December 2016
Volume 57, Issue 15
Open Access
Retina  |   December 2016
Early Postnatal Hyperoxia in Mice Leads to Severe Persistent Vitreoretinopathy
Author Affiliations & Notes
  • Paul G. McMenamin
    Department of Anatomy and Developmental Biology, Monash Biomedicine Discovery Institute, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Victoria, Australia
  • Rachel Kenny
    Department of Anatomy and Developmental Biology, Monash Biomedicine Discovery Institute, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Victoria, Australia
  • Sjakon Tahija
    Klinik Mata Nusantara, Jakarta, Indonesia
  • Jeremiah Lim
    Department of Optometry and Vision Sciences, The University of Melbourne, Melbourne, Victoria, Australia
  • Cecilia Naranjo Golborne
    Department of Anatomy and Developmental Biology, Monash Biomedicine Discovery Institute, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Victoria, Australia
  • Xiangting Chen
    Department of Anatomy and Developmental Biology, Monash Biomedicine Discovery Institute, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Victoria, Australia
  • Sheena Bouch
    Department of Anatomy and Developmental Biology, Monash Biomedicine Discovery Institute, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Victoria, Australia
  • Foula Sozo
    Department of Anatomy and Developmental Biology, Monash Biomedicine Discovery Institute, Faculty of Medicine, Nursing and Health Sciences, Monash University, Melbourne, Victoria, Australia
  • Bang Bui
    Department of Optometry and Vision Sciences, The University of Melbourne, Melbourne, Victoria, Australia
  • Correspondence: Paul G. McMenamin, Department of Anatomy and Developmental Biology, Monash University, Clayton, Victoria 3800, Australia; [email protected]
Investigative Ophthalmology & Visual Science December 2016, Vol.57, 6513-6526. doi:https://doi.org/10.1167/iovs.16-19928
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Paul G. McMenamin, Rachel Kenny, Sjakon Tahija, Jeremiah Lim, Cecilia Naranjo Golborne, Xiangting Chen, Sheena Bouch, Foula Sozo, Bang Bui; Early Postnatal Hyperoxia in Mice Leads to Severe Persistent Vitreoretinopathy. Invest. Ophthalmol. Vis. Sci. 2016;57(15):6513-6526. https://doi.org/10.1167/iovs.16-19928.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: To describe a mouse model of hyperoxia-induced vitreoretinopathy that replicated some of the clinical and pathologic features encountered in infants with severe retinopathy of prematurity and congenital ocular conditions such as persistent hyperplastic primary vitreous.

Methods: Experimental mice (C57BL/6J) were exposed to 65% oxygen between postnatal days (P)0 to P7 and studied at P10, P14, and 3, 5, 8, 20, and 40 weeks. Controls were exposed to normoxic conditions. Fundus imaging and fluorescein angiography were performed at all time points, and spectral-domain optical coherence tomography (SD-OCT) and electroretinography were performed at 8- and 20-week time points. Eyes were processed for resin histology, frozen sections, and retinal whole mounts. Immunostaining was performed to visualize vasculature isolectin B4 (Ib4), collagen type IV, glial fibrillary acidic protein, and α-smooth muscle actin.

Results: Early exposure to hyperoxia resulted in bilateral vitreous hemorrhages at 3 weeks. From 5 weeks onward there were extensive zones of retinal degeneration, scarring or gliosis, retinal folding, and detachments caused by traction of α-smooth muscle actin–positive vitreous membranes. Tortuous retinal vessels, together with hyperplastic and persistence of hyaloid vessels are evident into adulthood. In the early stages (P10–3 weeks), branches from the tunica vasculosa lentis (TVL) supplied the marginal retina until retinal vessels were established. The peripheral retina remained poorly vascularized into adulthood. Electroretinography revealed 50% to 60% diminution in retinal function in adult mice that strongly correlated with vitreal changes identified using SD-OCT.

Conclusions: This animal model displays a mixture of vitreoretinal pathologic changes that persist into adulthood. The model may prove valuable in experimental investigations of therapeutic approaches to blinding conditions caused by vitreous and retinal abnormalities.

