Autophagy modulates endothelial junctions to restrain neutrophil diapedesis during inflammation

Abstract

The migration of neutrophils from the blood circulation to sites of infection or injury is a key immune response and requires the breaching of endothelial cells (ECs) that line the inner aspect of blood vessels. Unregulated neutrophil transendothelial cell migration (TEM) is pathogenic, but the molecular basis of its physiological termination remains unknown. Here, we demonstrated that ECs of venules in inflamed tissues exhibited a robust autophagic response that was aligned temporally with the peak of neutrophil trafficking and was strictly localized to EC contacts. Genetic ablation of EC autophagy led to excessive neutrophil TEM and uncontrolled leukocyte migration in murine inflammatory models, while pharmacological induction of autophagy suppressed neutrophil infiltration into tissues. Mechanistically, autophagy regulated the remodeling of EC junctions and expression of key EC adhesion molecules, facilitating their intracellular trafficking and degradation. Collectively, we have identified autophagy as a modulator of EC leukocyte trafficking machinery aimed at terminating physiological inflammation.

Keywords: ATG16L1; ATG5; PECAM-1; autophagy; diapedesis; endothelium; extravasation; inflammation; junctions; neutrophils.

Graphical abstract

Figure 1

Acutely inflamed microvascular ECs exhibit induction of autophagy within vascular junctions (A–E) GFP-Map1lc3a^TG/+ or WT mice were subjected to local IR injury (A) Neutrophil extravasation at the indicated times post reperfusion, (n = 3–6 mice/group). (B–E) Representative confocal images (n = 6) of postcapillary venules (PCVs, PECAM-1), with arrows indicating EC junctional localization of GFP-LC3 puncta (scale bar, 5 μm) (B) and quantification of (C) GFP-LC3 puncta or (D) endogenous LC3 puncta per venular EC area at 4 h and (E) at the indicated times postreperfusion (n = 3–6 mice/group). (F–H) GFP-Map1lc3^TG/+ mice were treated intrascrotally (i.s.) with PBS or LPS. (F and G) Neutrophil extravasation (n = 3 mice/group) (F) and (G) representative (n = 3) confocal images of cremasteric PCVs (PECAM-1), with arrows indicating EC junctional localization of GFP-LC3 puncta (scale bar, 5 μm). (H) Quantification of GFP-LC3 puncta per venular EC area (n = 3 mice/group). Dashed boxes delineate magnified areas. Means ± SEMs. Statistically significant difference from controls or between indicated groups is shown by ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001; ns, not significant. See also Figure S1.

Figure 2

Modulation of EC autophagy controls neutrophil extravasation and cellular tissue damage (A and B) Neutrophil extravasation in chimeric Atg5^fl/fl and Atg5^ΔEC mice (A) treated i.s. with PBS or LPS and (B) subjected to local IR injury (n = 3–8 mice/group). (C) Representative (n = 8) confocal images of IR-stimulated cremasteric PCVs (PECAM-1) immunostained for neutrophils (MRP14) (scale bar, 30 μm). (D) Neutrophil extravasation in Atg5^fl/fl and Atg5^iΔEC mice subjected to local IR injury (n = 6 mice/group). (E) Propidium iodide (PI)⁺ cells in Atg5^fl/fl and Atg5^ΔEC mice subjected to IR injury, as quantified by confocal IVM (n = 3–5 mice/group). (F and G) GFP-LC3 puncta or endogenous LC3 puncta per venular EC area in cremasteric PCVs of (F) GFP-Map1lc3^TG/+ mice and (G) Atg5^fl/fl and Atg5^ΔEC mice treated i.s. with scrambled or Tat-Beclin 1 peptide (n = 3–5 mice/group). (H and I) Neutrophil extravasation in (H) WT mice and (I) Atg5^fl/fl and Atg5^iΔEC mice subjected to local IR injury and treated i.s. with scrambled or Tat-Beclin 1 (n = 4–6 mice/group). (J and K) Intravascular neutrophils in WT mice subjected to local IR injury and treated i.s. with scrambled or Tat-Beclin 1 (n = 3–4 mice/group) (J), and (K) representative confocal images (n = 3–6) of cremasteric PCVs (PECAM-1) immunostained for neutrophils (MRP14) (scale bar, 30 μm). Means ± SEMs. Statistically significant difference from controls or between indicated groups is shown by ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001; ns, not significant. See also Figure S2.

