Systemic Inflammation Perturbs Developmental Retinal Angiogenesis and Neuroretinal Function

Abstract

Purpose.: Perinatal inflammatory stress in preterm babies is associated with increased rates of severe retinopathy of prematurity (ROP) and adverse neurological dysfunction. In this study, we set out to determine the consequences of severe systemic inflammatory stress on developmental retinal vascularization and evaluate the subsequent outcome on retinal function in later life.

Methods.: Systemic inflammatory stress was induced in C57BL/6J mouse pups by an intraperitoneal injection of lipopolysaccharide (LPS; 1 mg/kg) at postnatal day 4. In response to LPS, retinal inflammation was confirmed by quantitative RT-PCR analysis of diverse inflammatory markers. A detailed and systematic analysis of retinal microglial infiltration, retinal vascular morphology, density, and growth rate was performed at key time points throughout retinal vascularization. Retinal function in adult life was assessed by using electroretinography at 6 weeks postinjection.

Results.: As early as 48 hours after intraperitoneal administration of LPS, a significant increase in retinal vascular density was noted throughout the retina. A pronounced increase in the number of activated microglial cell was observed in the retinal ganglion cell layer and in the outer plexiform layer just prior to their vascularization; direct physical contact between activated microglia and sprouting vessels suggested that microglia partake in promoting the aberrant retinal vascularization. With maturity, animals subjected to perinatal inflammatory stress displayed depleted retinal vascular beds and had significantly decreased retinal function as determined by electroretinography.

Conclusions.: Our data reveal that early severe postnatal inflammatory stress leads to abnormal retinal vascular development and increased vessel anastomosis and, ultimately, permanently compromises retinal function. The aberrant and initially exaggerated retinal vascularization observed is associated with microglial activation, providing a cellular mechanism by which perinatal sepsis predisposes to ROP.

Introduction

Retinopathy of prematurity (ROP) remains a major complication of preterm birth and a leading cause of blindness or profound visual impairment in childhood in the Western world. ^{1 –3} Severe ROP triggers significant sequelae including blinding retinal detachment, while milder forms increase the incidence of vision-impairing ametropia, refractive errors, strabismus, and disorders of color discrimination. ^{4 –7}  

ROP develops following a primary phase of retinal microvasculature degeneration and insufficient vascular development ^8,9 that is associated with interrupted and abnormal progression of vascular growth toward the retinal periphery. A secondary compensatory phase then ensues and results in abnormal pathological pre-retinal neovascularization. ^10,11 To date, the principal recognized factors associated with ROP are low birth weight, low gestational age, and relative hyperoxia associated with supplemental oxygen therapy. ^2,12,13  

Current preventive and therapeutic interventions are largely summed up by tight control of oxygen supplementation and laser therapy. ^{14 –17} While there are promising explorative strategies such as insulin growth factor 1 (IGF-1) therapy, ^{18 –20} tight serum glucose control, omega-3 supplementation paradigms, ^{21 –24} and prospective therapeutic approaches such as anti-vascular endothelial growth factor (VEGF) therapy (BEAT-ROP Trial ²⁵ ), a portion of ROP patients develop long-term visual impairment. Hence, ROP remains a significant complication of premature birth. Importantly, preterm infants with extremely low birth weight are highly prone to infection, and rates of infection inversely correlate with birth weight and gestational age. ²⁶ Neonatal infections in these infants are associated with poor neurodevelopmental and growth outcomes in early childhood. Recently, postnatal infection and systemic inflammation have been proposed as possible contributing factors to ROP pathogenesis. ^{26 –34} Perinatal inflammation may thus be an additional causal factor for ROP. At present, the cellular mechanisms by which systemic inflammation can impact retinal vascular development and influence retinal neuronal function remain ill-defined. It is therefore crucial to gain insight into this process in order to effectively therapeutically modulate the inflammatory stress that perturbs perinatal vascular development. 