Landmark clinical observations and animal research in the 1950s alerted neonatologists to the danger of providing high concentrations of supplemental O2 during intensive care of premature neonates in respiratory distress.13 It became apparent that hyperoxia followed by normoxia was a critical causative factor in retinopathy of prematurity (ROP) or retrolental fibroplasias, as it was then known. While supplemental O2 ensured survival of neonates with pulmonary insufficiency, relative hyperoxia in the eyes caused by high blood O2 saturation levels results in an interruption to normal retinal vascular development (phase I of ROP). When infants are returned to breathing room air, the ensuing relative hypoxia experienced by the growing and maturing retina results in excess vascular endothelial growth factor (VEGF) production that drives an aggressive neovascular response in the retina and vitreous (phase II of ROP) that can lead to retinal detachment, intravitreal hemorrhages, persistence of hyaloid vessels in a minority of cases, and subsequent vision loss.46 
Normal retinal vascularization in humans commences around weeks 16 to 17 gestational age (GA) and grows centrifugally at 0.1 mm/week to almost reach the margins of the retina at birth.7 The eye in some nonprimate mammals with holangiotic retinae is less mature at birth and retinal vascularization is often incomplete compared to the precocial state in the human eye. Researchers have exploited this situation by exposing newborn or very young animals, including cats, dogs, rats, and rabbits, to hyperoxic conditions at birth to “mimic” the human premature infant.2,5,8,9 As the high O2 exposure occurs at term, the pathology in these models is referred to as oxygen-induced retinopathy (OIR).10 At birth, mice still have a functioning and prominent hyaloid vasculature and the retina is largely avascular. Superficial vessels growing radially from the optic nerve head to reach the retinal margin by postnatal day 8 (P8) coincide with regression of hyaloid vasculature.11 Vertical sprouting occurs around P7 to form deep and intermediate plexi of capillaries, a process that continues centrifugally to reach the retinal margins between P12 and P15.12 This process is roughly analogous to the manner in which retinal vascularization occurs in humans in the third trimester. The timing of hyperoxia exposure between P7 and P12 in the well-described OIR mouse model13,14 was chosen to mimic the equivalent stage of ocular development in very preterm or moderate-preterm human infants. In this model, mouse pups are exposed to 75% oxygen between P7 and P12 and when examined around P17 to P21, the retina is characterized by a central quantifiable zone of avascularity and patches of neovascular tufts breaching the inner limiting membrane in the midperipheral retina and entering the vitreous, both resembling to some extent features of human ROP.13,14 Several pathognomonic features of severe (stage 4, 5) ROP such as retinal detachment, vitreous hemorrhages, cicatricial changes (retinal scarring), and avascularity of peripheral retina are not a feature of this model.11,15,16 Secondly, electroretinographic studies have shown that both photoreceptor (a-wave) and bipolar cell (b-wave) function while significantly reduced at P1817 have largely recovered by 8 weeks.18 Lastly, the hyaloid system has regressed as normal by week 5, and the retina and its vasculature, despite the earlier disturbances, are described as normal.14,17,19 
The hyaloid artery, entering the eye early in development, courses through and supplies the vitreous via branches, the vaso hyaloidea propria (VHP), before forming the tunica vasculosa lentis (TVL) around the lens posteriorly. The pupillary membrane, arising from developing iris vessels, covers the lens anteriorly. In humans, the hyaloid system undergoes regression in the third trimester and is absent at birth.7 In mice at P0 while the retinal vessels are just commencing growth, the hyaloid system has not commenced its regression.11 WNT7b, expressed by macrophages, has been identified as a signal for hyaloid vessel regression in the postnatal mouse eye, 20,21 as has the WNT receptor Frizzled. 
Ashton9 and later Browning et al.22 described severe vitreoretinopathy in mice when the period of hyperoxia exposure was immediately after birth (P0–P5). They noted the occurrence of vitreous hemorrhages, retinal detachments, and neovascular tufts in the vitreous that were thought to arise from superficial retinal vessels.9,22 However, there have been no comprehensive long-term studies or detailed analyses of hyaloid and retinal vascular changes and the functional implications of such severe disruptions to normal eye development. The present study aimed to characterize the clinical, histopathologic, and functional changes in the eyes of mice at multiple time points up to adulthood that had been exposed to 65% O2 from P0 to P7. Our data demonstrate that such early hyperoxia conditions produce severe vitreoretinal disturbances, some of which bear similarities to severe ROP as well as congenital ocular conditions such as persistent hyperplastic primary vitreous, sometimes now referred to as persistent fetal vasculature. 
Materials and Methods
Animals
All animal experiments were approved by the Monash University Animal Ethics Committee (MARP/2014/073 and MARP/2011/061) and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. A total of 134 C57BL/6J mice (Animal Resources Centre, Murdoch, WA, USA and Monash Animal Services, VIC, Australia) of both sexes were used in the present study (control normoxia, n = 69; hyperoxia exposed, n = 85), and sex made no difference to the outcome. Genotyping revealed the absence of the Crb1rd8 mutation in these mice.23 
Exposure to Hyperoxia
An established mouse hyperoxia model (P0–P7 65% O2 model) was used in the present study.24 Pregnant females were placed at embryonic (E) 14 in sealed clear plastic chambers (405 × 205 × 185 mm; length × width × height); floor area 501 cm2 (Tecniplast, Buguggiate [VA], Italy), in which a continuous gas flow could be introduced to maintain a uniform O2 concentration, and left to give birth to their litters naturally at term. Experimental litters were randomly assigned to normoxia (ambient room air; 21% O2) or hyperoxia exposure (65% O2). Hyperoxia exposure was achieved by the mixing of medical-grade dry air and medical-grade O2. Oxygen and carbon dioxide concentrations were continuously monitored by a gas analyzer (Servoflex MiniMP 5200; Servomex, Valley Point, Singapore). Litters were continuously exposed to hyperoxia or normoxia from birth for 7 days (P0–P7), with a brief interruption daily for animal care. Food and water were provided ad libitum for the duration of the experiment. For the 65% O2 groups, the total gas flow (medical O2 and medical air combined) into each chamber was 1.8 to 2.5 L/min. For the control groups (normoxia; air breathing) there was no gas flow artificially introduced into the chambers; gas exchange occurred passively across a microbiological filter. For litters exposed to 65% O2, each dam was paired with another dam and her litter, and the dams were rotated between normoxia and hyperoxia-exposed litters every 24 hours during the hyperoxic gas exposure; this usually prevents loss of body weight in dams and pups. After the 7-day exposure period, all mice were returned to conventional cages and breathed room air until necropsy. Animal house conditions were maintained on 12-hour day/night cycles throughout the experimental period. Light intensity in the cages housing control and hyperoxia-exposed mice in the animal facility never exceeded 103 lux (range, 23–103 lux) (Lux & Fc Light meter; Dick Smith, Melbourne, VIC, Australia). Animal weight was monitored at regular intervals, and no difference in growth of the mice up to 8 weeks was noted between control and hyperoxia mice or between sexes in both groups (data not shown). 
Retinal Fundus Imaging
At various time points (P10, P14, 3, 5, 8, 20, 40 weeks), the eyes of anesthetized mice from both hyperoxia and control groups were examined using the multimodal fundus Micron III camera (Phoenix Research Laboratories, Pleasanton, CA, USA) as previously described.25 
Calculation of Hemorrhage Area in Fundus Images
The fundus was traced using the circle tool in ImageJ (National Institutes of Health, Bethesda, MD, USA), and size of hemorrhage was calculated as a percentage of the total fundal area. The fundus was not readily visible in P10 and P14 hyperoxia mice due in part to the presence of the still-prominent hyaloid vasculature and in some cases the vitreous hemorrhages. 
Quantitation of Vessel Tortuosity in Retinal and Hyaloid Vasculature
Vessel tortuosity index was calculated on images of the retinal fundus (3 weeks onward) using a previously published protocol that is described for retinal whole mounts.26 
Assessment of Retinal Function Using Electroretinography (ERG)
At 8 weeks (hyperoxia, n = 8; controls, n = 9) and 20 weeks (hyperoxia, n = 8; controls, n = 6) of age, groups of mice in their habitual cages were dark-adapted overnight in a darkened room (maximum of 12–15 hours). After dark adaptation, the animals were anesthetized with an intraperitoneal injection of a combination of ketamine (67 mg/kg) and xylazine (13 mg/kg). Dilation of the pupil was induced with tropicamide eye drops, and corneal anesthesia was provided via proxymetacaine eye drops (Alcaine 0.5 %; Allergan, Frenchs Forest, NSW, Australia). Animals were wrapped in a blanket and placed on a heat pad (37°C) to maintain core temperature. Artificial tear lubricant (Celluvisc, Allergan) was placed on the cornea. Two custom-made chlorided (electroplated in 0.9% saline, for 15 seconds) silver (99% silver; A&E Metals, Marrickville, NSW, Australia) electrodes were then placed onto the animal: one on the surface of the central cornea, and a second differencing negative electrode placed around the circumference of the globe. Both electrodes were referenced to a ground needle electrode (F-E2-30; Grass Telefactor, West Warwick, RI, USA) inserted subcutaneously into the tail. Throughout functional testing, animals were continuously monitored for body temperature, arousal, and vital signs. A Ganzfeld sphere (Photometric Solutions International, Werribee, VIC, Australia) was used to deliver luminous energy ranging from −6.26 to 2.07 log cd.s/m2 calibrated with an IL1700 integrating photometer (International Light Technologies, Peabody, MA, USA). Signals were collected with band-pass settings of 0.3 to 1000 Hz (−3 dB) and digitized at an acquisition rate of 4 kHz. To ensure that differences in the ERG did not arise from impaired pupillary dilation, pupil size was measured by taking a photograph of the iris and pupil. 
Electroretinographic waveforms were analyzed using well-established methods.27 In brief, the photoreceptor response was modeled to return the amplitude of the photoreceptoral response by fitting the initial negative-going a-wave using a delayed-Gaussian function. Rod bipolar cell–mediated function was quantified in terms of its maximal amplitude (μV) by modeling the b-wave amplitudes as a function of stimulus energy with a hyperbolic curve. Oscillatory potentials (OPs), which provide an index of inner retinal inhibitory pathways involving amacrine cells, were isolated using a digital band-pass filter (−3 dB; 50–250 Hz). The amplitude of the largest OP was quantified at a stimulus energy of 0.72 to 2.07 log cd.s/m2. The ganglion cell dominant response was measured as the peak-to-trough amplitude of the scotopic threshold response (STR) recorded at −5.25 log cd.s/m2. For each of the above ERG parameters, the amplitude for the two eyes was averaged and then expressed as a percentage of the average amplitude for all control eyes. Group data are given as mean (± standard error of the mean). 
Intraocular pressure was measured using a rebound tonometer (Tonolab; iCare, Oy, Finland). 
Spectral-Domain Optical Coherence Tomography (SD-OCT)
Immediately following ERG recording, the 8-week (n = 8 hyperoxia, n = 9 controls) and 20-week (n = 8 hyperoxia, n = 6 controls) groups of mice were placed on the rodent alignment stage attached to the Envisu SD-OCT imaging device (Bioptigen, Inc., Morrisville, NC, USA). Lubricant was added to the cornea (Systane; Alcon, Fort Worth, TX, USA) and any excess liquid was absorbed using a cotton tip. Care was taken in alignment to allow volumetric scans centered on the optic nerve head (ONH; 1 .4 × 1.4 × 1.57 mm) to be obtained from both eyes, where a randomly selected eye was imaged first. The acquired volume consisted of 1000 A-scans per 100 B-scans. Imaging included the vitreous, which allowed us to visualize any abnormal vasculature or membranes. Animals were culled after examination and tissues collected as described below. 
Quantitation of Hyaloid Vasculature Using SD-OCT
Optical coherence tomography images were exported as a Tiff stack (100 B-scans, each 14 μm apart) and analyzed using FIJI software.28 The central 30 slices encompassing the nerve were collapsed into a single image by obtaining the maximum intensity of each slice. Next, images were converted to 8-bit binary using automatic thresholding algorithm Renyi Entropy.29 A median filter was applied to remove single pixel-wide specks before the hyaloid area in the vitreous was quantified in micrometers within an oval region of interest (width × height: 800 × 250 pixels). 
Quantification of Total Retinal Thickness
Total retinal thickness (TRT) was measured, using SD-OCT images, from the inner limiting membrane to Bruch's membrane as previously described elsewhere.30 In brief, measurements were made from the B-scans containing the optic nerve using FIJI. Retinal thickness was measured using calipers perpendicular to Bruch's membrane at a distance of 200 μm from the center of the optic nerve. Thickness values from nasal and temporal regions across five B-scans were averaged to give a global thickness measure for each eye. 
Tissue Processing and Histology
For histology and whole-mount immunostaining, euthanized animals were perfusion fixed (4% paraformaldehyde [PFA]) and eyes enucleated. The eyes were allocated to either retinal whole-mount staining, glycomethacrylate (GMA) resin histology, or frozen sections. Serial resin sections (5 μm) where possible including the optic nerve–pupil axis were collected and stained with hematoxylin and eosin (H&E). A selection of eyes (3, 5, 8, and 20 weeks, n = 4/time point) were frozen and sectioned (6 μm) for immunostaining to detect glial fibrillary acidic protein (GFAP; BD Pharmingen, San Diego, CA, USA), α-smooth muscle actin (α-SMA; Sigma-Aldrich Corp., St. Louis, MO, USA) to characterize the preretinal membranes, and collagen type IV (Coll IV; AbD Serotec, Oxford, UK) to stain superficial retinal and hyaloid vessels.31 Primary antibodies were visualized with AF546 goat anti-mouse secondary antibody (Molecular Probes, Mulgrave, VIC, Australia). 
Retinal Whole-Mount Immunohistochemistry
The fixed left eye from some animals was dissected in phosphate-buffered saline (PBS). Following the method previously described for processing whole mounts,32 the neural retinae were dissected from the retinal pigment epithelial–choroid–sclera and flattened by radial incisions.11 Using the staining regimen described by Tual-Chalot et al.,31 the retinae were permeabilized with an overnight incubation in 0.5% Triton X in PBS. They were incubated overnight in isolectin B4 (Ib4; Vector Laboratories, Inc., Burlingame, CA, USA) diluted in PBS and then incubated for 5 hours with a Strep-647 secondary antibody (Molecular Probes). Hoechst 33342 (Molecular Probes) was added to all preparations as a nuclear stain. All the steps outlined above were performed at room temperature on a rocker. Stained retinae were imaged using an SP5 five-channel inverted confocal microscope (Leica, North Ryde, NSW, Australia). Digital scanning of entire stained retinae were obtained using an AF6000LX wide-field microscope (Leica). 
Calculation of Area of Avascularity in Retinal Whole Mounts
The extent of avascularity in experimental mouse retinal samples was analyzed per a modified version of a method described by Connor et al.14 Tiled images of retinal samples were analyzed with the polygon selection tool available in the open-source FIJI software.28 
Statistical Analysis
The area of vitreous hemorrhages, number of hyaloid vessels, vessel tortuosity index, area of unvascularized retina, and ERG parameters in normoxia and hyperoxia mice were compared using 1-way or 2-way ANOVA (depending on experimental variables) and Tukey's post hoc comparison. Where data were obtained from both eyes (quantitation of vessels from Micron imaging, ERG, OCT), these were used to calculate a mean per animal. In cases where only one eye was available (retinal whole mounts), only one eye per animal contributed to the data point. 
Results
Clinical Fundus and Fluorescein Angiography Findings in P0 to P7 Hyperoxia-Exposed Mice
In vivo fundus examination of the eyes of P10 and P14 mice exposed to 65% O2 between P0 and P7 by bright-field and fluorescein angiography (FA) was obscured by large dilated vessels of the hyaloid system (TVL, VHP) (Fig. 1). By week 3 (P21), vitreous hemorrhages were commonly encountered (11/13 animals, 10 bilateral), as were extensive retinal detachments (12/13 animals, 11 bilateral). The hemorrhages occupied on average between 10% and 15% of the fundus field of view; however, this diminished rapidly by week 5 (Fig. 1; Supplementary Fig. S1). At later time points (5–40 weeks) there was evidence of hypopigmented zones of marked retinal degeneration and retinal folding (Fig. 1, left column; full data, Supplementary Fig. S2). Retinal vessels of hyperoxia-exposed mice were dilated, highly tortuous, and displayed evidence of nondichotomous branching when compared to normoxia (control) mice (Supplementary Fig. S3). This abnormal retinal vascular pathology persisted until 40 weeks in the hyperoxia-exposed mice (Fig. 1). At none of the time periods examined did there appear to be a central avascular zone. At later time points there appeared to be patchy peripheral avascular zones with leakage of vessels at the interface between the vascularized and avascular retina (Supplementary Fig. S4). 
Figure 1
 
