Directed transport of neutrophil-derived extracellular vesicles enables platelet-mediated innate immune response

Abstract

The innate immune response to bacterial infections requires the interaction of neutrophils and platelets. Here, we show that a multistep reciprocal crosstalk exists between these two cell types, ultimately facilitating neutrophil influx into the lung to eliminate infections. Activated platelets adhere to intravascular neutrophils through P-selectin/P-selectin glycoprotein ligand-1 (PSGL-1)-mediated binding, a primary interaction that allows platelets glycoprotein Ibα (GPIbα)-induced generation of neutrophil-derived extracellular vesicles (EV). EV production is directed by exocytosis and allows shuttling of arachidonic acid into platelets. EVs are then specifically internalized into platelets in a Mac1-dependent fashion, and relocated into intracellular compartments enriched in cyclooxygenase1 (Cox1), an enzyme processing arachidonic acid to synthesize thromboxane A₂ (TxA₂). Finally, platelet-derived-TxA₂ elicits a full neutrophil response by inducing the endothelial expression of ICAM-1, intravascular crawling, and extravasation. We conclude that critical substrate-enzyme pairs are compartmentalized in neutrophils and platelets during steady state limiting non-specific inflammation, but bacterial infection triggers regulated EV shuttling resulting in robust inflammation and pathogen clearance.

Figure 1. Thromboxane generation by platelets is enabled by interacting with neutrophils.

(a) PMNs and platelets were isolated from WT mice and TxB₂ production in both cell types alone as well as after co-incubation under control conditions or stimulation with ADP (10 μM) and fMLP (10 μM) was analysed in the presence or absence of blocking antibodies against P-selectin (clone RB40.34. 5 μg ml⁻¹) or PSGL-1 (4RA10, 5 μg ml⁻¹) (n=3). (b) PMNs and/or platelets from WT mice were radioactively labelled with C¹⁴-AA and TxB₂-C¹⁴ in control and ADP/fMLP-stimulated samples was measured (n=3). PMNs and platelets were isolated from WT mice, Cox1^−/− and Cox2^−/− mice and only PMN were radioactively labelled with C¹⁴-AA. (c) TxB₂-C¹⁴ in control and ADP/fMLP-stimulated samples (n=3). (d) Total TxB₂ production in control and ADP/fMLP-stimulated samples (n=3). PMNs and platelets were isolated from WT mice and treated with blocking antibodies against P-selectin (clone RB40.34, 5 μg ml⁻¹), PSGL-1 (clone 4RA10, 5 μg ml⁻¹) or tirofiban (100 μM). (e) TxB₂-C¹⁴ in control and ADP/fMLP-stimulated samples (n=3). (f) Total TxB₂ production in control and ADP/fMLP-stimulated samples (n=3). (g) Cox1 activity in platelets alone and after co-incubation with neutrophils (n=3). Mean±s.e.m., ANOVA plus Bonferroni testing, *P<0.05.

Figure 2. Platelets are required host defence during E. coli-induced pneumonia.

Glycol- and busulfan-treated wild-type mice were injected intratracheally with saline or viable E. coli. (a) Survival 24 h after instillation of 8 × 10⁶ viable E. coli (n=11–15). (b) Neutrophil recruitment into the alveoli and the c.f.u. count in the BAL (c), lung tissue (d) and the spleen (e) were analysed 24 h after intratracheal instillation of 6 × 10⁶ c.f.u.'s per mouse (n=4). Neutrophil accumulation in the lung was visualized by intravital microscopy of the middle right lung lobe by intravital microscopy. (f) Number of accumulated neutrophils per field of view (FOV) (n=3). (g) Neutrophils interacting with platelets in the lung capillaries in vivo (n=3). (h) The formation of circulating platelet–neutrophil aggregates in the blood of glycol- and busulfan-treated mice after intratracheal instillation of saline or viable E. coli (6 × 10⁶ c.f.u.'s per mouse) was measured by flow cytometry (n=4). (i) Ultrathin cross-sectioned lung tissue imaged by transmission electron microscopy from lung tissue of WT mice after inducing pneumonia showing a neutrophil (*) in close proximity to a platelet (#) and 2 erythrocytes (§) within the boundaries of the capillary wall (black arrow). (j) Confocal image of lung tissue from WT mice after induction of E. coli pneumonia with (k) 3D reconstruction and (l) exemplary display of a single confocal plane to identify neutrophils (Ly6G, green) and platelets (CD41, red) within pulmonary capillaries stained with PECAM-1 antibody (gray) (scale bars equal 20 μm). Mean±s.e.m., ANOVA plus Bonferroni testing, *P<0.05.

