and the clearance of senescent hepatocytes in
hepatocarcinoma ( 68 ). Induction of cellular
senescence in hepatic carcinomas through re-
activation of p53 promotes tumor regression
through the elimination of senescent tumor
cells by the innate immune system, and it has
been suggested that NK cells, macrophages,
and neutrophils contribute to similar degrees
through an unexplored mechanism ( 70 ). Our
findings suggest that prolonged exposure of
senescent ECs to NETs directly triggers a cyto-
toxic response leading to apoptotic death.
This likely occurs from direct exposure to pro-
tein constituents such as histones or MPO ( 54 ).
Ultimately, neutrophil-mediated NETosis might
facilitate further clearance through innate im-
mune phagocytosis.
Quantitative proteomics using SILAC revealed
that senescent ECs predominantly induce a
secretome involved in immune regulation. Sub-
sets of ECs in retinopathy adopt a SASP and
excrete cytokines such as CXCL1, IL-1b,and
others, similar to what has been described for
other senescent cell populations ( 53 ). Inhibi-
tion of these cytokines or their receptors cur-
tails senescence-induced NETosis and impedes
neovascular tuft regression, highlighting the
role of these inflammatory factors in vascularremodeling. We cannot exclude that other
mediators of the SASP response, such as re-
active oxygen species, might contribute to
NETosis. In addition to ECs, vascular-associated
cells such as pericytes and Müller glia also
trigger a SASP.
Although removal of diseased senescent vas-
culature through a neutrophil-mediated pro-
cess may be an intrinsic reparative mechanism
that favors tissue remodeling after vascular
lesions, uncontrolled retinal neutrophila and
abundant NETosis in the retina may be in-
compatible with proper retinal health. Retinal
NETosis may potentially exacerbate retinalBinetet al.,Science 369 , eaay5356 (2020) 21 August 2020 9of13
Fig. 6. NETs promote the regression of pathological vasculature in retinop-
athy.(A) Injections of DNAse I or anti-Ly6G at P17 of OIR prevented regression
of preretinal NV, leaving ~1.8-fold higher NV with DNAse I and ~1.5-fold higher
NV with anti-Ly6G at P19. (B) A similar decrease in vascular regeneration was
observed with ~1.5-fold higher avascular areas with DNAse I and ~1.7-fold higher
avascular areas with anti-Ly6G (n= 8-18 retinas). (C) LysM-cre+/+Pad4−/−and
LysM-cre+/+Pad4+/+mice developed similar amounts of preretinal NV and showed
similar rates of vascular regeneration (D)atP17ofOIR,butLysM-cre+/+Pad4−/−
mice showed retarded clearance of these vascular tufts at P19 (5.2-fold higher
NV) (E) and ~3.7-fold larger avascular areas (F)(n=10-14retinas)Scalebars,
0.5 mm. Data were analyzed with Student’sttest (*P< 0.05, **P< 0.01, ***P<
0.001) and are shown as means ± SEM.RESEARCH | RESEARCH ARTICLE