Long-Term Impairment of Retinal Ganglion Cell Function After Oxygen-Induced Retinopathy

Abstract

Premature infants with retinopathy of prematurity (ROP) have neovascularization of the retina, potentially resulting in low vision and even blindness. Some of these infants still have visual impairment, even if ROP resolves as they age. However, the mechanisms underlying the visual problems post-ROP are poorly understood. Because the pathological neovascularization in ROP infants can be mimicked in a mouse model with oxygen-induced retinopathy (OIR), we recapitulated post-ROP with post-OIR mice a few months after spontaneous regression of retinal neovascularization. Our pattern electroretinogram test demonstrates that post-OIR mice exhibit reduced P1-N2 responses, suggesting the impairment of retinal ganglion cells, the retina's output neurons. However, immunohistochemistry reveals that the density of retinal ganglion cells remains unchanged in post-OIR mice, indicating that the aforementioned pattern electroretinogram changes are functional. Our data further demonstrate that both light-adapted ex vivo electroretinogram a-waves (cone responses) and in vivo electroretinogram b-waves (ON cone bipolar cell responses) were significantly impaired in post-OIR mice. These results suggest that post-OIR impairment of the retinal cone pathway appears to result in the dysfunction of retinal ganglion cells, contributing to visual problems. A similar cellular mechanism could occur in post-ROP children, which is responsible for their visual impairment.

Keywords: bipolar cell; cone; oxygen–induced retinopathy; pattern electroretinogram; retinal ganglion cell; retinopathy of prematurity.

Figure 1

Post–OIR impairment of RGC function in vivo. The neural structure of the mammalian retina is illustrated (A), showing that cone signals pass through cone bipolar cells (cone-BC) directly to RGCs, but rod signals reach RGCs indirectly through the rod-cone coupling or via the rod—rod bipolar cell (rod-BC)—AII amacrine cell (AII-AC)—cone-BC pathway. Blue and red arrows indicate the flow of the rod and cone signals, respectively. The function of RGCs was evaluated by PERG in vivo. The PERG used 200 cycles of high–contrast (100%) horizontal gratings with a 1-Hz reversal rate, spatial frequency of 0.0589 cyc/deg, and mean luminance of 100 cd·s/m² as stimuli. Subfigures (A,B) illustrate typical recordings from an age–matched control mouse and a mouse 8 weeks after OIR, respectively. The P1–N2 amplitude (17 µV) in the control mouse (B) was larger than that (5 µV) in the post–OIR mouse (C). Data from 8– to 9–week post–OIR mice and aged–matched control mice were pooled and averaged (D), suggesting that the P1–N2 amplitude was significantly attenuated in 8–9 weeks post–OIR mice. The attenuation persisted in 22– to 24–week post–OIR mice (E).

Figure 2

No significant changes of the RGC density in post–OIR mice. RGCs were immuno–stained with an antibody against RPBMS in whole–mount retinas isolated from age–matched mice and mice 24 weeks after OIR. Images were taken from the center (A) and peripheral region (B) of the retina. The number of RGCs was manually counted and then divided by the counted area to find the RGC density. The average RGC density in the center and periphery was depicted in (C,D), respectively. n.s., no significance.

Figure 3

Reduced b–waves of in vivo light–adapted ERGs in post–OIR mice. Light–adapted full–field ERG a– and b–waves were recorded from age–matched controls (A) and 8–week post–OIR mice (B). Protocol details and ERG waveforms (a– and b–waves and OPs) are described in the main text. The intensity-response curve of b–waves was constructed by plotting its amplitude as a function of the stimulus intensity (C). * p < 0.05; **p < 0.01. 1 log unit = 10 cd·s/m².

Figure 4

Reduced a–waves of ex vivo light–adapted ERGs in post–OIR mice. Light–adapted a–waves were recorded from isolated retinas of age–control mice (A) and 8–week post–OIR mice (B), whereas b–waves were pharmacologically eliminated. The procedures of retina preparation and recording protocols are described in the main text. The peak amplitudes of a–waves were plotted as a function of the stimulus intensity to construct an intensity response curve (C). * p < 0.05, 0 log unit: 2.20 × 10¹⁶ photons/cm²/s.

Figure 5

No significant changes of the cone density in post–OIR mice. S–opsin–expressing and M–opsin–expressing cones were immunostained in whole–mount retinas isolated from age–matched mice and mice 24 weeks after OIR. Images were taken for M–opsin–expressing cones from the center (A) and peripheral region (B) of the dorsal retina. The number of M–opsin–expressing cones was manually counted and then divided by the counted area to find the density. The average M–opsin–expressing cone density for the center and periphery was depicted in (C,D), respectively. Similar procedures were used for S–opsin–expressing cone imaging from the ventral retina where S–opsin is predominantly expressed. Subfigures (E,F) show the average S–cone density for the center and periphery, respectively. n.s, no significance.

Figure 6

Reduced b–waves of in vivo and ex vivo scotopic ERGs in post–OIR mice. (A). In vivo dark–adapted full–field ERG a– and b–waves were recorded from 8– to 9–week post–OIR and age–matched control mice in response to 4 ms white light pulses with −1 or −2 log cd·s/m² (1 log unit = 10 cd·s/m²), only stimulating rod photoreceptors. The peak amplitude of a–waves remained unchanged at both intensities (top panel). Significant changes in b–waves were observed at high intensity but not at low intensity (bottom panel). (B). Ex vivo dark–adapted ERGs showed similar results as in vivo dark–adapted ERGs. Top panel, a–waves; bottom panel, b–waves. The 470 nm light pulses with a 20 ms duration were applied at −6.26 or −4.84 log photons/cm²/s. Both intensities only stimulated rod photoreceptors. *, p < 0.05; n.s., no significance.