The clinical appearance of the mouse eye as seen using Micron III retinal fundus imaging at various recovery times after P0 to P7 hyperoxia exposure. Left column: bright-field (BF) mode; middle column: fluorescein angiography (FA) mode focused on the retinal vessels; left column: FA focused on the hyaloid vessels. The retina and retinal vessels cannot be visualized clearly in P10 and P14 due the hyaloid artery and its branches, which are hypertrophic at both time points. Note the vitreous hemorrhages at 3 weeks, and retinal detachments (arrowheads) and focal degenerative changes (pale focal areas) from week 8 onward together with the highly tortuous retinal and persistent hyaloid vessels at all time points. Further time points are shown in Supplementary Figure S2.
Figure 1
 
The clinical appearance of the mouse eye as seen using Micron III retinal fundus imaging at various recovery times after P0 to P7 hyperoxia exposure. Left column: bright-field (BF) mode; middle column: fluorescein angiography (FA) mode focused on the retinal vessels; left column: FA focused on the hyaloid vessels. The retina and retinal vessels cannot be visualized clearly in P10 and P14 due the hyaloid artery and its branches, which are hypertrophic at both time points. Note the vitreous hemorrhages at 3 weeks, and retinal detachments (arrowheads) and focal degenerative changes (pale focal areas) from week 8 onward together with the highly tortuous retinal and persistent hyaloid vessels at all time points. Further time points are shown in Supplementary Figure S2.
Exposure of Mice to Early Hyperoxia (P0–P7) Disrupts the Normal Hyaloid Vascular Regression
During normal development in humans and other mammals, the hyaloid artery arises from the ONH and passes anteriorly through the vitreous, which it supplies via the VHP, before forming a regular capillary bed (the TVL) around the lens. This entire system undergoes regression in the last trimester in humans, and its abnormal persistence is a feature of a minority of cases of severe ROP33 and persistent hyperplastic primary vitreous.34 In the present study, while three to five major branches of the hyaloid artery were patent and highly visible during FA examination of normoxia (control) mice at P10, P14, and 3-week mice, these vessels had undergone normal regression by 5 weeks (Fig. 2A; Supplementary Fig. S4, right) as previously described.19 By contrast, in P0 to P7 hyperoxia-exposed mice the TVL, VHP, and hyaloid artery were more tortuous and dilated at 3 weeks compared to controls and furthermore persisted until 40 weeks of age (Fig. 1, right; Fig. 2A). 
Figure 2
 
Quantitative analysis of the number of hyaloid vessels observed by in vivo fundus imaging in P0 to P7 hyperoxia-exposed mice and controls (A); the tortuosity index of those hyaloid vessels (B) and retinal vessels (C). Values are mean ± SEM, ***P < 0.001 (1-tailed t-test). (A) Control group sizes, n = 11, 5, 20, 8, 16; experimental group sizes, n = 26, 9, 18, 8, 29. (B) Control group sizes, n = 11, 4, 13, 0, 3; experimental groups, n = 26, 9, 18, 8, 29. (C) Control group sizes, n = 11, 5, 20, 7, 16; experimental group sizes, n = 26, 9, 18, 8, 29.
Figure 2
 
Quantitative analysis of the number of hyaloid vessels observed by in vivo fundus imaging in P0 to P7 hyperoxia-exposed mice and controls (A); the tortuosity index of those hyaloid vessels (B) and retinal vessels (C). Values are mean ± SEM, ***P < 0.001 (1-tailed t-test). (A) Control group sizes, n = 11, 5, 20, 8, 16; experimental group sizes, n = 26, 9, 18, 8, 29. (B) Control group sizes, n = 11, 4, 13, 0, 3; experimental groups, n = 26, 9, 18, 8, 29. (C) Control group sizes, n = 11, 5, 20, 7, 16; experimental group sizes, n = 26, 9, 18, 8, 29.
Quantitation of Fundus Images Reveals Persistent Tortuosity of Retinal and Hyaloid Vessels in P0 to P7 Hyperoxia-Exposed Mice
Normally, mouse retinal vessels radiate in a spoke-like pattern from the ONH with approximately five or six arterioles and five or six venules in alternating pairs (Supplementary Fig. S4, middle).35 Since increased tortuosity of the retinal vessels is a predictor of clinical outcomes in human ROP,16 we quantified tortuosity in our P0 to P7 hyperoxia-exposed mice. Analysis of vessels (arteries and veins) in FA revealed highly significant differences between control (normoxia) and hyperoxia-exposed groups in tortuosity index of both hyaloid and retinal vessels (Figs. 2B, 2C). While hyaloid vessels became progressively less tortuous with recovery time (Fig. 2B), the index in retinal vessels remained approximately similar from 3 to 40 weeks (Fig. 2C). 
SD-OCT Reveals Severe and Persistent Pathology in the Vitreous and Retina of P0 to P7 Hyperoxia-Exposed Mice
Subgroups of control and P0 to P7 hyperoxia-exposed mice were investigated using SD-OCT at 8 and 20 weeks immediately prior to ERG studies. In control mice, normal retinal topography and anatomic arrangement of the layers corresponding to previous descriptions were observed (Fig. 3A).36 In 8-week hyperoxia-exposed mice there were conspicuous pathologic changes including retinal detachment, retinal thickening, disorganization of retinal layers, retinal folds, and rosettes (Figs. 3B, 3C). The most noticeable feature was the presence of vitreal opacities or preretinal membranes that had distinct adhesion points to the puckered or elevated retinal surface. These frequently overlaid areas of severe retinal disruption, folding, and detachment (Figs. 3B, 3C). This corresponded to the clinical fundus appearance in these mice described above. Isolated hyaloid vessel profiles representing persistent parts of the VHP were connected to the TVL around the lens (Fig. 3B). Three-dimensional reconstruction of the SD-OCT data demonstrated clearly the disorganized retina, vitreal membranes, retinal detachment, and persistent abnormal hyaloid vessels compared to controls (data not shown). By 20 weeks, while the changes were slightly less marked, there were still extensive regions of preretinal membranes, retinal tenting, and persistent hyaloid vessels extending to the posterior surface of the lens (Fig. 3D). 
Figure 3
 