Mouse retinal vascular systems develop in the first postnatal weeks and, hence, reliably mimic the final trimester of human retinal development, where preterm birth occurs. ³⁵ There is increasing evidence for the key role of myeloid cells in retinal vascular development, ^36,37 remodeling, ³⁸ repair, ^{39 –43} and anastomosis. ⁴⁴ Myeloid cells such as microglia associate directly with nascent vessels at the vascular front and modulate angiogenesis. ^37,39,45 Retinal microglia are the primary resident immune cells of the retina and are rapidly activated after an inflammatory insult. ^46,47 We therefore sought to determine whether activation of retinal microglia by systemic inflammatory stress could perturb retinal angiogenesis and lead to vascular aberrations similar to those observed in ROP. 

To induce inflammation and mimic a state of perinatal sepsis, we exposed neonatal mouse pups to bacterial lipopolysaccharide (LPS), the endotoxin, which stimulates a profound surge in proinflammatory cytokines. Here we provide evidence that systemic perinatal inflammation provokes accumulation of microglia in the retina. These immune cells associate with nascent retinal vasculature, disorganize the angiogenic front, and promote exaggerated vascular anastomosis resulting in severe vascular abnormalities, as observed in ROP. Importantly, in our paradigm, perinatal inflammatory stress at a period prior to vascular development leads to sustained and long-term visual impairment that persists into adulthood. Our data thus provide novel insight on how neonatal sepsis may predispose preterm infants to ROP and ensuing long-term visual anomalies. 

Methods

Animals

All animal procedures adhered to the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Animal Care Committee of the University of Montreal and Hôpital Maisonneuve Rosemont in accordance with the guidelines established by the Canadian Council on Animal Care. C57BL/6 wild-type mice were purchased from Jackson Laboratory and bred in-house. 

In Vivo Model of Sepsis

In order to induce systemic inflammatory process, mice pups were injected at postnatal day 4 (P4) with 1 mg/kg of LPS ( Escherichia coli O55:B5 lipopolysaccharide, L2880; Sigma-Aldrich, St. Louis, MO) intraperitoneally (IP) and compared to controls injected IP with normal saline. Mouse pups were maintained in normoxia until sacrificed, when eyes were collected. Vascular areas, vascular densities, and microglial numbers were measured at P6, P7, P8, P9, and P10. Adult retinal vascular densities were also assessed at P21. Electroretinogram recordings followed by immunostaining to measure the impact on adult retina were obtained at postnatal age of 6 weeks. 

Immunofluorescence Staining

Mice were sacrificed at 48-hour intervals following LPS injection until P10, P21, and P50 during adulthood. Eyes were collected and fixed in 4% paraformaldehyde for 20 minutes at room temperature. Retinas were dissected and permeabilized with 100% methanol for 10 minutes. Microglial cells and vasculature were identified in retinal flatmounts by incubating tissues with 1/200 dilution of anti-Iba1 (rabbit polyclonal 019-19741; Wako) in 0.4% Triton X-100, 3% bovine serum albumin, overnight at 4°C. Secondary antibody revelation was performed with 1/500 dilution of goat anti-rabbit (Gt anti-Rbt) immunoglobulin G (IgG) 488 Alexa Fluor (GT anti-Rbt 170-5046; Biorad, Mississauga, Ontario, Canada) and 1/100 dilution of Tetramethylrhodamineisothiocyanate (TRITC)-labeled lectin Griffonia simplicifolia (Sigma-Aldrich) for 2 hours at room temperature. Flatmounts were visualized by fluorescence microscopy (Zeiss observer microscope; Carl Zeiss Canada, Ltd., Toronto, Ontario, Canada). Retinal vascularization was quantified as the retinal surface covered at day 6, 7, or 8 relative to total retinal surface, using ImageJ software (http://rsbweb.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD). Vascular density was determined panretinally as the total capillary surface measured using ImageJ. For cryosections, eyes were enucleated from mice and fixed in 4% paraformaldehyde at room temperature for 4 hours, incubated in 30% sucrose overnight, embedded with optimal cutting temperature (OCT) medium, and then frozen in OCT compound. Sagittal cross-sections were permeabilized with 3% bovine serum albumin, 0.4% Triton X-100 over 30 minutes, incubated with 1/200 dilution of rabbit anti-Iba1 (rabbit polyclonal 019-19741; Wako) overnight at 4°C, followed by 1/1000 dilution of fluoresceinated secondary antibody (Gt anti-Rbt 170-5046; Biorad) and 1/100 TRITC-labeled lectin G. simplicifolia (Sigma-Aldrich) over 2 hours at room temperature for localization and vascular stain. 