Figure 3

EC ATG5 deficiency promotes increased and faster neutrophil TEM (A–C) Chimeric Atg5^fl/fl and Atg5^ΔEC mice were subjected to local IR injury and neutrophil responses in cremasteric PCVs analyzed by confocal real-time IVM. Neutrophil (A) adhesion, (B) TEM, and (C) duration of TEM. (D) Representative confocal images (n = 7) of cremasteric PCVs (PECAM-1) immunostained for neutrophils (MRP14), highlighting mosaic distribution of ECs, with arrows indicating Atg5^−/− and WT junctions (scale bars, 15 μm). (E–G) Neutrophil (E) paracellular and (F) transcellular TEM and (G) TEM through hotspots in WT and Atg5^−/− junctions of chimeric Atg5^ΔEC mice. (H and I) Pore (H) opening duration and (I) number within chimeric Atg5^ΔEC mice (n = 5–7 mice/group). (J) Representative (n = 5) confocal images illustrating an Atg5^−/− junction exhibiting multiple PECAM-1 pores (arrows; scale bars, 15 μm). Dashed boxes delineate magnified areas. Means ± SEMs. Statistically significant difference from controls or between indicated groups is shown by ^∗p < 0.05 and ^∗∗p < 0.01; ns, not significant. See also Figure S3.

Figure 4

ATG5-dependent autophagy regulates the architecture and molecular composition of EC contacts (A–F) Atg5^fl/fl and Atg5^ΔEC mice were subjected to local IR injury. (A and B) Representative confocal images (n = 3–4) of cremasteric PCVs (PECAM-1 and VE-cadherin) showing aberrant, thickened junctional structures (arrowheads) (A) and (B) frequency of thickened junctions (n = 3–4 mice/group). (C–F) Quantification of PECAM-1 and VE-cadherin (C and D) junctional width and (E and F) junctional enrichment (n = 3 mice/group). (G and H) Cell surface proteins under (G) basal and (H) endotoxemia conditions in WT and ATG5-deficient lung ECs from Atg5^ΔEC mice (n = 3–8 mice/group). (I–N) Correlative light electron microscopy (CLEM) analysis of a venular segment in an IR-stimulated Atg5^ΔEC mouse (n = 1). (I) Serial-block face scanning electron microscopy (SBF-SEM) micrograph of the region of interest (ROI) (targeted as shown in Video S3), illustrating segmentation of ECs. (J) Confocal image showing WT and Atg5^−/− ECs within the ROI, with the latter exhibiting thickened PECAM-1 junctional structures (arrowhead and asterisk). (K) 3D reconstruction of segmented ECs and cell-cell contacts of the venular area depicted in (J). (L–N) Enlargements of the 3D model illustrating (L) WT-WT and (M and N) Atg5^−/−-Atg5^−/− cell contacts showing areas of enlarged contacts (arrowhead) and membrane flaps (asterisk, scale bars, 1 μm). (O) Neutrophil extravasation in Atg5^fl/fl and Atg5^ΔEC mice subjected to IR injury and treated with an isotype control or anti-PECAM-1-blocking mAb (n = 3–4 mice/group). Dashed boxes delineate magnified areas. Means ± SEMs. Statistically significant difference from controls or between indicated groups is shown by ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, and ^∗∗∗∗p < 0.0001. Scale bars, 5 μm, unless otherwise specified. See also Figure S4.

