Lysyl Oxidase Promotes Actin-Dependent Neutrophil Activation and Cytotoxicity Toward Retinal Endothelial Cells in Diabetes

Abstract

Activated neutrophils contribute to retinal endothelial cell (EC) death and capillary degeneration associated with early diabetic retinopathy (DR), a major vision-threatening complication of diabetes. However, the factors and mechanisms driving neutrophil activation and cytotoxicity in diabetes remain insufficiently understood. Here, we show that lysyl oxidase (LOX), a matrix cross-linking and stiffening enzyme that increases retinal EC susceptibility to activated neutrophils, simultaneously activates neutrophils in its soluble form. Specifically, treatment of diabetic mice with LOX inhibitor β-aminopropionitrile (BAPN) prevented the diabetes-induced increase in neutrophil activation (extracellular release of neutrophil elastase and superoxide) and cytotoxicity toward cocultured mouse retinal ECs. Mouse neutrophils and differentiated (neutrophil-like) human HL-60 cells treated with recombinant LOX alone exhibited significant activation and cytotoxicity. Mechanistically, this LOX-induced neutrophil activation was associated with biphasic F-actin remodeling, with the initial and rapid (∼10 min) F-actin depolymerization followed by a significant increase in F-actin polymerization and polarization. Preventing the initial F-actin depolymerization blocked LOX-induced neutrophil activation and cytotoxicity toward cocultured retinal ECs. Finally, this biphasic F-actin remodeling was found to be essential for LOX-induced membrane aggregation of azurophilic granule marker CD63 and NADPH organizer p47phox, which are associated with extracellular release of neutrophil elastase and superoxide, respectively. By revealing a previously unrecognized causal link between LOX and actin-dependent neutrophil activation in diabetes, these findings provide fresh mechanistic insights into the proinflammatory role of LOX in early DR that goes beyond its canonical matrix-stiffening effects.

Article highlights: Activated neutrophils kill retinal endothelial cells (ECs) in early diabetic retinopathy, but how neutrophils become activated in diabetes is not well understood. We found that lysyl oxidase (LOX), whose matrix-localized form activates retinal ECs, can also directly activate neutrophils in its soluble form. LOX-induced release of neutrophil elastase and superoxide is mediated by actin remodeling and membrane aggregation of azurophilic granules. The dual ability of LOX to activate neutrophils (in its soluble form) and retinal ECs (in its matrix-localized form) implicates it as a key proinflammatory target for early diabetic retinopathy.

Figure 1

LOX mediates diabetes-induced neutrophil activation and cytotoxicity toward retinal ECs. A: RT-qPCR analysis of mouse whole retina indicates that LOX inhibitor BAPN prevents the diabetes-induced increase (by 1.5-fold; P < 0.01) in LOX mRNA level (n = 6 mice/group; 10-week diabetes). B: Flow cytometry–based analysis of FITC-annexin V–labeled mRECs following a 16-h coculture with neutrophils isolated from ND mice, D mice, or D + BAPN mice (3 mg/kg body weight) (n = 6 mice/group; 10-week diabetes) revealed that the significant increase in mREC apoptosis caused by neutrophils from D mice is prevented by LOX inhibition. C: Representative Western blot bands and cumulative densitometric analysis of mouse neutrophils (n = 6 mice/group; 10-week diabetes) indicate that the diabetes-induced approximately twofold increase (P < 0.05) in NE protein expression is prevented by LOX inhibition (using BAPN). D: fMLP (10 nmol/L) stimulation of neutrophils isolated from ND, D, or D + BAPN mice (n = 6 mice/group; 10-week diabetes) and subsequent EnzChek elastase assay revealed that the diabetes-induced ∼40% increase (P < 0.05) in extracellular NE activity is prevented by LOX inhibition. E: fMLP (10 nmol/L) stimulation of neutrophils isolated from ND mice or 4-week D mice (i.e., 4 weeks’ duration of diabetes) (n = 7 mice/group) and subsequent EnzChek elastase assay indicate a lack of change in extracellular NE activity in 4-week D mice. F: Flow cytometry–based analysis of FITC-annexin V–labeled mRECs following coculture with mouse neutrophils isolated from ND or 4-week D mice (n = 7 mice/group) revealed no increase in neutrophil-mediated mREC apoptosis in shorter-term diabetes. G: RT-qPCR analysis of mouse whole retina indicates that LOX mRNA levels do not increase in 4-week D mice (n = 7 mice/group). Dot plots indicate mean ± SD.