Microglial Quantification

Numbers of cells expressing Iba1 and/or stained with lectin were determined from P6, P8, P10, and P21 eyes. In each retinal flatmount, 3 representative fields in the central zone and the peripheral zone were selected for counting (6 fields in total per layer [see Fig. 1B]). In each field, cells were counted separately for the deep layer, the intermediate layer, and the superficial layer while observing them in the microscope at high resolution (×200). The layers could be clearly separated with deeper focus point. Activated microglia cells were identified by their altered morphology (short and thick processes) and would have been Iba1-lectin double-stained. For each retina, the mean number of microglia cells counted on each layer was calculated for central and peripheral zones. Then, the means and standard errors of the mean from at least four different retinas were calculated in each zone of all layers. 

Figure 1

 

Postnatal systemic inflammatory stress provokes anomalies in retinal angiogenesis. (A) Timeline of the experimental paradigm used in the study. Each point of sacrifice is represented by a red arrow. (B) Schematic representation of a retinal flatmount depicting typical vascularization of the developing retina. Green (central) and blue (peripheral) squares represent the zones where images were taken throughout development. (C) Representative high magnification images of Griffonia simplicifolia lectin-stained vessels and their quantification (D) in the central retina. (E) Representative high-magnification images and quantification (F) in the peripheral retina. (G) Quantitative RT-PCR for mVEGF mRNA expression in whole-retina extracts confirmed enhancement of vascular proliferation between P6 and P8 (n = 4). (H) Vascular front in LPS-injected retina. (I) Representative photomicrographs of lectin-stained whole-mount retinas at P6, P7, and P8 and their quantification (J) are shown; (n = 4–7 in each group; *P < 0.05; **P < 0.01; ***P < 0.001). (C, E) Bars: 50 μm; (H) 500 μm.

Figure 1 (cont.)

RT-PCR and Quantitative Real-Time PCR

Eyes were rapidly enucleated and whole retinas processed for RNA isolation by using GenElute RNA mini-prep kit (RTN10; Sigma-Aldrich) after tissue homogenization in RNA lysis buffer. We reversed transcribed the RNA by using Moloney murine leukemia virus reverse transcriptase (M-MLV RT) and analyzed gene expression using SybrGreen in an real-time pCR machine (ABI). β-Actin was used as a reference gene. Amplification of various mouse retinal inflammatory target genes used to measure the local response to LPS are listed below. Primer sequences were designed using primer BLAST sequences by NCBI: mIba1 (forward [F]: 5′-CCTGATTGGAGGTGGATGTCAC; reverse [R]: 5′-GGCTCACGACTGTTTCTTTTTTCC), mIL-1β (F: 5′-CTGGTACATCAGCACCTCACA; R: 5′-GAGCTCCTTAACATGCCCTG), mIL-6 (F: 5′-TAGTCCTTCCTACCCCAATTTCC; R: 5′-TTGGTCCTTAGCCACTCCTTC), mTNFα (F: 5′-CCCTCACACTCAGATCATCTTCT; R: 5′-GCTACGACGTGGGCTACAG), mTLR2 (F: 5′-GTTGGACGGCAGTCTCTGCG; R: 5′-GGCCACCAAGATCCAGAAGAGCC), mTLR4 (F: 5′-TGCCTGACACCAGGAAGCTTGA; R: 5′-TGAGAGGTGGTGTAAGCCATGCCA), mVEGF (F: 5′-GCCCTGAGTCAAGAGGACAG; R: 5′-CTCCTAGGCCCCTCAGAAGT). Quantitative analysis of gene expression was performed, and gene expression was calculated relative to that of β-actin expression by using the cycle threshold (ΔC_t ) method. 

