Antenatal IL-1-dependent inflammation persists postnatally and causes retinal and sub-retinal vasculopathy in progeny

Antenatal inflammation as seen with chorioamnionitis is harmful to foetal/neonatal organ development including to eyes. Although the major pro-inflammatory cytokine IL-1β participates in retinopathy induced by hyperoxia (a predisposing factor to retinopathy of prematurity), the specific role of antenatal IL-1β associated with preterm birth (PTB) in retinal vasculopathy (independent of hyperoxia) is unknown. Using a murine model of PTB induced with IL-1β injection in utero, we studied consequent retinal and choroidal vascular development; in this process we evaluated the efficacy of IL-1R antagonists. Eyes of foetuses exposed only to IL-1β displayed high levels of pro-inflammatory genes, and a persistent postnatal infiltration of inflammatory cells. This prolonged inflammatory response was associated with: (1) a marked delay in retinal vessel growth; (2) long-lasting thinning of the choroid; and (3) long-term morphological and functional alterations of the retina. Antenatal administration of IL-1R antagonists – 101.10 (a modulator of IL-1R) more so than Kineret (competitive IL-1R antagonist) – prevented all deleterious effects of inflammation. This study unveils a key role for IL-1β, a major mediator of chorioamnionitis, in causing sustained ocular inflammation and perinatal vascular eye injury, and highlights the efficacy of antenatal 101.10 to suppress deleterious inflammation.

homogenized and RNA was isolated as per manufacturer's instructions. Using the spectrophotometer Nanodrop 1000, RNA concentration and purity (ratio A260:A280 > 1.6) were measured. For each sample, 500 ng of RNA was converted into complementary DNA (cDNA) using the synthesis kit iScript Reverse Transcription Supermix (Bio-Rad; Hercules CA, United States). Primers were designed with NCBI Primer Blast. Gene expression was quantified with Stratagene MXPro3000 (Stratagene) with SYBR Green Master Mix (Bio-Rad). Gene expression levels were standardized with the universal primer 18 S (Ambion Life Technology; Burlington ON, Canada). Dissociation curves were obtained to verify primer specificity. We analysed the following genes: Il1b, Il6, Il8, Il12, Ccl2, Casp1, Tnfa, Il4, Il10 and Il27, as previously described 22,28 ; primer sequences are detailed in Table 1.
Immuno-enzymology method ELISA. ELISA was performed as documented 22,28 and as recommended by the manufacturer for the following kits: mouse IL-1β/IL-1F2 Quantikine (#MLB00C; R&D systems), mouse IL-6 Quantikine (#M6000B; R&D systems) and mouse IL-8 (#MBS261967; Mybiosource; recognizes the IL-8 homologue CXCL2). Tissue was equally distributed and briefly lysed in a RIPA solution containing proteases to inhibit protein degradation. Each 50 μL sample was then loaded on a 96-well plate previously embedded with specific primary antibodies, then incubated for 2 hours at room temperature. Then, wells were washed 5 times and incubated for 2 hours with an enzyme-linked polyclonal secondary antibody. After further washes, substrate solution was added. Reaction was stopped after 30 min and plates were read at 450 nm with a wavelength correction of 570 nm.
Eye collection and fixation. Foetuses and pups (<Pt4) were decapitated or sacrificed with CO 2 and cervical dislocation (>Pt7). By pressing around the orbits with small scissors, eyes were pulled out of the skull and collected by pinching the optic nerve. The cornea was pierced with a syringe and the eyes were fixated for 20-30 min in PFA 4%, then transferred in PBS and preserved at 4 °C until they were processed for flat mounts or cryosections.