SD-OCT imaging of control (A) and hyperoxia-exposed mice at 8 weeks (B, C) and 20 weeks (D). Note the dense membranes in the vitreous (arrowheads) in hyperoxia-exposed mice and the manner in which the retina is puckered or elevated where these membranes attach to the inner retina. Extensive retinal detachment is clearly evident (large arrow [B]). Isolated profiles of hyaloid vessels are also present in 8- and 20-week-old eyes. Note the large rosette in the retina at 8 weeks (black arrowhead [C]).
Figure 3
 
SD-OCT imaging of control (A) and hyperoxia-exposed mice at 8 weeks (B, C) and 20 weeks (D). Note the dense membranes in the vitreous (arrowheads) in hyperoxia-exposed mice and the manner in which the retina is puckered or elevated where these membranes attach to the inner retina. Extensive retinal detachment is clearly evident (large arrow [B]). Isolated profiles of hyaloid vessels are also present in 8- and 20-week-old eyes. Note the large rosette in the retina at 8 weeks (black arrowhead [C]).
Significantly greater areas of the vitreous were occupied by persistent abnormal hyaloid vessels and preretinal membranes in the hyperoxia-exposed groups compared with control at both 8 and 20 weeks of age (Supplementary Fig. S5). Total retinal thickness, as measured by SD-OCT, showed no difference between hyperoxia and control groups at 8 or 20 weeks; however, there was a trend toward higher TRT values in the hyperoxia-exposed mice at 20 weeks due to retinal detachment and tenting due to upward pressure from hyaloid scar tissue at the vitreous–retinal interface (Supplementary Fig. S5). 
ERG Shows Abnormal, Longstanding Functional Changes in Hyperoxia Mice
In light of the importance of the loss of a lifetime of visual function for infants with severe ROP or congenital vitreoretinopathies, we sought to determine if the persistence of marked retinal pathology into adulthood that was noted in the P0 to P7 hyperoxia-exposed mice was reflected by long-term altered retinal function. Figure 4A shows averaged responses from a control (thin trace) and a treated eye (thick trace) in an 8-week-old mouse. The top traces show the response to a bright flash of light, which provides information about the integrity of photoreceptors (a-wave), ON bipolar cells (b-wave), and inhibitory pathways involving amacrine cells (OPs). The lower trace shows the response to a very dim flash. This produces the STR, which is known to be dominated by ganglion cell responses in rodents.37 Figure 4B shows that photoreceptor (a-wave), bipolar cell (b-wave), amacrine cell (OPs), and ganglion cell (STR) function are all significantly attenuated in P0 to P7 hyperoxia-exposed mice (P < 0.01) when compared to control groups at 8 weeks. 
Figure 4
 
Retinal function as assessed by ERG in control and OIR mice (65% O2 P0–P7) at 8 weeks (A, B) and 20 weeks (C, D). (A) Average group waveforms for control (thin traces) and OIR mice (thick traces) elicited with a bright (a-wave, b-wave, oscillatory potentials46 at 1.2 log cd.s/m2) and dim (scotopic threshold response, −5 log cd.s/m2) flash. (B) Size (average ± SEM) of each of the major ERG components expressed relative to the group average of control eyes (OIR response/average control, %). (C) Average waveforms at 20 weeks. (D) Relative response at 20 weeks. At 8 weeks, n = 8 hyperoxia, n = 9 controls; 20 weeks, n = 8 hyperoxia, n = 6 controls.
Figure 4
 
Retinal function as assessed by ERG in control and OIR mice (65% O2 P0–P7) at 8 weeks (A, B) and 20 weeks (C, D). (A) Average group waveforms for control (thin traces) and OIR mice (thick traces) elicited with a bright (a-wave, b-wave, oscillatory potentials46 at 1.2 log cd.s/m2) and dim (scotopic threshold response, −5 log cd.s/m2) flash. (B) Size (average ± SEM) of each of the major ERG components expressed relative to the group average of control eyes (OIR response/average control, %). (C) Average waveforms at 20 weeks. (D) Relative response at 20 weeks. At 8 weeks, n = 8 hyperoxia, n = 9 controls; 20 weeks, n = 8 hyperoxia, n = 6 controls.
All components of the ERG remained significantly attenuated in hyperoxia-exposed mice compared with controls at 20 weeks of age (Figs. 4C, 4D) (P < 0.01). In particular, Bonferroni post hoc analysis shows that the STR was significantly more affected than all other ERG components (P < 0.05), suggesting a further compromising of ganglion cell function with age in hyperoxia-exposed mice. No significant difference was found in pupil size (1.5 ± 0.1 mm) between the hyperoxia-exposed and control groups of mice at either 8 or 20 weeks. In addition, there was no significant difference in intraocular pressure between the control and the hyperoxia-exposed groups at 8 weeks (hyperoxia, 9 ± 1 mm Hg versus control, 8 ± 1 mm Hg) or 20 weeks (hyperoxia, 10 ± 1 mm Hg versus control, 9 ± 1 mm Hg). 
Given that the hyperoxia-exposed group showed more vitreal opacities and hyaloid vascular remnants, we performed a correlation analysis in order to determine if there was a relationship between both measures and retinal function. There was a poor relationship between retinal function and retinal thickness (Supplementary Fig. S4C); however, there was a very strong correlation between the photoreceptoral (Supplementary Fig. S4D), bipolar cell (Supplementary Fig. S4E), and ganglion cell (Supplementary Fig. S4F) function and vitreal opacities identified using SD-OCT. 
Early (P0–P7) Hyperoxia Induces Severe Vitreoretinopathy
Ocular histology of P10 and P14 eyes (Fig. 5) revealed that full differentiation of all retinal layers was complete although the photoreceptor layer was not fully mature, a situation that appeared ostensibly similar to that for normoxia controls (not shown). However, there was minimal evidence of retinal vasculature but marked hyperplasia of the VHP and the TVL, particularly at the P14 time point at which the TVL formed a thick condensation on the posterior lens surface (Figs. 5D, 5E). The vessels in the periphery of this abnormal TVL sent branches onto the peripheral retinal surface (Fig. 5F). This was confirmed in retinal whole mounts (not shown). Observations in 3-week-old hyperoxia-exposed mice confirmed the pathologic changes observed clinically by fundus bright-field imaging, FA imaging, and SD-OCT. Namely, these eyes were characterized by vitreous hemorrhages, vitreous or preretinal membranes, persistent hyaloid artery and associated TVL and VHP, retinal folds, and rosettes, together with various degrees of retinal degeneration including complete loss of the outer retina (Figs. 6A–C). Most of these changes were consistently observed in 5-week-old (data not shown) and 8-week-old mice (Figs. 5D–F) with the exception that vitreous hemorrhages had partly resolved along with the preretinal membranes and persistent VHP (Figs. 6D, 6E), which were less conspicuous than in P10, P14, and 3-week-old mice (Fig. 5). By 20 and 40 weeks, vitreal or preretinal membranes were not as frequently encountered in sections (data not shown); however, capillaries were still present on the posterior lens surface (the persistent TVL) and in the vitreous (the persistent VHP) near the ONH (Fig 6E; Supplementary Fig. S5G). In eyes from 3 to 40 weeks there was marked focal loss of outer retinal layers (Figs. 6B, 6C, 6F) and distinct patches of RPE loss leading to retinal tissue abutting directly onto the choroid (not shown). 
Figure 5
 
Histopathology of representative eyes from P10 (AC) and P14 (DF). At P10, note the prominent hyaloid vessels and TVL. The retinal morphology appears ostensibly normal for the age of the mouse. At P14 the hyaloid artery and TVL are hyperplastic and extend onto the retina near the optic nerve head and form a thick distinct capsule on the posterior lens surface (E) that extends laterally and appears to send branches to supply the peripheral retina (F). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Figure 5
 
Histopathology of representative eyes from P10 (AC) and P14 (DF). At P10, note the prominent hyaloid vessels and TVL. The retinal morphology appears ostensibly normal for the age of the mouse. At P14 the hyaloid artery and TVL are hyperplastic and extend onto the retina near the optic nerve head and form a thick distinct capsule on the posterior lens surface (E) that extends laterally and appears to send branches to supply the peripheral retina (F). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Figure 6
 