Figure 5

EC autophagy machinery regulates cell-surface PECAM-1 intracellular trafficking and degradation (A–C) Schematic illustrating cell-surface biotinylation method. Control and ATG5 siRNA-silenced HUVECs were stimulated with LPS (A) and (B) immunoblotted for total and biotinylated PECAM-1, ATG5, and β-actin at the indicated times postbiotin incubation and (C) analyzed for fold change in cell surface-derived PECAM-1 (n = 7). (D–G) GFP-LC3 transfected HUVECs were stimulated with LPS before antibody feeding using a nonblocking anti-PECAM-1 mAb. (D) Number of GFP-LC3⁺/ PECAM-1⁺ vesicles at the indicated times after incubation with anti-PECAM-1 mAb (n = 3–5; 40–100 cells analyzed per condition). (E) Time-lapse confocal images (Video S6) showing the formation of GFP-LC3⁺/PECAM-1⁺ vesicles (scale bars, 10 μm). (F and G) Representative (n = 4) confocal images of GFP-LC3 transfected HUVECs immunostained for PECAM-1 and WIPI2 (scale bars, 10 μm and enlargements, 3 μm) (F) and (G) quantification of the number of GFP-LC3⁺/ PECAM-1⁺ vesicles WIPI2^-/+ (n = 4; >100 cells analyzed per condition). Means ± SEMs. Statistically significant difference from controls is shown by ^∗p < 0.05 and ^∗∗p < 0.01. See also Figure S5.

Figure 6

Non-canonical autophagy operates in microvascular ECs and regulates PECAM-1 distribution in response to IR injury (A and B) GFP-Map1lc3^TG/+ mice were subjected to local IR injury. (A and B) Representative (n = 3) confocal images of cremasteric PCVs (PECAM-1) immunostained for WIPI2, showing GFP-LC3⁺ and WIPI2⁺ puncta (scale bar, 5 μm) (A) and (B) number of GFP-LC3⁺/WIPI2^-/+ puncta per venular EC area (n = 3 mice/group). (C and D) Atg5^ΔEC, WT, and Atg16L1^E230 mice were subjected to local IR injury. (C) Representative (n = 3–5) confocal images of cremasteric PCVs (PECAM-1) immunostained for endogenous LC3, with arrows indicating localization of LC3 puncta (scale bar, 5 μm). (D) Number of LC3 puncta per venular EC area (n = 3–5 mice/group). (E–G) WT and Atg16L1^E230 mice were subjected to local IR injury. (E) Representative (n = 3) confocal images of cremasteric PCVs (PECAM-1, VE-cadherin) (scale bars, 5 and 3 μm for enlargements) and associated quantification of PECAM-1. (F and G) Junctional width (F) and (G) junctional enrichment (n = 3 mice/group). Means ± SEMs. Statistically significant difference from controls is shown by ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗∗p < 0.0001. See also Figure S6.

Figure 7

Genetic ablation of vascular autophagy (canonical and non-canonical) promotes exaggerated and more rapid leukocyte trafficking (A and B) Chimeric Atg5^fl/fl and Atg5^ΔEC mice were subjected to (A) LPS-induced peritonitis (n = 6–12 mice/group) or (B) skin inflammation (n = 5–6 mice/group), and neutrophil infiltration was quantified by flow cytometry or myeloperoxidase (MPO) activity, respectively. (C) Atg5^fl/fl and Atg5^iΔEC mice were subjected to LPS-induced peritonitis, and neutrophil infiltration was quantified by flow cytometry (n = 4–9 mice/group). (D) WT and Atg16L1^E230 mice were subjected to MDP-induced peritonitis, and neutrophil infiltration was quantified by flow cytometry (n = 4–5 mice/group). (E and F) Non-chimeric (E) and (F) chimeric WT and Atg16L1^E230 mice were subjected to local IR injury, and neutrophil extravasation was assessed by confocal microscopy (n = 6–8 mice/group). (G–I) Chimeric Atg5^fl/fl and Atg5^ΔEC mice were subjected to zymosan-induced peritonitis, and infiltration of (G) neutrophils, (H) monocytes, and (I) eosinophils at the indicated times was quantified by flow cytometry (n = 3–9 mice/group). Means ± SEMs. Statistically significant difference from controls or between indicated groups is shown by ^∗p < 0.05, ^∗∗p < 0.01, and ^∗∗∗p < 0.001.