Electroretinogram Recordings and Analysis

The retinal function of adult (45-day-old) mice injected with IP injection of saline or LPS at P4 was investigated with electroretinography. Mice were housed in the animal care facility under a 12L:12D cycle (80 lux). Electroretinogram recordings were obtained as described previously. ⁴⁸ Before recordings were made, animals were dark adapted for a period of 12 hours, following which they were anesthetized with an intramuscular injection of a mixture of ketamine (85 mg/kg) and xylazine (5 mg/kg). Drops of tropicamide 1% (Mydriacyl; Alcon, Mississauga, ON) and proparacaine hydrochloride 0.5% (Alcaine) were used to dilate the pupils and anesthetize the eye, respectively. Animals were placed on a heating pad (Harvard Apparatus, Holliston, MA) during the entire recording session to maintain their body temperature at 37°C. All the preparations were performed under dim red light. The animals were then placed in a chamber equipped with a photostimulator (model PS22; Grass Instruments, Warwick, RI) and a rod-desensitizing background of 30 cd/m². The active electrode (DTL fiber electrode; 27/7 X-Static silver-coated conductive nylon yarn; Sauquoit Industries) was placed on the cornea, a reference electrode was positioned in the mouth (E5 disc electrode; Grass Instruments), and a ground electrode (E2 subdermal electrode; Grass Instruments) was inserted in the tail of the animal. Recordings of full-field electroretinograms (bandwidth, 1-1000 Hz; 10,000×, 6-db attenuation; P-511 amplifiers; Grass Instruments) were performed with a data acquisition system (MP 100 WS; Biopac System). Scotopic electroretinograms were evoked in response to a flash of white light of 0.9 log cd·s·m⁻² in intensity (Grass PS-22 photostimulator, interstimulus interval of 10 seconds, flash duration of 20 μs, average of 5 flashes). Photopic electroretinograms were also evoked to flashes of 0.9 log cd·s·m⁻² (photopic background, 30 cd/m²; interstimulus interval, 1 second; flash duration 20 μs; average of 20 flashes). In order to avoid the previously reported light adaptation effect, photopic recordings were obtained 20 minutes after the opening of the background light. ^49,50 Electroretinogram amplitudes were measured according to a method described previously. ⁵¹ Briefly, the amplitude of the a-wave was measured from baseline to trough and the b-wave from the trough of the a-wave to the highest peak of the b-wave. All recordings were made from the left eye. 

Statistical Analysis

Results are expressed as means ± SEM. Electroretinogram data are given as means ± SD. Statistical analyses were performed using two-tailed unpaired Student t tests and one-way analysis of variance (ANOVA). Statistical significance was set at P value of <0.05. 

Results

Induction of Postnatal Inflammatory Stress Alters Retinal Vascular Development

There is mounting evidence suggesting that neonatal sepsis is a major risk factor in developing severe ROP. ^{26 –32} In order to determine how a surge in systemic inflammation during the early postnatal period influences retinal vascular development, we exposed mouse pups to systemic LPS and assessed retinal vascular density and rates of retinal vascularization. 