Histopathology of representative eyes from 3-week (AC) and 8-week (DF) time points in hyperoxia-exposed mice. Low magnification (insets) reveals the general layout of the mouse eyes and shows normal anterior segment morphology. At 3 weeks of age, a large prominent hyaloid artery along with vitreal membranes (black arrow) is clearly identifiable bridging between the connective tissue around the hyaloid artery and the elevated retinal surface. Extensive retinal folding and rosettes are also visible. Vitreous hemorrhages (red arrows) are clearly identifiable (A, B), as are focal regions of extensive outer retinal degeneration (B, C). At 8 weeks of age, vitreous membranes (arrow in [D]) are still evident associated with elevated and disrupted retinal layers. Numerous hyaloid or epiretinal vessels (arrows in [E]). Note extensive outer retinal disruption and folding and loss of outer retinal layers (arrowhead in [F]). Control normoxia mice were normal and revealed no abnormal pathology at all time points examined (Supplementary Fig. S3). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Figure 6
 
Histopathology of representative eyes from 3-week (AC) and 8-week (DF) time points in hyperoxia-exposed mice. Low magnification (insets) reveals the general layout of the mouse eyes and shows normal anterior segment morphology. At 3 weeks of age, a large prominent hyaloid artery along with vitreal membranes (black arrow) is clearly identifiable bridging between the connective tissue around the hyaloid artery and the elevated retinal surface. Extensive retinal folding and rosettes are also visible. Vitreous hemorrhages (red arrows) are clearly identifiable (A, B), as are focal regions of extensive outer retinal degeneration (B, C). At 8 weeks of age, vitreous membranes (arrow in [D]) are still evident associated with elevated and disrupted retinal layers. Numerous hyaloid or epiretinal vessels (arrows in [E]). Note extensive outer retinal disruption and folding and loss of outer retinal layers (arrowhead in [F]). Control normoxia mice were normal and revealed no abnormal pathology at all time points examined (Supplementary Fig. S3). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Immunolabeling of frozen sections demonstrated moderate inner retinal gliosis as evidenced by increased immunoreactivity for GFAP at 3 weeks (data not shown) and 8 weeks (Fig. 7). In 3- and 8-week hyperoxia-exposed eyes, preretinal membranes were found to contain extensive quantities of type IV collagen, which is normally detectable only around larger-caliber vessels in the superficial retina and is absent in the normal vitreous (Fig. 7). Immunolabeling also detected α-SMA associated with vitreous membranes (Fig. 7). 
Figure 7
 
Immunostained frozen sections of control and hyperoxia-exposed eyes at 8 weeks showing astrogliosis in the superficial retina (abnormal GFAP immunoreactivity, top right) and abnormal deposition of type IV collagen (Coll IV; middle right) together with the presence of α-smooth muscle actin (α-SMA)-positive staining in the superficial retina and vitreal membranes (bottom right). Bar: 50 μm.
Figure 7
 
Immunostained frozen sections of control and hyperoxia-exposed eyes at 8 weeks showing astrogliosis in the superficial retina (abnormal GFAP immunoreactivity, top right) and abnormal deposition of type IV collagen (Coll IV; middle right) together with the presence of α-smooth muscle actin (α-SMA)-positive staining in the superficial retina and vitreal membranes (bottom right). Bar: 50 μm.
Early Hyperoxia Exposure in Mice Leads to Marked Changes in Retinal Vasculature
One of the critical issues in ROP in humans (stage 2 upward) is the incomplete vascularization of the peripheral retina and leakage into the vitreous from abnormal superficial retinal vessels at the margins of the vascularized zone.38 In normoxia control mice at 5 to 8 weeks onward, the retinal vasculature could be seen to have grown radially and reached the peripheral margins of the neural retina (Fig. 8A) and consisted of three laminae of vessels or capillary beds (superficial, deep, and intermediate layers) (Figs. 8B, 8C), which conforms to previous descriptions of mouse retinal vasculature.35 While a classical feature of the conventional 75% O2 P7 to P12 OIR model when studied at around P17 to P21 is a central avascular or vaso-obliterative zone, this was not a characteristic of the present 65% O2 P0 to P7 model (Figs. 8D–I). Instead, a less dense plexus of smaller superficial retinal vessels had grown more haphazardly in a radial direction toward the periphery but had failed to completely vascularize the margin by week 3 (day 21). In addition, there was largely an absence of the deep capillary plexus (Figs. 8D–I). Only in the central retina were occasional larger-caliber vessels noted penetrating into the deeper layers of the retina (Figs. 8F, 8I). By week 5 the pattern of highly tortuous morphology with nondichotomous branching radiating arteries and veins observed in fundus imaging (Fig. 1) was confirmed in lectin- and antibody-stained retinal whole mounts (Figs. 8D–F). In 20-week-old hyperoxia-exposed mice, an avascular zone persisted in the peripheral retina (Fig. 8G); vessels were not arranged in the normal distinct laminae (Fig. 8H) and had failed to form the normal deeper or intermediate branches (Fig. 8I), and the density of vessels was abnormally low. Quantitative analysis confirmed that there was consistent failure of vascularization of the peripheral retina (Supplementary Fig. S5H). 
Figure 8
 
Retinal whole mounts from control and hyperoxia-exposed mice at 5 and 20 weeks stained with Ib4 (red/magenta). The entire retinal vasculature is shown in the left column (A, D, G), and confocal three-dimensional reconstructions or Z profiles of retinal vessels in the central and peripheral retina are in the right column. The latter show the extent of penetration of the capillaries into the neural retina. The nuclear layers of the retina (ganglion cell layer [GCL], inner nuclear layer [INL], and outer nuclear layers [ONL]) are visualized by Hoechst nuclear staining (blue) and best seen in control retinal profiles (B, C). The vitread aspect of the retina is orientated superiorly. Note that at 5 weeks of age in normal eyes there are already three layers of vessels in the central and peripheral retina: a superficial, an intermediate (inner plexiform layer), and a deep (outer plexiform layer) (labeled 1, 2, 3, respectively in [B]). Note the mainly superficial vessels from 5 weeks onward in the hyperoxia mice; however, large-caliber vessels are seen penetrating the central retina in 20-week hyperoxia-exposed mice. Scale bars: 1 mm, right column (A, D, G); 50 μm (B, C, E, F, H, I).
Figure 8
 
