Introduction
Retinopathy of prematurity (ROP), which often presents with bronchopulmonary dysplasia (BPD), is among the most common morbidities, affecting approximately 60% of very low-birthweight infants needing oxygen therapy [
1,
2]. ROP is the leading cause of childhood vision impairment and blindness worldwide [
3‐
5]. The pathological hallmarks of ROP are characterized in 2 phases, with delayed vascular development in phase 1 and intravitreal neovascularization in phase 2 [
1,
2]. Laser and anti-VEGF therapy are current treatments for severe ROP [
6,
7]. Laser therapy can reduce intravitreal neovascularization and decrease the possibility of progression to retinal detachment. However, some reported side effects of laser therapy include induced myopic refractive error and visual field loss [
6,
8]. Anti-VEGF therapy is considered in some forms of severe ROP, but VEGF also plays critical physiological functions in vital organ development, such as in the lungs and brains of these infants [
9,
10]. It has been reported that anti-VEGF therapy for ROP may have side effects on system vascular development and adverse effects on brain and lung development [
9,
11,
12]. Thus, new effective, preventive, and treatment strategies with fewer side effects are needed.
The mouse model of oxygen-induced retinopathy (OIR) is useful for testing the effects of high oxygen exposure on both the initial developing neonatal retinal vasculature and subsequent intravitreal neovascularization. First high oxygen damages newly developed capillaries and then following the return to room air (RA), the avascular retina stimulates the development of intravitreal tufts. Recent studies report that neurons, astrocytes, microglia inflammation, and pyroptosis contribute to vasculopathy due to elevated reactive oxygen species (ROS) and inflammation in OIR models [
13‐
15]. Most recently, researchers reported that retinal microglia play important roles in the protection of vascular damage in a mouse model of OIR [
16].
Inflammasomes are multi-protein complexes that mediate proteolytic cleavage of gasdermin D (GSDMD), pro-IL-1β, and pro-IL-18 by caspase-1 [
17‐
19]. Apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) is pivotal in inflammasome assembly and activation of caspase-1 [
20,
21]. A hallmark of inflammasome activation is ASC assembly into large oligomer protein complexes, called ASC specks (~ 1–2 µm). Macromolecular ASC speck formation by oligomerization of pyrin domains (PYD) creates a multitude of activation sites for caspase-1. Activated caspase-1 cleaves GSDMD to release a 30-kDa N-terminal domain, GSDMD-p30, which oligomerizes in the cell membrane to form pores that cause pyroptosis [
22,
23]. In addition, the GSDMD pores allow rapid release of active IL-1β and IL-18, resulting in secondary inflammation [
24]. Moreover, inhibition of the NLRP3 inflammasome with MCC950 suppresses the development of retinal neovascularization in OIR mice [
25]. We have reported that GSDMD gene knockout prevents both vaso-obliteration and intravitreal neovascularization in OIR mice [
26]. However, there are no reports on the effects of ASC inhibitory strategies in treating ROP or OIR mice.
IC100 is a humanized IgG4 monoclonal antibody that targets ASC, disrupts ASC oligomerization and speck formation, and inhibits IL-1β release [
27]. It has been reported that IC100 suppresses the immune-inflammatory response in experimental autoimmune encephalomyelitis models of multiple sclerosis and traumatic brain injury [
27,
28].
In this study, we tested the hypothesis that hyperoxia induces ASC speck formation, which leads to microglial activation and retinopathy, and that inhibition of ASC speck formation by IC100 will ameliorate microglial activation and abnormal retinal vascular formation. To test these hypotheses, we used the mouse OIR and BPD models and provided insights into the mechanistic functions of ASC in the pathogenesis of OIR and the utility of ASC inhibition.
Discussion
In this study, we found that hyperoxia exposure induced ASC specks in the mouse retinas. In an OIR mouse model, we showed that inhibition of ASC speck formation by IC100, a humanized monoclonal anti-ASC antibody, by IVT or IP injection improved vaso-obliteration and intravitreal neovascularization in OIR retinas. These vascular changes were associated with reduced microglial activation, restored retinal structure, and improved retinal function. These structural and functional improvements by IC100 correlated with corrections of hyperoxia-modulated gene pathways associated with eye development, angiogenesis, inflammatory response, and neurogenesis. To the best of our knowledge, this study is the first to demonstrate the efficacy of IC100 in successfully treating OIR mice. Given that IC100 is a humanized antibody, and it is being manufactured for possible use in clinical trials, it may have a potential therapeutic use in the treatment of preterm infants with ROP.
ROP, characterized by avascular retina and intravitreal neovascularization, continues to be a major cause of childhood vision impairment and blindness [
1,
4] and current therapies using laser and VEGF inhibitors have unwanted side effects [
6‐
9]. Anti-VEGF therapy by IVT injection is an effective therapy with several advantages compared with laser therapy, including the potential for increased visual field and lower degrees of refractive error [
7]. However, there is a paucity of safety data about the effects on organ systems beyond the eye. IVT bevacizumab lowers serum VEGF levels for up to 8 weeks [
38], and thus may inhibit vascular development in rapidly developing organs such as the lungs and brain of neonatal infants. In fact, anti-VEGF has been reported to be associated with lower motor scores and higher rates of neurodevelopmental disability in children [
39,
40]. Thus, novel, safe, and effective therapies are needed for treating ROP.