LPS (1 mg/kg) was administered to postnatal mouse pups by a single IP injection on P4, and retinal vascular phenotypes were analyzed on P6, P8, P10, and P21 (Fig. 1A). Vascular density was quantified on lectin-stained whole-retinal flatmounts that were divided into central or peripheral areas and imaged accordingly (Fig. 1B). In central zones (Fig. 1B, green square), a significant increase in vascular density was noted at P6 (29.4% ± 5.9% increase, P = 0.0002) and P8 (22.9% ± 7.2% increase, P = 0.005), whereas vessel density returned closer to normal levels and no significant change was observed at P10 (Figs. 1C, 1D). In peripheral zones (Fig. 1B, blue square), the augmented vascular density persisted to P6 (48.1% ± 4.6% increase, P < 0.0001), P8 (22.9% ± 5.9% increase, P = 0.0009), and P10 (48.7% ± 12.3% increase, P = 0.004) (Figs. 1E, 1F). Consistent with a heightened state of angiogenesis, analysis of whole-retinal mRNA by quantitative PCR from P6 to P8 revealed an increase in VEGF expression in LPS-treated mice (Fig. 1G). Vessel density at the vascular front was profoundly increased, and vessels showed high rates of anastomosis. The peripheral superficial retinal vasculature was reminiscent of fibrotic neovascular scars observed in ROP (Fig. 1H). ^12,13  

The aberrant increase in vascular density had minimal effects on the overall rate of vascularization, other than at 72 hours postinjection (at P7), when the total vascularized area was transiently decreased by 11.2% ± 2.5% (P = 0.0003) compared to controls (Figs. 1I, 1J). Together, these data demonstrate that systemic postnatal inflammatory stress perturbs the initial retinal vascularization primarily by increasing vascular density. 

Vascular density was also assessed at P21, a developmental time point after all retinal plexuses have formed. ⁵² Mice that received LPS had tenuous vessels and significantly decreased densities in all areas analyzed (central decrease of 22.4% ± 7.5%, P = 0.01; peripheral decrease of 32.8% ± 8.6%, P = 0.003) (Figs. 1C–1F). These data suggest that the initial rise in vascular density receded and was followed by excessive vascular pruning in later stages of retinal vascular maturation. 

Systemic Postnatal Inflammation Induces Massive and Sustained Pan-Retinal Microglial Activation

Retinal microglia respond to stress and injury and upon activation retract their processes and become increasingly lectin-positive. ⁵³ Subjecting neonatal mice to a single systemic injection of LPS at P4 (Fig. 1A) robustly activated microglia throughout the retina (Figs. 2, 3) by directly or indirectly inducing a state of local retinal inflammation. ^54,55 Activated microglia resided in three distinct layers loosely corresponding to the retinal vascular plexuses when observed by immunofluorescence in sagittal cross-sections (Figs. 2A, 2B). At the points analyzed between P6 and P8, the increase in number of microglia was accompanied by a rise in mRNA expression of the microglial marker Iba1 and LPS-specific receptors TLR2 and TLR4 (Fig. 2C). In addition, the response to systemic LPS resulted in a 2- to 5-fold increase in retinal expression of mRNAs for proinflammatory cytokines (IL-1β, IL-6, TNF-α) as determined between P6 and P8 (Fig. 2D). These data correlate with the incremental accumulation of activated microglia (Fig. 3). 

Figure 2

 

Systemic administration of LPS provokes an increase in retinal microglia and levels of pro-inflammatory cytokines. (A, B) Immunofluorescence on retinal cryosections shows the rise in microglial cells in the ganglion cell layer in the inner and outer plexiform layers at P8 in both central and peripheral zones of the retina. (C) Quantitative RT-PCR results are shown for mIba-1, mTLR4, and mTLR2 on whole retinas from P6, P7, and P8 (n = 4). (D) Focal retinal inflammatory state was confirmed by quantitative RT-PCR analysis showing increased mRNA expression of mIL-1β, mIL-6 , and mTNFα on whole retinas from P6 to P8 (n = 4). (**P < 0.01; ***P < 0.001). (A, B) Bars: 500 μm.

Figure 3

 

Systemic administration of LPS induces a major increase in activated microglia throughout the retina. (A) Graphical depiction of the superficial vessel plexus being analyzed. (B–E) Representative images and quantification at P6 and P8 of double-stained (Iba1 in green and lectin in red) whole-retinal flatmounts in central and peripheral zones of the superficial retinal layer after LPS injection compared to controls. (F) Graphical depiction of the nascent intermediate plexus is shown. (G–J) Representative images and quantification at P6 and P8 of double-stained (Iba1 in green and lectin in red) whole-retinal flatmounts in central and peripheral zones of the nascent intermediate plexus after LPS injection compared to controls. (K) Graphical depiction of the deep vascular plexus. (L–O) Representative images and quantification at P6 and P8 of double-stained (Iba1 in green and lectin in red) whole-retinal flatmounts in central and peripheral zones of the deep vascular plexus after LPS injection compared to controls. (n = 4–8 in each group; *P < 0.05; **P < 0.01; ***P < 0.001). (A, B) Bars: 500 μm; (F, H, K, M, P, R) 50 μm.