Figure 3. Blocking thromboxane receptors aggravates E. coli-induced pneumonia.

WT control mice, mice after injection of a thromboxane receptor antagonist (SQ 29548), following PMN depletion or after administration of blocking antibodies against P-selectin (clone RB40.34, 50 μg per mouse) or PSLG-1 (clone 4RA10, 50 μg per mouse), were instilled i.t. with viable E. coli (6 × 10⁶ c.f.u.'s per mouse) or saline. (a) Serum TxB₂ levels in glycol- and busulfan-treated mice, after PMN depletion and after blocking P-selectin (clone RB40.34, 50 μg per mouse) or PSLG-1 (clone 4RA10, 50 μg per mouse) (n=4). (b) Neutrophil recruitment into the alveoli and the c.f.u. count in the BAL (c), lung tissue (d) and the spleen (e) were analysed after 6 h (n=4). (f) The number of accumulated neutrophils in the lung was visualized by intravital microscopy (n=3). (g) Survival 24 h after instillation of 8 × 10⁶ viable E. coli (n=6–15). Mean±s.e.m., ANOVA plus Bonferroni testing, log rank test in 3 g *P<0.05.

Figure 4. EV-mediated shuttling of arachidonic acid into platelets is necessary for host defence.

Isolated PMNs, PLTs or both were pretreated with blocking antibodies against P-selectin (clone RB40.34, 5 μg ml⁻¹) or GPIbα (clone Xia.B2, 5 μg ml⁻¹) and stimulated with ADP (10 μM) and fMLP (10 μM) at 37 °C for 30 min. (a) EVs were quantified in the supernatant (n=4). (b) The arachidonic concentration in the EV fraction was quantified by ELISA (n=4). (c) Uptake of fluorescently-labelled, isolated neutrophil EVs (labelled with a green fluorescent cell tracker) in isolated platelets labelled with a CD41-PE antibody (dilution 1:200) was analysed by confocal microscopy (exemplary micrograph, scale bar equals 10 μm). (d) Uptake of neutrophil-derived EVs into platelets (exemplary micrograph, scale bar equals 10 μm). (e) Colocalization of EVs and Cox1 in platelets (exemplary micrograph, scale bar equals 5 μm). PMNs and platelets were isolated from WT mice and treated with a blocking antibody against GPIbα. (f) TxB₂-C¹⁴ in control and ADP (10 μM)/fMLP (10 μM)-stimulated samples after pretreatment with a blocking GPIbα-antibody (clone Xia.B2, 5 μg ml⁻¹) with or without substitution of isolated EVs (n=3). (g) Total TxB₂ production in control and ADP/fMLP-stimulated samples (n=3). (h) Cox1 activity after treatment with isotype, blocking antibody against GPIbα (clone Xia.B2, 5 μg ml⁻¹) or blocking GPIbα antibody and isolated EVs (n=3). Wild-type mice pretreated with the isotype or blocking antibody against GPIbα (clone Xia.B2, 50 μg per mouse) were injected intratracheally with viable E. coli and (i) the amount of circulating platelet-neutrophil aggregates in the blood, (j) neutrophil recruitment into the alveoli and the c.f.u. count in the BAL (k), lung tissue (l) and the spleen (m) were analysed after 24 h (n=4). (n) Wild-type mice were pretreated with GPIbα blocking antibody (clone Xia.B2, 50 μg per mouse) and received isolated neutrophil EVs or control and survival was assessed 24 h after instillation of 8 × 10⁶ viable E. coli (n=7–8). Isolated PMNs and platelets were pre-incubated with a blocking antibody against GPIbα (clone Xia.B2, 5 μg ml⁻¹) or isotype control and stimulated with ADP (10 μM) and fMLP (10 μM) at 37°C for 30min. The concentration of LTB4 (o) and LTC4 (p) was measured in the supernatant (n=4). Mean±s.e.m., ANOVA plus Bonferroni testing, two-tailed t-test in 4b, log rank test in 4n *P<0.05.