Retinal whole mounts from control and hyperoxia-exposed mice at 5 and 20 weeks stained with Ib4 (red/magenta). The entire retinal vasculature is shown in the left column (A, D, G), and confocal three-dimensional reconstructions or Z profiles of retinal vessels in the central and peripheral retina are in the right column. The latter show the extent of penetration of the capillaries into the neural retina. The nuclear layers of the retina (ganglion cell layer [GCL], inner nuclear layer [INL], and outer nuclear layers [ONL]) are visualized by Hoechst nuclear staining (blue) and best seen in control retinal profiles (B, C). The vitread aspect of the retina is orientated superiorly. Note that at 5 weeks of age in normal eyes there are already three layers of vessels in the central and peripheral retina: a superficial, an intermediate (inner plexiform layer), and a deep (outer plexiform layer) (labeled 1, 2, 3, respectively in [B]). Note the mainly superficial vessels from 5 weeks onward in the hyperoxia mice; however, large-caliber vessels are seen penetrating the central retina in 20-week hyperoxia-exposed mice. Scale bars: 1 mm, right column (A, D, G); 50 μm (B, C, E, F, H, I).
Discussion
The present study describes the long-term clinical, functional, and histopathologic changes in a small animal model that result from early hyperoxia exposure. Some of the features of this oxygen-induced vitreoretinopathy (OIVR) partly mimic some of the features of severe ROP while others are closer in nature to various congenital vitreoretinal disorders. The noteworthy features in this new OIVR model produced by a hyperoxia regimen of 65% O2 exposure between P0 and P7 include vitreous hemorrhage; retinal detachment; retinal degeneration; loss of outer and inner retinal function; cicatricial changes or glial scarring; persistence of the primary vitreous or VHP and TVL; and a peripheral avascular zone with leakage of vessels at the interface of vascularized and nonvascularized retina. These changes produced by early exposure to hyperoxia cause a permanent vitreoretinal degeneration, which ERG studies confirmed resulted in the marked loss of retinal function that fails to recover by 8 weeks. Indeed, the inner retina or ganglion cell function deteriorates further by 20 weeks. 
The well-studied 75% O2 P7 to P12 mouse OIR model has been pivotal in elucidating the mechanism of vessel regression during the vaso-obliterative phase of ROP (phase I), and the role of VEGF and other factors such as insulin-like growth factor-1 (IGF-1) in the subsequent neovascular response to relative hypoxia when animals or neonates are returned to room air (phase II of ROP).11 However, the OIR model does not display several of the other characteristic features of severe ROP.11,13 Specifically, the presence of a central avascular area or vaso-obliterative zone in this mouse OIR model is unlike human ROP, where by contrast it is the peripheral retina that remains avascular due to impaired radial growth of retinal vessels.5 Further, the retinal vasculature in the P7 to P12 75% O2 mouse OIR largely remodels during the postnatal reparative phase (P21 onward), and the hyaloid system regresses as normal by week 5 or 6.14,17,19 While remodeling occurs in mild ROP, severe untreated ROP can lead to a lifetime of visual impairment. The OIVR model described herein leads to abnormal persistence of the hyaloid system, not only a feature of a minority of cases severe ROP33 but also a characteristic of a spectrum of other sight-threatening conditions including the congenital abnormality known as persistent hyperplastic primary vitreous (PHPV).34 This new OIVR model thus presents an opportunity to dissect the molecular and cellular mechanisms that lead to persistence of the hyaloid system. Such persistence may relate to the delayed growth of retinal vasculature and hence the reduced oxygenation levels in the vitreous when animals were returned to normoxia at P7. Perhaps this relative hypoxia interferes with the normal mechanisms that cause hyaloid regression. 
We acknowledge that the timing of the hyperoxia exposure in the present OIVR model is not completely comparable to premature human infants at risk of ROP as retinal vascularization has barely commenced in the mouse eye at P0,9 while in the human eye at 23 to 25 weeks of gestation, between 50% and 60% of the retina is already vascularised.7 However, in both situations there is developing growing retina that is unvascularized. Whether the current model can inform us on the mechanisms of some of the pathologic changes observed will prove fertile ground for future studies. 
Ashton and other investigators9,39,40 noted many years ago that the retinae of mice exposed to hyperoxia between P0 and P5 had more severe changes than those of mice exposed to hyperoxia at later time points. Ashton attributed this to a complete obliteration of the developing retinal vessel complex after 5 days of hyperoxia. Upon return to room air for 5 days (P10), mice exposed to early hyperoxia showed abnormal retinal neovascularization and vasoproliferation into the adjacent vitreous accompanied by an overgrowth and proliferation of hyaloid vessels similar to those described in the present study. These observations went largely unnoticed until Gole and colleagues,10,22 in a study of P10 mice exposed to 98% to 100% O2 from P0 to P5, noted ultrastructural changes in vessels of the retina, vitreous, or hyaloid system that mimicked the neovascular phase of ROP and concluded that to produce proliferative vitreoretinopathy in mouse, the O2 concentration must be in the order of 98% as had been previously postulated. In our laboratory, exposure to concentrations of O2 over 75% at P0 onward leads to death of the pups (data not shown). We conclude therefore that it is not only the timing of the hyperoxia but also the choice of O2 concentration of 65% that may be responsible for the observations presented herein. 
As long ago as 1957, Ashton41 commented in relation to the value of animal models of ROP that “[t]he main divergence is in the course of their development, for while the vasoproliferations in the premature infant frequently (but by no means inevitably) progress to retinal detachment and cicatrization, they regress in the animal eye and retinal detachment does not develop” (450). The clinical and correlated histopathologic data in the present study convincingly show that our P0 to P7 65% O2 protocol does indeed produce an animal model with vitreous hemorrhages, vitreous contractile membranes (as evidenced by the presence of α-SMA), retinal detachment, cicatrial changes (as evidenced by increased and longstanding GFAP expression), and profound retinal degeneration. While in premature infants only 6% of ROP patients may progress to threshold or severe disease, even apparently involuted ROP is still associated with changes in the eye including abnormal nondichotomous branching of retinal vessels, failure to vascularize the peripheral retina, vitreous membranes with vitreoretinal traction, retinal detachment, and persistent tortuosity of vessels,42 all of which are seen as predictors of unfavorable clinical sequelae.43 In the original description of the P7 to P12 75% O2 mouse OIR model, Smith et al.13 stated it was not their intention to produce a model with cicatricial changes or retinal detachment but rather a model of retinal neovascularization, and indeed, it is for this reason that most investigators choose to look at this model at the early time points (around P17). The presence of “neovascular tufts” (small vessels that appear elevated from the inner retinal surface, which arise from retinal vessels) is an important characteristic of the conventional P7 to P12 75% O2 OIR model. However, even at our earliest time point (3 weeks), we did not observe obvious examples of neovascular tufts in sections or whole mounts. We concluded that the numerous vessels in the vitreous identified on SD-OCT and histologic sections were in fact abnormally persistent hyaloid vessels and in the main were not originating from the retinal vessels. 
From ERG studies in humans it is known that the magnitude the photoreceptoral response is smaller and kinetics of activation of phototransduction are prolonged in rod photoreceptors in children (10–16 years of age) who have suffered severe ROP due to preterm birth (mean gestational age [MGA] of 26 weeks, mean birth weight 700 g).44 We found both a significant reduction in photoreceptoral response amplitude and phototransduction activation kinetics at 8 weeks in hyperoxia-exposed mice, an effect that persisted at 20 weeks. In addition to outer retinal deficits, previous studies in the rat model of OIR have indicated decreased inner retinal responses as evidenced by a diminished oscillatory potentials model at P31 when retinopathy is maximal.45 This is in agreement with our observation of inner retinal deficits in OPs and the ganglion cell STR. Functionally, while both the a- and b-waves are significantly reduced at P18 in the mouse P7 to P12 75% O2 OIR model,17,23 recent evidence describes their near recovery by 8 weeks.18 Not only has our present study revealed that the hyperoxia between P0 and P7 produces prolonged and severe changes (60%–70% loss) in the mouse retina ERG, but we have specifically demonstrated the continued deterioration in ganglion cell function between 8- and 20-week-old mice, which indicates that the pathologic changes are not merely restricted to the photoreceptor layer in the early phase of the disease but that there is an ongoing deterioration in inner retinal function. The significant correlations between vitreal opacities and ERG data we observed strongly suggest that this compromised inner retinal function may in part be due to the ongoing effects of contractile changes in preretinal membranes. To our knowledge, this is the first study not only to demonstrate contractile elements in preretinal membranes in a mouse model of vireoretinopathy but also to show that their presence is likely strongly correlated with functional deterioration in retinal function. 
While the lack of mechanistic data may be regarded as a limitation of the present study, we propose that this model presents many opportunities for such studies in the future, and indeed these are currently in progress. 
Acknowledgments
The authors thank technical support staff of the Monash Micro Imaging facility, the histology platform, and animal house staff. 
Supported by Jean and Julius Tahija Foundation, National Health and Medical Research Council of Australia (1046203), Australian Research Council (FT130100338). 
Disclosure: P.G. McMenamin, None; R. Kenny, None; S. Tahija, None; J. Lim, None; C. Naranjo Golborne, None; X. Chen, None; S. Bouch, None; F. Sozo, None; B. Bui, None 
References
Campbell K. Intensive oxygen therapy as a possible cause of retrolental fibroplasia; a clinical approach. Med J Aust. 1951; 2: 48–50.
Patz A, Hoeck LE, De La Cruz E. Studies on the effect of high oxygen administration in retrolental fibroplasia. I. Nursery observations. Am J Ophthalmol. 1952; 35: 1248–1253.
Ashton N. Pathological basis of retrolental fibroplasia. Br J Ophthalmol. 1954; 38: 385–396.
Hellstrom A, Smith LE, Dammann O. Retinopathy of prematurity. Lancet. 2013; 382: 1445–1457.
Lutty GA, Chan-Ling T, Phelps DL, et al. Proceedings of the Third International Symposium on Retinopathy of Prematurity: an update on ROP from the lab to the nursery (November 2003, Anaheim California). Mol Vis. 2006; 12: 532–580.
Rivera JC, Sapieha P, Joyal JS, et al. Understanding retinopathy of prematurity: update on pathogenesis. Neonatology. 2011; 100: 343–353.
Ashton N. Retinal angiogenesis in the human embryo. Br Med Bull. 1970; 26: 103–106.
Phelps DL, Rosenbaum AL. Effects of marginal hypoxemia on recovery from oxygen-induced retinopathy in the kitten model. Pediatrics. 1984; 73: 1–6.
Ashton N. Some aspects of the comparative pathology of oxygen toxicity in the retina. Ophthalmologica. 1970; 160: 54–71.
Gole GA, Browning J, Elts SM. The mouse model of oxygen-induced retinopathy: a suitable animal model for angiogenesis research. Doc Ophthalmol. 1990; 74: 163–169.
Stahl A, Connor KM, Sapieha P, et al. The mouse retina as an angiogenesis model. Invest Ophthalmol Vis Sci. 2010; 51: 2813–2826.
Dorrell MI, Friedlander M. Mechanisms of endothelial cell guidance and vascular patterning in the developing mouse retina. Prog Retin Eye Res. 2006; 25: 277–295.
Smith LE, Wesolowski E, McLellan A, et al. Oxygen-induced retinopathy in the mouse. Invest Ophthalmol Vis Sci. 1994; 35: 101–111.
Connor KM, Krah NM, Dennison RJ, et al. Quantification of oxygen-induced retinopathy in the mouse: a model of vessel loss vessel regrowth and pathological angiogenesis. Nat Protoc. 2009; 4: 1565–1573.
Hartnett ME, Penn JS. Mechanisms and management of retinopathy of prematurity. N Engl J Med. 2013; 368: 1162–1163.
International Committee for the Classification of Retinopathy of Prematurity. The International Classification of Retinopathy of Prematurity revisited. Arch Ophthalmol. 2005; 123: 991–999.
Vessey KA, Wilkinson-Berka JL, Fletcher EL. Characterization of retinal function and glial cell response in a mouse model of oxygen-induced retinopathy. J Comp Neurol. 2011; 519: 506–527.
Nakamura S, Imai S, Ogishima H, Tsuruma K, Shimazawa M, Hara H. Morphological and functional changes in the retina after chronic oxygen-induced retinopathy. PLoS One. 2012; 7: e32167.
Ritter MR, Aguilar E, Banin E, Scheppke L, Uusitalo-Jarvinen H, Friedlander M. Three-dimensional in vivo imaging of the mouse intraocular vasculature during development and disease. Invest Ophthalmol Vis Sci. 2005; 46: 3021–3026.
Lobov IB, Rao S, Carroll TJ, et al. WNT7b mediates macrophage-induced programmed cell death in patterning of the vasculature. Nature. 2005; 437: 417–421.
Stefater JAIII, Lewkowich I, Rao S, et al. Regulation of angiogenesis by a non-canonical Wnt-Flt1 pathway in myeloid cells. Nature. 2011; 474: 511–515.
Browning J, Wylie CK, Gole G. Quantification of oxygen-induced retinopathy in the mouse. Invest Ophthalmol Vis Sci. 1997; 38: 1168–1174.
Mattapallil MJ, Wawrousek EF, Chan CC, et al. The Rd8 mutation of the Crb1 gene is present in vendor lines of C57BL/6N mice and embryonic stem cells and confounds ocular induced mutant phenotypes. Invest Ophthalmol Vis Sci. 2012; 53: 2921–2927.
O'Reilly M, Hansbro PM, Horvat JC, Beckett EL, Harding R, Sozo F. Bronchiolar remodeling in adult mice following neonatal exposure to hyperoxia: relation to growth. Anat Rec (Hoboken). 2014; 297: 758–769.
Chen X, Kezic JM, Forrester JV, et al. In vivo multi-modal imaging of experimental autoimmune uveoretinitis in transgenic reporter mice reveals the dynamic nature of inflammatory changes during disease progression. J Neuroinflammation. 2015; 12: 17.
Scott A, Powner MB, Fruttiger M. Quantification of vascular tortuosity as an early outcome measure in oxygen induced retinopathy (OIR). Exp Eye Res. 2014; 120: 55–60.
He Z, Vingrys AJ, Armitage JA, Nguyen CT, Bui BV. Chronic hypertension increases susceptibility to acute IOP challenge in rats. Invest Ophthalmol Vis Sci. 2014; 55: 7888–7895.
Schindelin J, Arganda-Carreras I, Frise E, et al. Fiji: an open-source platform for biological-image analysis. Nat Methods. 2012; 9: 676–682.
Kapur JS, Sahoo PK, Wong ACK. A new method for gray-level picture thresholding using the entropy of the histogram. Comput Gr Image Process. 1985; 29: 273–285.
Zhao D, He Z, Vingrys AJ, Bui BV, Nguyen CT. The effect of intraocular and intracranial pressure on retinal structure and function in rats. Physiol Rep. 2015; 3 (8): e12507.
Tual-Chalot S, Allinson KR, Fruttiger M, Arthur HM. Whole mount immunofluorescent staining of the neonatal mouse retina to investigate angiogenesis in vivo. J Vis Exp. 2013; 77: e50546.
McMenamin PG. Optimal methods for preparation and immunostaining of iris, ciliary body, and choroidal wholemounts. Invest Ophthalmol Vis Sci. 2000; 41: 3043–3048.
Eller AW, Jabbour NM, Hirose T, Schepens CL. Retinopathy of prematurity. The association of a persistent hyaloid artery. Ophthalmology. 1987; 94: 444–448.
Goldberg MF. Persistent fetal vasculature (PFV): an integrated interpretation of signs and symptoms associated with persistent hyperplastic primary vitreous (PHPV). LIV Edward Jackson Memorial Lecture. Am J Ophthalmol. 1997; 124: 587–626.
McLenachan S, Magno AL, Ramos D, et al. Angiography reveals novel features of the retinal vasculature in healthy and diabetic mice. Exp Eye Res. 2015; 138: 6–21.
Ruggeri M, Wehbe H, Jiao S, et al. In vivo three-dimensional high-resolution imaging of rodent retina with spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2007; 48: 1808–1814.
Bui BV, Fortune B. Ganglion cell contributions to the rat full-field electroretinogram. J Physiol. 2004; 555 (pt 1): 153–173.
Tahija SG, Hersetyati R, Lam GC, Kusaka S, McMenamin PG. Fluorescein angiographic observations of peripheral retinal vessel growth in infants after intravitreal injection of bevacizumab as sole therapy for zone I and posterior zone II retinopathy of prematurity. Br J Ophthalmol. 2014; 98: 507–512.
Gyllensten LJ, Hellstrom BE. Experimental approach to the pathogenesis of retrolental fibroplasia. I. Changes of the eye induced by exposure of newborn mice to concentrated oxygen. Acta Paediatr Suppl. 1954; 43: 131–148.
Patz A, Eastham A, Higginbotham DH, Kleh T. Oxygen studies in retrolental fibroplasia. II. The production of the microscopic changes of retrolental fibroplasia in experimental animals. Am J Ophthalmol. 1953; 36: 1511–1522.
Ashton N. Experimental retrolental fibroplasia. Annu Rev Med. 1957; 8: 441–454.
Kaiser RS, Trese MT, Williams GA, Cox MSJr. Adult retinopathy of prematurity: outcomes of rhegmatogenous retinal detachments and retinal tears. Ophthalmology. 2001; 108: 1647–1653.
Multicenter trial of cryotherapy for retinopathy of prematurity. One-year outcome--structure and function. Cryotherapy for Retinopathy of Prematurity Cooperative Group. Arch Ophthalmol. 1990; 108: 1408–1416.
Hansen RM, Moskowitz A, Tavormina JL, Bush JN, Soni G, Fulton AB. Temporal summation in children with a history of retinopathy of prematurity. Invest Ophthalmol Vis Sci. 2015; 56: 914–917.
Liu K, Akula JD, Hansen RM, Moskowitz A, Kleinman MS, Fulton AB. Development of the electroretinographic oscillatory potentials in normal and ROP rats. Invest Ophthalmol Vis Sci. 2006; 47: 5447–5452.
Milic R, Banfi G, Del Fabbro M, Dopsaj M. Serum creatinine concentrations in male and female elite swimmers. Correlation with body mass index and evaluation of estimated glomerular filtration rate. Clin Chem Lab Med. 2010; 49: 285–289.
Figure 1
 