Figure 2

LOX induces neutrophil activation and CD63 aggregation. A: Neutrophils isolated from ND mice were either left untreated or treated with recombinant mouse LOX (75 ng/mL for 6 h) in the presence or absence of brief fMLP stimulation (10 nmol/L). Fluorometric analysis of the supernatant/EnzChek elastase substrate mix revealed a 40% increase (P < 0.05) in extracellular NE activity in LOX-treated/fMLP-stimulated neutrophils. B: LOX-treated ND mouse neutrophils exhibited a twofold increase (P < 0.001) in adhesion to cultured mRECs. C: Flow cytometry quantification of the total number of FITC-annexin V–labeled mRECs following 16 h of coculture with LOX-treated or untreated ND neutrophils revealed a 1.5-fold increase (P < 0.001) in mREC apoptosis caused by LOX-treated neutrophils. D: Flow cytometry–based analysis of FITC-annexin V–labeled HRECs following 16 h of coculture with recombinant human LOX-treated (75 ng/mL for 6 h) or untreated dHL-60 cells in the absence or presence of NE inhibitor sivelestat (50 µmol/L) revealed that LOX-treated dHL-60 cells cause a significant increase in HREC apoptosis, which is blocked in the presence of sivelestat. E: dHL-60 cells were either left untreated or treated with LOX ± NE inhibitor sivelestat (50 µmol/L) for 6 h before brief stimulation with fMLP (10 nmol/L). Fluorometric analysis of the supernatant/EnzChek elastase substrate mix revealed that the LOX-induced ∼1.6-fold increase (P < 0.001) in extracellular NE activity is blocked by sivelestat. F: Untreated or LOX-treated dHL-60 cells were labeled with anti-CD63 (to visualize NE-containing granules [green]) and Phalloidin-594 (to visualize F-actin [red]). Representative confocal images and subsequent quantitative analysis, performed as depicted in the schematic, demonstrated that LOX causes a significant increase in colocalized CD63 aggregation and F-actin polarization (arrowheads). Column scatter plots indicate mean and distribution from ≥80 cells. G: AFM stiffness measurement of untreated or LOX-treated dHL-60 cells (n = 30) revealed that LOX induces significant softening (by ∼70%; P < 0.001) of dHL-60 cells. Dot plots indicate mean ± SD. Box plot represents the 25th to 75th percentiles. The black line and square within the box represent the median and mean, respectively. Whiskers extend to the minimum and maximum values. Scale bar, 5 µm. Sivel, sivelestat; Unt., untreated.

Figure 3

LOX induces rapid and transient actin remodeling in neutrophils. A: dHL-60 cells were either left untreated or treated with LOX (75 ng/mL) for the indicated durations prior to labeling with Phalloidin-594 to visualize F-actin. Representative fluorescence images and subsequent quantitative analysis revealed that LOX causes a rapid and transient decrease in F-actin intensity followed by a significant increase at 360 min (6 h). Line graph indicates mean ± SD from ≥200 cells; intensity was normalized with respect to untreated cells. B: Untreated or LOX-treated (6 h) dHL-60 cells were labeled with Phalloidin-594 (red) and anti-CD63 (to visualize NE-containing granules [green]). Representative confocal images and subsequent quantitative analysis from ≥90 cells demonstrated that LOX-induced CD63 aggregation is spatiotemporally associated with F-actin polarization (arrowheads) at 360 min (6 h). C: dHL-60 cells were treated with 75 ng/mL LOX either continuously for 360 min (6 h) or transiently for 10 min prior to labeling with Phalloidin-594 at the indicated time points. Representative fluorescence images and subsequent quantitative analysis from ≥110 cells revealed that removal of LOX after 10 min does not inhibit F-actin repolymerization seen at 6 h. F-actin intensity was normalized with respect to the value at 0 min. Dot plots indicate mean ± SD. Scale bar, 5 µm. a.u, arbitrary unit; Cont., continuous; Norm., normalized; Trans., transient; Unt., untreated.

Figure 4

Transient actin depolymerization is necessary for LOX-induced neutrophil activation. A and B: dHL-60 cells were treated with 75 ng/mL LOX ± actin stabilizing drug JASP (1 µmol/L) for the indicated durations prior to labeling with Phalloidin-594 to visualize F-actin. Representative fluorescence images (A) and subsequent quantitative analysis (B) revealed that the LOX-induced F-actin remodeling is blocked by JASP. Line graph in B indicates mean ± SD from ≥80 cells; intensity was normalized with respect to untreated dHL-60 cells. $P < 0.001 for untreated vs. LOX treatment; #NS for untreated vs. LOX + JASP treatment. Dot plot indicates mean ± SD from ≥65 cells. C: AFM stiffness measurement of dHL-60 cells (n = 40) that were either left untreated or treated with 75 ng/mL LOX ± actin stabilizing drug JASP (1 µmol/L) for 6 h revealed that the LOX-induced softening of dHL-60 cells is significantly inhibited (by 40%; P < 0.05) by JASP. D: Untreated or LOX ± JASP–treated (6 h) dHL-60 cells were labeled with Phalloidin-594 (red) and anti-CD63 (to visualize NE-containing granules [green]). Representative confocal images and subsequent quantitative analysis from ≥70 cells revealed that JASP treatment concurrently blocks LOX-induced F-actin polarization and CD63 aggregation (arrowheads). E: Untreated or LOX ± JASP–treated (6 h) dHL-60 cells were briefly stimulated with fMLP (10 nmol/L). Fluorometric analysis of the supernatant/EnzChek elastase substrate mix revealed that the LOX-induced increase in extracellular NE activity is significantly inhibited (by ∼70%; P < 0.05) by JASP. F: HRECs were cocultured with LOX ± JASP–pretreated or untreated dHL-60 cells for 16 h prior to FITC-annexin V/propidium iodide labeling and flow cytometry. Quantification of the total number of annexin V–positive HRECs (plotted as percentage of untreated) revealed that the increase in HREC apoptosis caused by LOX-pretreated dHL-60 cells is significantly inhibited (by 70%; P < 0.001) by JASP. Dot plots indicate mean ± SD. Box plots represent the 25th to 75th percentiles. The black line and square within the box represent the median and mean, respectively. Whiskers extend to the minimum and maximum values. Scale bar, 5 µm. Unt., untreated.