Retinal and choroidal flat mounts. For each group, eyes of 5 mice at Pt1, Pt4, Pt8, Pt15, Pt21 and Pt30 were dissected as reported 43 , to isolate the retina and choroid/retinal pigment epithelium (RPE) complex. Flat mounts were then labelled with a fluorescent antibody against lectin (Sigma-aldrich; 1:100) and Iba-1 (ab5076, Abcam; 1:500). Fluorescent secondary antibody was applied for the Iba-1 labelling (Alexa fluor 488, Abcam; 1:200). On a microscope slide, 4 cuts were performed with a scalpel to flatten the retina and choroid/RPE complex. With a paintbrush soaked with PBS, hyaloid vessels were removed. Images were obtained in confocal microscopy using a 20 × (lectin) or 100 × (Iba-1) objective. For Iba-1 imaging, the confocal was focused on the superficial retinal vascular layer at Pt 1 (corresponding to the nerve fibre layer [ Histological quantification. Vascular surface was measured with ImageJ and expressed in percentage of total retinal surface, and vessel density was analysed using ImageJ software in the mid-periphery of the retina, as described 43 . Briefly, lectin immunolabelling (blood vessels, showed as red on the retina) was isolated from other structures using the colour deconvolution tool in ImageJ. The detection threshold was established to reduce artefacts and a semi-quantitative comparison of vascular density was performed. Iba-1 positive (Iba1 + ) cells were counted using the ImageJ software in the central retina at Pt1 and in the mid-periphery at Pt15 and 30, as reported 45 .
In order to measure the thickness of the different retinal layers and of the choroid, ImageJ software was used to draw a line across the region to measure. Measures were taken at the central, middle and periphery of the retina and were repeated to obtain results that were more representative. A ratio was calculated for each eye to make a comparison between the groups. The retinal thickness was measured between the outer nuclear layer (ONL) and the ganglion cell layer (GCL) using the same method than the choroid, as we reported 21,39,46 . Electroretinography. Electroretinogram (ERG) recording was performed with Espion ERG diagnosys machine with ColorDome Ganzfeld stimulator (Diagnosys LLC, Lowell, MA), as reported 39 . Mice were dark-adapted overnight and anesthetized prior to ERG with an intraperitoneal injection of Ketamine (100 mg/kg) and Xylazine (20 m/kg) mixture and mouse body temperature was kept at 37 °C using heated pad. DTL Plus electrodes (Diagnosys LLC) were positioned at the surface of the cornea after pupil's dilatation with drops of 1% atropine and 2.5% phenylephrine (Alcon) and flash ERG were measured. Scotopic responses were simultaneously recorded at light intensity 0.9 cd-s/m². ERG was performed under red light in a dark room. After recording, eye hydration was carefully verified, and mice were kept at 37 °C until they awoke. Statistical analysis. The term dam/group refers to the dams per group regardless of number of progeny (pups) of at least 1; hence n = 1 more for a single dam (with ≥1 progeny). Parametric analysis was performed since the power analysis is generally greater for continuous parametric variables, and given the spread of data between groups 47 . Comparisons between two variants was analysed by t-test. Comparisons between several groups were performed using one-way variance analysis (ANOVA); Dunnett's multiple comparison method was

Gene
Forward Reverse used when many treatments were compared with one control. The value p < 0.05 was considered statistically significant. Data is presented as means ± S.E.M 47 .
Data availability. Data sets generated during this study can be obtained from the corresponding author upon request.

Results
In utero exposure to IL-1β causes an acute perinatal inflammatory response in the foetal retina and sub-retina. To study the retinal and sub-retinal vascular development in foetuses and pups after exposure to IL-1β in utero, we employed an established murine model of IL-1β-induced chorioamnionitis 34,42 . As expected, intrauterine IL-1β (G16.5 days) shortened gestation and induced marked neonatal mortality; 101.10 prolonged gestation to term and markedly augmented foetal survival (Suppl. Fig. 1); whereas Kineret exerted no protective effect, as previously reported 34 . Eyes (undissected) of foetuses exhibited increased mRNA expression of pro-inflammatory mediators with a variable profile depending on timing after exposure to IL-1β-induced chorioamnionitis ( Fig. 2A-C); at birth, increased protein levels of IL-1β, IL-6 and IL-8 were detected in the eyes (Fig. 2D-F); these were normalized by 101.10, but not by Kineret with the exception of IL-8 attenuation. Interestingly, early (at G17.5) IL-1β-induced expression of anti-inflammatory Il27 was suppressed by Kineret but preserved by 101.10; while expression of Il10 was diminished by both Kineret and 101.10 (Suppl. Fig. 2). To better localize the intraocular inflammatory response, we isolated the retina and choroid of foetuses. IL-1β triggered early (G17.5) mRNA expression of Il1b, Il6, Il8, Il12, Ccl2 and Tnfa in retina more than in choroid (Fig. 2G,H), which was associated with increased intra-ocular accumulation of Iba1 + cells (activated macrophage/microglia) 48 (Fig. 2I), as reported in inflammation-associated retinopathy 49 ; this Iba1 + cell accumulation is clearly distinct from the relatively low level of Iba1 + cells in control animals wherein it plays a role in normal retinal development 50 . Notably, antenatal 101.10 and Kineret significantly attenuated marked IL-1β-elicited cytokine induction and Iba1 + cell accumulation ( Fig. 2G-I).