Figure 3 (cont.)

Figure 3 (cont.)

In order to quantify the distribution of activated microglia, we divided the retina into central or peripheral zones and into layers corresponding to the superficial plexus (ganglion cell layer [GCL] [Fig. 3A]), the nascent intermediate layer (beneath the GCL and loosely corresponding to the forming inner plexiform layer [Fig. 3F]) and the deeper layer (corresponding to the outer plexiform layer [Fig. 3K]). Superficial layers were separated to better define how microglial distribution evolves after systemic inflammatory insult. Activated microglial cells were identified as coexpressing Iba-1 (green) and lectin (red) and having an amoeboid morphology as previously described. ⁵³  

The superficial (Figs. 3A–E) and deep (Figs. 3K–O) layers, which are the first 2 vascular plexuses to form, showed a significant increase in the number of activated microglia at P6 (48 hours post-LPS) and P8 (96 hours post-LPS) in both peripheral and central zones (Figs. 3A–E, 3K–O). The intermediate layer (last vascular layer to form) did not initially (at P6) contain more microglia; this might be due to massive activation (and migration) of microglia into the superficial layers at P6; however, by P8, there was significant accumulation of activated microglial cells in the intermediate layer in LPS-treated animals (Figs. 3F–J). At P10 (144 hours post-LPS), the number of activated microglia subsided, and only superficial peripheral zones still contained significantly more activated microglia (Figs. 3D, E). Our data reveal an acute yet relatively transient rise in activated microglia in retinal zones corresponding vascular plexuses. 

Activated Microglia in Postnatal Sepsis Are Associated With Retinal Vessels

Retinal microglia play a crucial role in modeling and organizing the nascent vascular plexus. ^36–41,44 Throughout normal development, microglia are associated with vascular tip cells and partake in the angiogenic process (Fig. 4A). During neonatal sepsis, a large number of microglial cells (Iba1-positive green cells) localize to central retinal zones in the wake of the vascular front at branch points and likely partake in vascular anastomosis ⁴⁴ (Figs. 4B, 4F). An abnormally elevated number of microglia is also observed at the vascular front and potentially disorganize vascular growth (Fig. 4B). These activated microglia are intimately associated with retinal vessels (Fig. 4D) and are found either around branch points of sprouting vessels or at tip cells, predominantly in peripheral zones at P6 and P8. These microglia remain after retinal vascularization is completed as observed at P21 and persist mainly in the central portion of the retina at P50, the latest time point analyzed (Figs. 4F–4J). No overt morphological changes were noted between groups at P50. 

Figure 4

 

Microglia activated by systemic inflammation are associated with sprouting retinal vessels. (A, B) High magnification of P8 retinal flatmounts reveal an increased number of activated Iba1-positive microglial cells (green) in direct association with the vascular front or tip and stalk cells after LPS treatment within central and peripheral zones of the retina. Both superficial retinal regions, central and peripheral, revealed an increase number of microglial cells in close proximity with growing vessels. (C, D). Immunofluorescence on retinal cryosections confirmed increased numbers of microglial cells (green) in association with vessels of the GCL. (E, F) High magnification images of double-labeled (Iba1 and lectin) whole-mount retinas at P6, P8, and P21 shows elevated numbers of microglia in close proximity with branching point of sprouting vessels (white arrows) in LPS-treated. (G, H) Comparison of P8 and P50 retinal immunofluorescence of microglial cells (green) in central and peripheral zones of the retina. Central and peripheral microglial cells in the ganglion cell layer are increased in both time points and reveal a long-term impact of LPS on microglia distribution at P50. (I, J) Three-dimensional reconstruction of Z-stack images from central retinal zones in LPS-treated animals at P8 and P50 confirmed a sustained rise of microglia (green) in the GCL. (G, H) Bars: 50 μm; (A,C,D,E,F) 25 μm; (B) 10 μm.