Inflammasomes are multi-protein complexes that mediate proteolytic cleavage of GSDMD, pro-IL-1β, and pro-IL-18 by caspase-1 [
18]. ASC is pivotal in inflammasome assembly and activation of caspase-1, which subsequently activates GSDMD that leads to pyroptosis, and fast release of IL-1β and IL-18 that results in inflammation [
21]. There is increasing evidence that inflammasome activation plays a role in ROP. NLRP3 inflammasome activation is linked to OIR in mice [
41], and treatment with MCC950, an inhibitor of NLRP3, reduced the pathological neovascularization in OIR mice [
25]. We recently published a study showing that GSDMD gene knockout ameliorated hyperoxia-induced retinal injury including, vaso-obliteration and intravitreal neovascularization, retinal inflammation, retinal layer thinning, and transcriptional regulation of retinal gene pathways related to inflammation, cell death, tissue remodeling, and vascular development [
26]. However, there is no report on whether ASC plays a role in OIR or if inhibition of ASC reduces retinal damage in OIR mice.
Herein, we tested if hyperoxia activates ASC in the developing vasculature and retinas. ASC speck formation is widely used as a readout for detecting inflammasome activation [
42]. We utilized ASC-citrine reporter mice to examine if hyperoxia activates ASC speck formation in the retinas. We first exposed these mice to RA and 85% O
2 from P1 to P14, which commonly used to induce BPD and multi-organ injuries in neonatal mice [
26,
36]. We found that ASC specks were increased in hyperoxia-exposed lungs and brains in these mice (data not shown). When we examined the retinas, we found no ASC specks in RA-exposed mice. However, the hyperoxia-exposed retinas had increased ASC specks associated with disorganized vasculature. We also found a similar increase in retinal ASC specks and vascular association when neonatal ASC-citrine reporter mice and wildtype mice were exposed to 75% O
2 from P7 to P11, which is commonly used in mice to induce ROP-like OIR models. Moreover, ASC specks were colocalized with both M1 and M2 activated microglia in the intravitreal neovascular areas of the OIR retinas. Additional studies with the OIR model in wildtype mice also detected ASC specks in retinal ganglion cells and microglia cells, as well as the vitreous fluid of the OIR mice. These novel data highlight that ASC is activated in the retinas of OIR mice, which sets a foundation for us further test the efficacy of ASC inhibition in treating OIR mice.
We tested IC100 for treatment of mouse OIR by IVT injection at P12 and IP injection at P12, P14, and P16 during the post hyperoxia exposure and room air recovery period. We performed detailed assessments of intravitreal vascularization, microglial expression, retinal layer thickness, retinal function, and transcriptome profiling. We have demonstrated that IC100 effectively treated both phases of OIR as IC100-IVT and IC100-IP had decreased retinal vaso-obliteration and intravitreal vascularization, with IC100-IVT being more effective. Similar to our recent publication on the role of GSDMD in OIR, we found that IC100 treatment reduced hyperoxia-induced retinal inflammation as assessed by reduced numbers of activated microglial cells detected by AIF-1 (M1 microglia/macrophages) and CD206 (M2 microglia/macrophages) in the retinas. Similar findings of M1 and M2 microglial cells both existing in OIR mouse retinas have been previously reported [
33]. Microglial cells play active roles in maintaining the normal structure and functioning of the retina under normal physiological conditions. Microglia become pathologically activated in a chronic pro-inflammatory environment and release excessive inflammatory mediators that promote retina damage and disease progression [
43‐
45]. Preventing chronic microglial activation by IC100 would certainly reduce retina injury and the progression of OIR caused by chronic hyperoxia exposure.
Previous studies have demonstrated that mouse OIR models have reduced retina thickness in multiple layers [
26,
46]. We sought to understand the role of ASC in retinal structure development by analyzing retinal thickness. We found that under hyperoxia exposure, the placebo-treated mice had retinal thinning compared to the IC100-IVT and IC100-IP-treated hyperoxia-exposed retinas, with major differences observed in the INL, ONL and total retinal thickness. We also observed disorganized GCL with missing ganglion cells in oxygen-exposed, placebo-treated retinas. However, treatment with IC100 significantly reduced the thinning of these layers and improved the organization of the GCL. The INL has second order neuron bipolar cells, and the ONL has cone photoreceptors [
47]. Bipolar cells separate visual signals evoked by light and dark contrasts and encode them to ON and OFF pathways, respectively [
48]. Photoreceptors are specialized neurons that convert light into electrical signals that stimulate physiological processes. Signals from the photoreceptors are sent through the optic nerve to the brain for processing [
49]. Thus, any therapeutic that reduces thinning of these tissue layers and improves GCL organization as we have shown that IC100 does, should help maintain or restore retina structural development and improve ROP functional outcomes.