The clinical appearance of the mouse eye as seen using Micron III retinal fundus imaging at various recovery times after P0 to P7 hyperoxia exposure. Left column: bright-field (BF) mode; middle column: fluorescein angiography (FA) mode focused on the retinal vessels; left column: FA focused on the hyaloid vessels. The retina and retinal vessels cannot be visualized clearly in P10 and P14 due the hyaloid artery and its branches, which are hypertrophic at both time points. Note the vitreous hemorrhages at 3 weeks, and retinal detachments (arrowheads) and focal degenerative changes (pale focal areas) from week 8 onward together with the highly tortuous retinal and persistent hyaloid vessels at all time points. Further time points are shown in Supplementary Figure S2.
Figure 1
 
The clinical appearance of the mouse eye as seen using Micron III retinal fundus imaging at various recovery times after P0 to P7 hyperoxia exposure. Left column: bright-field (BF) mode; middle column: fluorescein angiography (FA) mode focused on the retinal vessels; left column: FA focused on the hyaloid vessels. The retina and retinal vessels cannot be visualized clearly in P10 and P14 due the hyaloid artery and its branches, which are hypertrophic at both time points. Note the vitreous hemorrhages at 3 weeks, and retinal detachments (arrowheads) and focal degenerative changes (pale focal areas) from week 8 onward together with the highly tortuous retinal and persistent hyaloid vessels at all time points. Further time points are shown in Supplementary Figure S2.
Figure 2
 
Quantitative analysis of the number of hyaloid vessels observed by in vivo fundus imaging in P0 to P7 hyperoxia-exposed mice and controls (A); the tortuosity index of those hyaloid vessels (B) and retinal vessels (C). Values are mean ± SEM, ***P < 0.001 (1-tailed t-test). (A) Control group sizes, n = 11, 5, 20, 8, 16; experimental group sizes, n = 26, 9, 18, 8, 29. (B) Control group sizes, n = 11, 4, 13, 0, 3; experimental groups, n = 26, 9, 18, 8, 29. (C) Control group sizes, n = 11, 5, 20, 7, 16; experimental group sizes, n = 26, 9, 18, 8, 29.
Figure 2
 