Figure 5

Transient actin depolymerization is sufficient to cause neutrophil activation and cytotoxicity. A and B: dHL-60 cells were either left untreated or treated with the actin depolymerization drug cytoD (2.5 µmol/L) for the indicated durations prior to labeling with Phalloidin-594 to visualize F-actin. Representative fluorescence images and subsequent quantitative analysis revealed that cytoD causes a transient decrease in F-actin intensity followed by a significant increase in F-actin polymerization and polarization at 360 min (6 h). Line graph indicates mean ± SD from ≥200 cells; intensity was normalized with respect to untreated cells. Dot plot indicates mean and distribution from ≥70 cells. C: Untreated or cytoD-treated (6 h) dHL-60 cells were labeled with Phalloidin-594 (red) and anti-CD63 (to visualize NE-containing granules [green]). Representative confocal images and subsequent quantitative analysis from ≥65 cells revealed that cytoD treatment leads to a significant increase in colocalized F-actin polarization and CD63 aggregation (arrowheads) at 360 min (6 h). D: Untreated or cytoD-treated (6 h) dHL-60 cells were briefly stimulated with fMLP (10 nmol/L). Fluorometric analysis of the supernatant/EnzChek elastase substrate mix revealed that cytoD-treated dHL-60 cells exhibit a significant increase (by 30%; P < 0.01) in extracellular NE activity when compared with untreated cells. E: HRECs were cocultured with cytoD-pretreated or untreated dHL-60 cells for 16 h prior to FITC-annexin V/propidium iodide labeling and flow cytometry. Quantification of the total number of annexin V–positive HRECs (plotted as percentage of untreated) revealed that cytoD-treated dHL-60 cells cause a significant increase (by ∼40%; P < 0.05) in HREC apoptosis. Dot plots indicate mean ± SD. Box plots represent the 25th to 75th percentiles. The black line and square within the box represent the median and mean, respectively. Whiskers extend to the minimum and maximum values. Scale bar, 5 µm. Unt., untreated.

Figure 6

LOX inhibition blocks diabetes-induced F-actin remodeling and CD63 aggregation in neutrophils. Mouse bone marrow neutrophils isolated from ND mice, D mice, or D + BAPN mice (3 mg/kg body weight) (n ≥ 90 cells/group; 10-week diabetes) were labeled with Phalloidin-594 (to visualize F-actin [red]) and anti-CD63 (to visualize NE-containing granules [green]). A and B: Representative confocal images and subsequent quantitative analysis from ≥90 cells revealed that LOX inhibition using BAPN prevents the diabetes-induced concurrent increase in F-actin intensity, F-actin polarization, and CD63 aggregation. Arrowheads indicate F-actin polarization or CD63 aggregation. C: AFM stiffness measurement of bone marrow neutrophils (n ≥15 cells) isolated from ND, D, or D + BAPN mice (3 mg/kg body weight) (n = 6 mice/group; 10-week diabetes) revealed that the diabetes-induced reduction in neutrophil stiffness (by 40%; P < 0.01) is significantly inhibited (by ∼50%; P < 0.05) by LOX inhibitor BAPN. Dot plots indicate mean ± SD. Scale bar, 5 µm.

Figure 7

Schematic illustration of LOX-induced regulation of neutrophil actin dynamics, activation, and cytotoxicity in diabetes. Based on our current findings, we propose that LOX is a mechanical determinant of neutrophil activation wherein it alters actin cytoskeleton and cell stiffness to increase extracellular NE and superoxide release that, together, exert cytotoxic effects on retinal ECs. This newly identified mechanism of neutrophil activation begins with LOX-induced rapid (within ∼10 min) F-actin depolymerization into G-actin. At this early time, the membrane distribution of CD63 and p47^Phox, which reflects the cell’s ability to release extracellular NE and superoxide, respectively, remains uniform and unchanged. However, over time (6 h), as F-actin recovers through repolymerization, it simultaneously becomes polarized. Interestingly, this late-phase F-actin remodeling causes redistribution (aggregation) of CD63 (and associated azurophilic granules) and p47^Phox that predictably leads to increased extracellular release of NE and superoxide, respectively, the key cytotoxic mediators of retinal EC death in diabetes.