Figure 4(cont.)

Systemic Postnatal Inflammatory Stress Alters Neuroretinal Function in Adult Mice

Given the severity of retinal vascular aberrations resulting from the LPS-induced model of perinatal sepsis (Fig. 1), we sought to determine whether visual function was affected in adulthood. We assessed neuroretinal function by performing short-flash electroretinography analysis 6 weeks post-LPS treatment. Scotopic recordings (mixed cone-rod) from animals having undergone LPS-induced perinatal sepsis-like injury showed a pronounced 28.0% decrease in a-wave amplitude, indicative of outer-retinal dysfunction (P = 0.01) (Figs. 5A, 5B). A similar 28.5% attenuation in b-wave amplitude was observed, suggesting inner retinal dysfunction (P = 0.02) (Figs. 5A, 5B); no significant change in peak times were noted for either scotopic a- or b-waves (Fig. 5C). In line with scotopic measurements, LPS-treated animals showed a 22.1% decrease in b-wave amplitude (P = 0.04) under photopic conditions; peak time amplitudes did not change significantly (Figs. 5D, 5E). These results demonstrate that lesions incurred secondary to postnatal sepsis are sustained into adulthood and result in functional neuroretinal deficits. 

Figure 5

 

Adult neuroretinal function is altered by postnatal inflammatory stress. (A) Representative recordings of full-field scotopic ERGs in response to bright flashes of white light at an intensity of 0.9 log cd · s · m⁻². (B) P4 LPS-injected mice (6 weeks old at recording time) show a significant decrease in scotopic a-wave and b-wave responses, respectively, compared to controls. (C) No significant increase in peak time was observed between groups. (D) Representative recordings of full-field photopic ERGs in response to bright flashes of white light at an intensity of 0.9 log cd · s · m⁻². In LPS-treated mice, photopic ERG b-waves were significantly decreased compared to those in controls, whereas b-wave peak times remained significantly unchanged (E) (n = 5–6 in each group; scotopic and photopic; *P < 0.05).

Discussion

While neonatal sepsis is being recognized as a major risk factor for developing severe ROP, ^27,28,33 to date, the underlying mechanism of how pronounced perinatal systemic inflammation influences retinal angiogenesis is not well understood. In this study, we provide evidence that severe systemic perinatal inflammation provokes activation of microglia within the retina and leads to aberrant vascular development with excessive anastomosis. These vascular abnormalities are similar to those observed in ROP ^2,13,56 and are associated with long-term deficits in visual function. 

We observed that a single systemic dose of LPS induced a sepsis-like state and provoked a profound increase in the number of activated microglia in intimate association with the nascent vascular plexuses of the retina. LPS is an endotoxin ligand for Toll-like receptor (TLR)-4 family members ⁵⁷ and is frequently used to study the inflammatory response. Exposure of endothelium to LPS results in its activation and production of proinflammatory mediators that ultimately increase vasopermeability. ^58,59 Moreover, microglia are also equipped with a broad range of pattern recognition receptors of the TLR family to detect microbial intruders. ⁶⁰ Microglia carry out functions of immune surveillance ^61,62 and neuronal homeostasis ^63,64 and can have either protective or detrimental effects depending on the immunological state of the tissue and the local cytokine signature. ^62–65 In our experimental paradigm, systemic LPS provoked microglial activation across retinal layers and significantly increased total numbers of activated microglia in the retina. The observed rise in production of proinflammatory cytokines correlated with the observed peak in microglial numbers and their activation state. This protracted response to a single injection of LPS is likely related to long-term microglial activation rather than an immediate direct response by circulating macrophages and endothelial cells to systemic LPS. These data agree with those of previous reports describing LPS-provoked microglial activation, ⁶⁶ subsequent migration, ⁶⁷ and a sustained inflammatory response. ^62,65,68  