Iba1 + cell accumulation persists post-natally in retina and choroid of newborns subjected in utero to inflammation. Abnormally high numbers of Iba1 + cells in retina over extended duration can affect vascular development and in turn cause long-term deficits of retinal function 37 . Since perinatal inflammation resulting in retinopathy is associated with prolonged inflammation 37 with IL-1β-dependent microglial accumulation 38 , we determined if Iba1 + cells persisted in retina and choroid of animals exposed to IL-1β-induced chorioamnionitis. We detected a 3-fold increase in Iba1 + cell density on post-term (Pt) day 1 in eyes of animals exposed to antenatal IL-1β, which persisted on postnatal days 15 and 30 in retina and choroid ( Fig. 3  week was reduced in animals subjected to IL-1β-induced chorioamnionitis (consistent with previous reports 53 ) (Fig. 4A,B), but by Pt 8 retinal surface was fully vascularized in all groups as vessels reached the periphery (Fig. 4C); concordantly, Il1b expression was only marginally increased at Pt 8 (p < 0.06 vs sham-treated; Suppl. Fig. 4A). On the other hand, a small decrease in pan-retinal vascular density was detected in mice exposed to chorioamnionitis at Pt 8, which corresponds to the period wherein intra-retinal vessels are actively forming. This hypo-vascularization was aggravated at Pt 15 (Fig. 4D) when Il1b (mRNA) expression markedly rose (Suppl. Fig. 4A); ultimately retinal vascularization normalized at Pt 30 (Suppl. Fig. 4B) along with Il1b (mRNA) expression (Suppl. Fig. 4A). Antenatal 101.10 fully rescued retinal vascularization at all ages analysed, whereas Kineret was only partially effective at Pt 1 and ineffective subsequently at Pt 4, Pt 8 and Pt 15 (Fig. 4).
Since the choroid is an exclusive blood supply to the outer retina and its involution was recently linked to ROP and retinal functional deficits 39,46 , we measured choroid thickness. As anticipated, choroid thickness was normal soon after IL-1β treatment (at G17.5) (Suppl. Fig. 4C). By Pt 1 choroidal thinning was observed; these changes persisted beyond the neonatal period (at Pt 21) (Fig. 5A,B). Antenatal 101.10, but not Kineret, preserved choroidal (normal) thickness.
To demonstrate that the effects of IL-1β are unrelated to prematurity per se, we showed that PTB induced with the progesterone receptor antagonist RU486 did not significantly affect ocular cytokine profile or postnatal vascular development (Suppl. Fig. 5), inferring an important role for perinatal inflammation on the latter.
Retinal structural and functional deficits in offspring secondary to gestational tissue-triggered inflammation. Retinal development depends upon normal blood supply 54 , such that both retinal structure and function are affected by vasculopathy as seen in ROP 21,39 . Accordingly, we measured retinal thickness from the outer neuroblastic layer to the nerve fibre layer at G17.5 and Pt 1, and between the ONL and the GCL at Pt 21 (of the same eyes used to assess choroidal thickness). As expected 39 , full retinal thickness was not affected within the first 2 weeks after antenatal-induced inflammation (Fig. 5C). By Pt 15, following retinal vasculopathy (Fig. 4), there was a tendency for decreased thickness of the retina (Suppl. Fig. 6A), which was significantly manifested at Pt 21 (Fig. 5D), and was largely attributed to thinning of the inner nuclear layer (Fig. 5F and Suppl. Fig. 6B); this resulted in a corresponding decreased b-wave amplitude (largely contributed by the inner nuclear layer) (Fig. 6). The ONL thickness and corresponding a-wave amplitude remained intact at the end of the first postnatal month (Figs 5E and 6), as reported in ROP-associated choroidopathy 39 . Retinal morphometry and inner retinal function were fully preserved by antenatal 101.10, but only partially by Kineret.