Figure 5. EV shuttling involves directed release and uptake mechanisms.

(a) Caveolin-1 and Clathrin in supernatant and EV fraction from control and ADP (10 μM)/fMLP (10 μM)-stimulated samples was detected by western blot (exemplary blot from three experiments). (b) Isolated platelets and neutrophils (ration 1:10) were co-incubated and control samples and stimulated samples were pretreated with vehicle or 5 μM BAPTA-AM and the number of generated EVs was quantified (n=3). (c) Total TxB₂ production in control and ADP (10 μM)/fMLP (10 μM)-stimulated samples after pretreatment with vehicle or 10 μg ml⁻¹ chlorpromazine (n=3). (d) Mac-1 (CD11b) in supernatant and EV fraction from control and ADP (10 μM)/fMLP (10 μM)-stimulated samples was detected by western blot (exemplary blot from 3 experiments). (e) Total TxB₂ production in control and ADP (10 μM)/fMLP (10 μM)-stimulated samples after pretreatment with a blocking Mac-1 antibody (clone M1/70, 5 μg ml⁻¹) or antibody plus 10 μg ml⁻¹ chlorpromazine (n=3). (f) Platelet adhesion in fibrinogen-coated flow chambers in control and ADP/fMLP-stimulated samples after pretreatment with a blocking Mac-1 antibody (clone M1/70, 5 μg ml⁻¹) or antibody plus 10 μg ml⁻¹ chlorpromazine (n=3). (g) Western Blot of Mac-1 in platelets co-incubated with EV fraction from unstimulated and stimulated neutrophils (exemplary blot from three experiments). (h) Wild-type mice were pretreated with GPIbα blocking antibody (clone Xia.B2, 50 μg per mouse) and received isolated neutrophil EVs pretreated with or without a blocking Mac-1 antibody. Survival was assessed 24 h after instillation of 8 × 10⁶ viable E. coli (n=6–7). (i) Cox1 activity in WT neutrophils alone or after co-incubation with activated WT or Cox1^−/− platelets (n=5). Mean±s.e.m., ANOVA plus Bonferroni testing, log rank test in 5 h *P<0.05.

Figure 6. Cox1 modulates neutrophil recruitment during pneumonia.

MLMVEC were isolated from WT mice and co-incubated with platelets and PMNs and ICAM-1 mRNA expression (a) and ICAM-1 surface expression (b) were analysed (n=4). (c) Exemplary immunofluorescence staining of ICAM-1 on the surface of MLMVEC after co-incubation with platelets and PMNs (scale bar equals 50 μm). (d) ICAM-1 (green) and PECAM-1 (red) immunofluorescence staining in fixed lung sections from WT and Cox1^−/− mice after instillation of E. coli (scale bar equals 100 μm). WT mice or Cox1^−/− mice received a blocking ICAM-1 antibody (clone YN1, 50 μg per mouse) or isotype control and were injected intratracheally with viable E. coli (6 × 10⁶ c.f.u.'s per mouse). (e) Exemplary 3D confocal image of E. coli infected lung samples stained against PMNs (clone RB6-8C5, green) and PECAM (clone 390, red). Migration velocity (f) and distance (g) of PMNs in WT mice, Cox1^−/− mice and WT mice after pretreatment with a blocking ICAM-1 antibody (clone YN1, 50 μg per mouse) (n=3). (h) Neutrophil recruitment into the alveoli and the c.f.u. count in the BAL (i), lung tissue (j) and the spleen (k) were analysed after 24 h (n=4). Mean±s.e.m., ANOVA plus Bonferroni testing, *P<0.05.