Retinal microglia are resident ocular immune cells derived from myeloid progenitor cells. ⁶⁹ They enter the retina either from the peripheral margins via blood vessels of the ciliary body or centrally from the embryonic hyaloid artery via optic nerve head and vitreous. ^46,47 The postnatal flux of retinal microglia occurs according to a bell-shaped distribution across the period of vascular developmental (P0–P21) with the highest number of microglia reported at P7. ⁴⁶ Our data agree with this reported developmental peak of retinal microglia; however, the systemic inflammation paradigm used in our study massively increased the total number of these cells. Importantly, the most significant changes in microglial numbers were observed in the superficial (GCL) and deep (OPL) layers immediately preceding vascularization. Systemic neonatal inflammation primarily impaired vascular arborization and provoked a significant increase in peripheral vascular density that in the most severe cases resulted in a neovascular membrane-like structure. The rate of retinal vascularization was negligibly affected. Throughout our experimental paradigm, an intimate proximity was observed between microglia and several vascular structures including tip, stalk, and phalanx cells of nascent vessels. This microglial-vascular relationship likely promoted the fusion of developing vascular structures as has been previously demonstrated. ⁴⁴ Taken together, the timing of the increased influx of activated microglia into retinal layers undergoing vascularization and their direct interaction with nascent vessels is consistent with their role in perturbing retinal vascularization following systemic inflammation. ^26,33,70,71  

Of direct clinical relevance, analysis of retinal function by electroretinography suggests that heightened systemic perinatal inflammation during retinal development leads to long-term visual sequelae and altered visual function. The electroretinographic changes noted appear to have affected the rod and cone function as well as the outer (a-wave) and inner (b-wave) retina similarly (identical attenuation factors), a finding suggestive of a panretinal effect. Furthermore, although there were no earlier measurements taken of retinal function (i.e., closer to the time of injection), the fact that only the amplitude of the electroretinogram was affected (and not the peak time or morphology of the electroretinogram wave) would suggest a destructive acute effect (that occurred soon after sepsis), rather than a degenerative chronic condition (that would have been triggered following the injection). Notwithstanding the above, our findings agree with the notion that in addition to hyperoxia postnatal inflammatory stress may be an another factor that predisposes preterm infants to ROP and precipitates longer term loss of visual acuity. ^34,35  

In summary, this is the first study to demonstrate that systemic inflammatory stress occurring during retinal development may lead to permanent retinal damage, likely through activated microglia. Numerous actions of microglia are implicated in normal and aberrant development of the neurovascular unit. ^68,72 While the underlying mechanisms of microglial activation and recruitment remain to be elucidated, our study suggests that perinatal inflammatory stress should be monitored and hence provide insight for future therapeutic interventions to counter ROP. 

Acknowledgments

Supported by operating grants from the Canadian Institutes of Health Research (CIHR) (221478) (PS), the Canadian Diabetes Association (OG-3-11-3329) (PS), the Natural Sciences and Engineering Research Council of Canada (418637), and the Foundation Fighting Blindness Canada. Additional funding was provided by the Réseau de Recherche en Santé de la Vision du Québec-FRQS. ST holds scholarships from the Fonds de la Recherche en Santé du Québec (FRSQ) and CIHR. PS holds a Canada Research Chair in Retinal Cell Biology and the Alcon Research Institute Young Investigator Award. SC holds a Canada Research Chair (Translational Research in Vision) and the Leopoldine Wolfe Chair in translational research in age-related macular degeneration. FB holds a FRSQ postdoctoral fellowship. 

Disclosure: S. Tremblay, None; K. Miloudi, None; S. Chaychi, None; S. Favret, None; F. Binet, None; A. Polosa, None; P. Lachapelle, None; S. Chemtob, None; P. Sapieha, None