Discussion
Gestational inflammation is an independent risk factor for the development of neonatal morbidities, including to the eye 11,16,19 . Yet, the identification of a causal factor playing a pivotal role remains a challenge. Studies have suggested that the pro-inflammatory cytokine IL-1β is involved in the pathophysiology of many perinatal complications 55,56 . A link between chorioamnionitis and neonatal brain injury is described 19 , and associated with vascular damage 42 . Of relevance, IL-1β plays a critical role in central nervous system insult during the perinatal period 42,57,58 . In retinal injury the role of IL-1β has mostly been ascribed to inflammation resulting from exposure to hyperoxia [37][38][39][40] , an important predisposing factor to ROP. However, there is still no evidence demonstrating a direct role for IL-1β in retinal/sub-retinal injury of the immature subject. Given that retinal damage represents a major complication of prematurity, itself often associated with inflammation, we set out to discriminate the effects of antenatal (intra-uterine) IL-1β-triggered inflammation on retinal and sub-retinal vascular development in offspring during the perinatal period and adolescence; in this process we also determined the efficacy of 101.10 (compared to Kineret). By inducing utero-placental inflammation as reported 34,42 with intra-uterine IL-1β (which does not cross the placental barrier 59 ), we found a prolonged inflammation in the eye that begins ante-natally and ensues in choroidal and retinal vascular, structural and ultimately sustained functional deficits; these detrimental changes were prevented by antenatal 101.10 (but not [recommended dose of] Kineret). The cascade of inflammatory mediators amplified by IL-1β exerts cytotoxicity on its own. As alluded to above macrophage/microglial invasion is a major contributor of IL-1β generation which in turn exerts endothelial cytotoxicity via neuronally-generated Semaphorin3A 38 , and causes direct neurotoxicity including to photoreceptors 60,61 . In cultured retinal endothelial cells, chronic exposure (5 days) to IL-1β leads to increased caspase-3 activity and apoptosis 36 . Concordantly, inhibition of IL-1β signalling using pharmacological or genetic approaches decreases caspase activity and apoptosis in retina, and prevents microvascular degeneration in hyperglycaemia-induced retinopathy models 62 . Comparably, IL-12 is a cytokine produced by macrophages, neutrophils, and other inflammatory cells, and acts by triggering T-cell differentiation 63 ; IL-12 exhibits antiangiogenic properties 64 , and has been associated with serious ocular diseases including uveitis 65 . Interestingly, IL-1-induced  Representative images (top panels) and quantification (bottom panels) of lectin-stained cross-sections of choroids from pups at Pt 1 (A) and Pt 21 (B) previously exposed to the indicated treatments in utero (Fig. 1). Vertical bars represent the average choroidal thickness. Scale bar for A, 30 μm; scale bar for B, 50 μm. n = 3-4 pups/group. (C,D) Representative images (top panels) and quantification (bottom panels) of DAPI-stained cross-sections of retinas from pups at Pt1 (C) and Pt21 (D) previously exposed to the indicated treatments in utero (Fig. 1) IL-12 retinal production was inhibited by 101.10 which may represent an additional mechanism through which 101.10 protects retinal vasculature. The role of TNF in cytotoxicity is more complex as it depends on the receptor it acts upon which may convey opposing actions 66,67 .