To that end, we tested if decreased pathological vascularization, reduced inflammation, and restored retinal layer development by IC100 treatment led to improved retinal function as assessed by PERG in our OIR model. We found that placebo injected hyperoxia-exposed retinas had drastically decreased amplitude, however, treatment with IC100 by IVT injection significantly increased amplitude in the hyperoxia-exposed mice. Previous studies in OIR mice have shown that this model not only has abnormal vasculature, but also significantly reduced neuronal function [
30,
50]. In a study similar to ours it was recently reported that a humanized monoclonal antibody Fab fragment to the angiogenic factor secretogranin III, effectively corrected retinal dysfunction in OIR mice [
30]. These data suggest an potential role for antibodies against both inflammasome-related molecules and angiogenic factors in regulating retinal functions under hyperoxia that result in ROP.
Our study provided molecular insights into the underlying mechanisms by which IC100 treatment improves retinal structure and function in this OIR model. We showed by qRT-PCR that IC100 specifically downregulated hyperoxia-induced gene expression of inflammasome-related molecules such as Asc, Gsdmd, and Il1b, inflammatory mediators, Il6 and Tnf, and an angiogenic factor, Vegf. Our retinal RNA-seq data provided a better characterization of how IC100 affects hyperoxia-regulated transcriptomes and biological pathways related to OIR. We found that hyperoxia upregulated and downregulated distinctive gene pathways in the PBS-treated retinas but IC100 treatment corrected some of these gene pathways under hyperoxia. In O
2-PBS retinas, hyperoxia-induced gene pathways in cluster 4 were corrected by IC100 to about room air levels, including extracellular structure organization, mesenchyme development, regulation of angiogenesis, regulation of eye development, endothelial cell differentiation and proliferation, and leukocyte migration. Some specific genes in these pathways included TGF beta induced (Tgfbi) [
51], collagen type III alpha 1 chain (Col3a1) [
52], lysyl oxidase like-4 (Loxl4) [
53], and fibronectin (Fn1) [
54] (extracellular structure organization); Loxl2 and Loxl3 [
53], and integrin subunit beta 3 (Itgb3) [
55] (mesenchyme development); Sox9 [
56], placenta growth factor (Pigf) [
57], and G protein-couple receptor 4 (Gpr4) [
58] (angiogenesis); vimentin (Vim) [
59], aryl hydrocarbon receptor (Ahr) [
60], chloride intracellular channel 4 (Clic 4) [
61], and ATP/GTP binding protein 1 (Agtpbp1) [
62] (eye development); and putative serine protease 56 (Prss56) [
63], moesin (Msn) [
64], integrin alpha M (Itgam) [
65], amine oxidase copper containing 3 (Aoc3) [
66], and switching B cell complex subunit (Swap70) [
67] (leukocyte migration). Many of these genes have been associated with adult retinal diseases. Our data are the first to show that these genes could be direct or indirect target genes of the inflammasome cascade, which play important roles in OIR. We also identified some of the hyperoxia-suppressed gene pathways that were corrected by IC100 in cluster 6 including synapse assembly, regulation of neuron differentiation, projection, and recognition. Some of specific genes in these pathways included fibroblast growth factor 13 (Fgf13) [
68], cadherin 18 (Cdh18) [
69], and growth associated protein 43 (Gap43) [
70] (synapse assembly); nucleosome assembly protein 1-like 2 (Nap1l2) [
71], fasciculation and elongation protein zeta-1 (Fez1) [
72], BEN domain containing 6 (Bend6) [
73], serpine family I member 1 (Serpini1) [
74], alpha-(1,3)-fucosyltransferase (Fut9) [
75], and hepatocyte growth factor (Hgf) [
76] (regulation of neuron differentiation and projection development). The known functions of some of these genes are related to photoreceptor degeneration, retinal degeneration, retinal development, neural circuit establishment, and self-renewal of neural stem cells. Their roles in the development of OIR or ROP have not been established. The fact that IC100 treatment improved retinal structure and function and corrected expression of these genes further supports the critical role of ASC in the pathogenesis of OIR by transcriptional regulation of hyperoxia-modified genes in mouse models.
We recognize that treatment with IC100 partially corrected the transcriptional changes induced by hyperoxia compared to normal-developed retinas in the RA-PBS group. These results are expected, given that IC100 treatment was provided on P12 when the retinas had been exposed to hyperoxia for 5 days, and the vascular injury and inflammation were induced by hyperoxia at this time. This hyperoxia exposure and treatment sequence closely resemble the clinical practice in ROP diagnosis and treatment.
In conclusion, this study demonstrates that hyperoxia activates ASC in mouse retinas. Treatment with IC100 largely attenuates abnormal vascularization, inflammation, retinal thinning, and retinal dysfunction in a mouse model of OIR. We provide transcriptional effects of hyperoxia and IC100 treatment in addition to the structural and functional changes in this model. The results from this study, combined with our recently published data on the effects of GSDMD deficiency in ameliorating hyperoxia-induced lung and retinal injury in neonatal mice [
26], highlight that the inflammasome-cascade is central to hyperoxia-induced premature multi-organ damage including the retinas. Additional studies are needed to test this antibody in other rodent models of OIR to gain evidence that targeting ASC may be beneficial for treating ROP in premature infants.