Quantitative analysis of the number of hyaloid vessels observed by in vivo fundus imaging in P0 to P7 hyperoxia-exposed mice and controls (A); the tortuosity index of those hyaloid vessels (B) and retinal vessels (C). Values are mean ± SEM, ***P < 0.001 (1-tailed t-test). (A) Control group sizes, n = 11, 5, 20, 8, 16; experimental group sizes, n = 26, 9, 18, 8, 29. (B) Control group sizes, n = 11, 4, 13, 0, 3; experimental groups, n = 26, 9, 18, 8, 29. (C) Control group sizes, n = 11, 5, 20, 7, 16; experimental group sizes, n = 26, 9, 18, 8, 29.
Figure 3
 
SD-OCT imaging of control (A) and hyperoxia-exposed mice at 8 weeks (B, C) and 20 weeks (D). Note the dense membranes in the vitreous (arrowheads) in hyperoxia-exposed mice and the manner in which the retina is puckered or elevated where these membranes attach to the inner retina. Extensive retinal detachment is clearly evident (large arrow [B]). Isolated profiles of hyaloid vessels are also present in 8- and 20-week-old eyes. Note the large rosette in the retina at 8 weeks (black arrowhead [C]).
Figure 3
 
SD-OCT imaging of control (A) and hyperoxia-exposed mice at 8 weeks (B, C) and 20 weeks (D). Note the dense membranes in the vitreous (arrowheads) in hyperoxia-exposed mice and the manner in which the retina is puckered or elevated where these membranes attach to the inner retina. Extensive retinal detachment is clearly evident (large arrow [B]). Isolated profiles of hyaloid vessels are also present in 8- and 20-week-old eyes. Note the large rosette in the retina at 8 weeks (black arrowhead [C]).
Figure 4
 
Retinal function as assessed by ERG in control and OIR mice (65% O2 P0–P7) at 8 weeks (A, B) and 20 weeks (C, D). (A) Average group waveforms for control (thin traces) and OIR mice (thick traces) elicited with a bright (a-wave, b-wave, oscillatory potentials46 at 1.2 log cd.s/m2) and dim (scotopic threshold response, −5 log cd.s/m2) flash. (B) Size (average ± SEM) of each of the major ERG components expressed relative to the group average of control eyes (OIR response/average control, %). (C) Average waveforms at 20 weeks. (D) Relative response at 20 weeks. At 8 weeks, n = 8 hyperoxia, n = 9 controls; 20 weeks, n = 8 hyperoxia, n = 6 controls.
Figure 4
 
Retinal function as assessed by ERG in control and OIR mice (65% O2 P0–P7) at 8 weeks (A, B) and 20 weeks (C, D). (A) Average group waveforms for control (thin traces) and OIR mice (thick traces) elicited with a bright (a-wave, b-wave, oscillatory potentials46 at 1.2 log cd.s/m2) and dim (scotopic threshold response, −5 log cd.s/m2) flash. (B) Size (average ± SEM) of each of the major ERG components expressed relative to the group average of control eyes (OIR response/average control, %). (C) Average waveforms at 20 weeks. (D) Relative response at 20 weeks. At 8 weeks, n = 8 hyperoxia, n = 9 controls; 20 weeks, n = 8 hyperoxia, n = 6 controls.
Figure 5
 
Histopathology of representative eyes from P10 (AC) and P14 (DF). At P10, note the prominent hyaloid vessels and TVL. The retinal morphology appears ostensibly normal for the age of the mouse. At P14 the hyaloid artery and TVL are hyperplastic and extend onto the retina near the optic nerve head and form a thick distinct capsule on the posterior lens surface (E) that extends laterally and appears to send branches to supply the peripheral retina (F). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Figure 5
 
Histopathology of representative eyes from P10 (AC) and P14 (DF). At P10, note the prominent hyaloid vessels and TVL. The retinal morphology appears ostensibly normal for the age of the mouse. At P14 the hyaloid artery and TVL are hyperplastic and extend onto the retina near the optic nerve head and form a thick distinct capsule on the posterior lens surface (E) that extends laterally and appears to send branches to supply the peripheral retina (F). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Figure 6
 
Histopathology of representative eyes from 3-week (AC) and 8-week (DF) time points in hyperoxia-exposed mice. Low magnification (insets) reveals the general layout of the mouse eyes and shows normal anterior segment morphology. At 3 weeks of age, a large prominent hyaloid artery along with vitreal membranes (black arrow) is clearly identifiable bridging between the connective tissue around the hyaloid artery and the elevated retinal surface. Extensive retinal folding and rosettes are also visible. Vitreous hemorrhages (red arrows) are clearly identifiable (A, B), as are focal regions of extensive outer retinal degeneration (B, C). At 8 weeks of age, vitreous membranes (arrow in [D]) are still evident associated with elevated and disrupted retinal layers. Numerous hyaloid or epiretinal vessels (arrows in [E]). Note extensive outer retinal disruption and folding and loss of outer retinal layers (arrowhead in [F]). Control normoxia mice were normal and revealed no abnormal pathology at all time points examined (Supplementary Fig. S3). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Figure 6
 
Histopathology of representative eyes from 3-week (AC) and 8-week (DF) time points in hyperoxia-exposed mice. Low magnification (insets) reveals the general layout of the mouse eyes and shows normal anterior segment morphology. At 3 weeks of age, a large prominent hyaloid artery along with vitreal membranes (black arrow) is clearly identifiable bridging between the connective tissue around the hyaloid artery and the elevated retinal surface. Extensive retinal folding and rosettes are also visible. Vitreous hemorrhages (red arrows) are clearly identifiable (A, B), as are focal regions of extensive outer retinal degeneration (B, C). At 8 weeks of age, vitreous membranes (arrow in [D]) are still evident associated with elevated and disrupted retinal layers. Numerous hyaloid or epiretinal vessels (arrows in [E]). Note extensive outer retinal disruption and folding and loss of outer retinal layers (arrowhead in [F]). Control normoxia mice were normal and revealed no abnormal pathology at all time points examined (Supplementary Fig. S3). Magnifications: ×50, insets ×10 (A, D); ×100 (B, C, E, F).
Figure 7
 
Immunostained frozen sections of control and hyperoxia-exposed eyes at 8 weeks showing astrogliosis in the superficial retina (abnormal GFAP immunoreactivity, top right) and abnormal deposition of type IV collagen (Coll IV; middle right) together with the presence of α-smooth muscle actin (α-SMA)-positive staining in the superficial retina and vitreal membranes (bottom right). Bar: 50 μm.
Figure 7
 
Immunostained frozen sections of control and hyperoxia-exposed eyes at 8 weeks showing astrogliosis in the superficial retina (abnormal GFAP immunoreactivity, top right) and abnormal deposition of type IV collagen (Coll IV; middle right) together with the presence of α-smooth muscle actin (α-SMA)-positive staining in the superficial retina and vitreal membranes (bottom right). Bar: 50 μm.
Figure 8
 
Retinal whole mounts from control and hyperoxia-exposed mice at 5 and 20 weeks stained with Ib4 (red/magenta). The entire retinal vasculature is shown in the left column (A, D, G), and confocal three-dimensional reconstructions or Z profiles of retinal vessels in the central and peripheral retina are in the right column. The latter show the extent of penetration of the capillaries into the neural retina. The nuclear layers of the retina (ganglion cell layer [GCL], inner nuclear layer [INL], and outer nuclear layers [ONL]) are visualized by Hoechst nuclear staining (blue) and best seen in control retinal profiles (B, C). The vitread aspect of the retina is orientated superiorly. Note that at 5 weeks of age in normal eyes there are already three layers of vessels in the central and peripheral retina: a superficial, an intermediate (inner plexiform layer), and a deep (outer plexiform layer) (labeled 1, 2, 3, respectively in [B]). Note the mainly superficial vessels from 5 weeks onward in the hyperoxia mice; however, large-caliber vessels are seen penetrating the central retina in 20-week hyperoxia-exposed mice. Scale bars: 1 mm, right column (A, D, G); 50 μm (B, C, E, F, H, I).
Figure 8
 
Retinal whole mounts from control and hyperoxia-exposed mice at 5 and 20 weeks stained with Ib4 (red/magenta). The entire retinal vasculature is shown in the left column (A, D, G), and confocal three-dimensional reconstructions or Z profiles of retinal vessels in the central and peripheral retina are in the right column. The latter show the extent of penetration of the capillaries into the neural retina. The nuclear layers of the retina (ganglion cell layer [GCL], inner nuclear layer [INL], and outer nuclear layers [ONL]) are visualized by Hoechst nuclear staining (blue) and best seen in control retinal profiles (B, C). The vitread aspect of the retina is orientated superiorly. Note that at 5 weeks of age in normal eyes there are already three layers of vessels in the central and peripheral retina: a superficial, an intermediate (inner plexiform layer), and a deep (outer plexiform layer) (labeled 1, 2, 3, respectively in [B]). Note the mainly superficial vessels from 5 weeks onward in the hyperoxia mice; however, large-caliber vessels are seen penetrating the central retina in 20-week hyperoxia-exposed mice. Scale bars: 1 mm, right column (A, D, G); 50 μm (B, C, E, F, H, I).
Supplement 1
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×