In addition to retinopathy, choroidopathy has lately also been regularly observed in subjects formerly afflicted with ROP as recently reviewed 68,69 ; and this feature was recently reproduced in models of ROP 39 . Involution of the choroid was found to be sustained and subsequently led to retinal pigment epithelium and photoreceptor degeneration 39 . In this context, IL-1β exerts a major contribution to choroidal thinning of the newborn subjected to perinatal oxidative stress, which in turn results in long-term injury to the sub-and outer retina. In line with this concept, we observed sustained presence of inflammatory cells post-natally, resulting in choroidal thinning as early as Pt 1, which we had found to predispose to subsequent photoreceptor injury. Together, current and previous observations 39 related respectively to antenatal and postnatal inflammation and ensued retinal damage, reveal efficacy of anti-IL-1β treatment. Our findings are consistent with brain neuronal cell death of pups exposed to IL-1β-induced chorioamnionitis 42 . Hence, chorioamnionitis triggers an inflammatory cascade that spreads to the foetus causing retinal and choroidal injury.
An important feature reported herein is the long-term deficit in retinal structure (inner nuclear layer thinning) and function caused by exposure to IL-1β in utero, despite recovery of retinal vascular density at Pt 30 which coincided with abated IL-1β expression, consistent with diminished expression of inflammatory cytokines in residing macrophages upon resolution of the pro-inflammatory state 51,52 . Inner retinal functional deficits (depicted by b-wave amplitude suppression) are well described in former ROP subjects and believed to result prominently from retino-vascular injury 21,68 . Although we also observed early involution of the choroid upon exposure to utero-placental inflammation, time is needed until developmentally increased photoreceptor metabolism can no longer be supported by insufficient O 2 and nutrient supply from the damaged choroid, resulting in photoreceptor damage 39 . Hence, we surmise a gradual degeneration of photoreceptors (and retinal pigment epithelium) after exposure to antenatal inflammation. Together, the retinal and sub-retinal damage observed in response to antenatal IL-1β are indicative of an important role for inflammation, found to be prolonged.
A relevant aspect of this study applies to the efficacy of 101.10 compared to Kineret in protecting the choroid and retina. We recently showed that inhibition of IL-1 receptor using a novel small peptide labelled 101.10 was fully effective in preventing PTB, improving survival and preserving foetal lung, gut and brain integrity 42 ; whereas current commercially available IL-1 receptor antagonist (namely Kineret) were mostly ineffective. These findings represented an unparalleled therapeutic which improved foetal/newborn outcome. Kineret is a competitive antagonist of the IL-1 receptor. Doses utilized corresponded to those clinically recommended (4 mg/kg/dose). Although higher doses of Kineret are more effective, a 7-fold-increase in dosage is needed to augment gestation and ensuing foetal maturation 42 , which would seriously compromise immune response, an established side effect of Kineret. As a competitive antagonist of IL-1 receptor Kineret suppresses both JNK/p38/c-jun/AP-1 and NF-κB pathways; whereas 101.10 acts as a modulator that biases IL-1R-induced signal transduction resulting in inhibition of the JNK/p38/c-jun/AP-1 pathway, while desirably preserving the NF-κB pathway 23,34 , important for immune vigilance 70,71 . Along these lines, in contrast to Kineret, 101.10 maintained activity of some endogenous anti-inflammatory cytokines, notably IL-27 known to neutralize macrophage activity 72 , thus augmenting its anti-inflammatory activity; interestingly, IL-10 release is p38-dependent 73 and IL-27 generation is NF-kB-dependent 74 , consistent with signalling actions of 101.10 and Kineret. Another disadvantage of Kineret applies to its relatively large size (17.5 kDa) curtailing passage across the placental barrier to sufficiently block inflammation in the foetal compartment 59 ; whereas 101.10 is a small molecule (0.85 kDa) that can distribute in gestational tissues 42 , consistent with its preferred antenatal efficacy in suppressing detrimental inflammation and preserving foetal organ integrity. In summary, this study unveils an unprecedented key role for IL-1β during gestation in the development of long-lasting inflammation and injury to the eye of the progeny, independently of prematurity per se. Antenatal inflammation as seen in chorioamnionitis induces damage to the choroid and retinal vasculature, structure and function, all of which is remarkably preserved by the small IL-1 receptor modulator 101.10. 101.10 could represent a novel therapeutic approach not only to tackle PTB 34 but also in preventing major foetal/neonatal organ injury including to the eye of premature subjects exposed to antenatal inflammation.