serum albumin.(F-H) Animals were challenged with fMLF or vehicle i.t.; 6 h later (F) perfused lung single-cell digests were prepared as detailed in Materials and methods and analysed using flow cytometry.Lung neutrophils are plotted.(G and H) Representative images of (G) H&E-stained and (H) S100A9-labelled lung sections.Scale bars, 100 μm (G) and 50 μm (H).(A-F) Each symbol represents one experimental animal.Data obtained in at least two separately conducted experiments were amalgamated in these graphs.Data were analysed by (A and B) two-way ANOVA with Tukey's multiple comparison test and by (D-F) one-way ANOVA with Tukey's multiple comparison test.*p < 0.05; **p < 0.01; ***p < 0.001.

Figure 6 .
Figure 6.Arap3 À/À lungs contain NETs following fMLF challenge and mice develop more severe disease following influenza A virus (IAV) infection.(A-C) Mice were injected i.p. with inhibitors or their vehicle as indicated 30 min before i.t.fMLF or saline challenge.BALf was analysed for (A) total protein, (B) citrullinated histone H3, and (C) neutrophil elastase.(D-F) Influenza was induced with 20 PFU PR8 or saline given i.t. on day 0 and mice monitored thereafter as detailed in Materials and methods.(D) Body weights expressed as percentage of starting weight and (E) clinical scores are plotted.(F) Representative histology sections from lungs fixed at day 7 after infection.Arrowheads identify lung oedema; scale bar, 50 μm.(A-C) Comparison to fMLF-stimulated Arap3 À/À mouse are indicated; each symbol represents one mouse.(Dand E) Graphs depict mean ± SD (seven mice/group).Data were analysed using one-way ANOVA with Dunnett's multiple comparison test (A-C) and two-way ANOVA with Tukey's multiple comparison test (D and E).Comparisons were made to fMLF-stimulated wild-types (wts) (A-C).In (D) and (E), black asterisks denote comparisons between IAV and saline-treated controls; red asterisks denote comparisons between IAV and saline-treated Arap3 À/À mice; red number signs denote comparisons between IAV-treated Arap3 À/À and control mice.*, # p < 0.05; **p < 0.01; ***p < 0.001; ****, #### p < 0.0001.